(updated 8.4.2022)

8.4.2022 6:40pm - Hilarious conversation! "American Jews think/act like Christians without Jesus."

8.4.2022 6:02pm - (Replying to @thinkyparts @ADL and @JGreenblattADL) Thanks for the warning. I usually us "mother-milk deep" but, given what I see and hear from our neighbors, I'm not sure which is deeper.

8.4.2022 5:57pm - (Replying to @Rondo2 @ADL and @JGreenblattADL) I wouldn't give up so easily on @ADL . We recently heard some sparks of understanding from @JGreenblattADL , and he might even clarify here what @ADL 's position is on "pure" Zionophobic bigotry, one that does not rattle any anti-Semitic tropes American Jews habitually fear.

8.4.2022 2:05pm - (Replying to @thinkyparts @ADL and @JGreenblattADL) 2-State solution has always been Zionists goal and, while settlements make this goal difficult, what makes it IMPOSSIBLE is Arab's rejectionism: "not even the size of a postage stamp". There's ample ambiguity in defining antisemitism, anti-Zionism is crisp: No Jewish sovereignty.

8.4.2022 1:21pm - I was happy to see the @ADL 's report pointing to the danger of anti-Zionist candidates. Alas, the only fault @ADL finds in those candidates is invoking "Jewish power and control" rhetoric. Does @ADL deem Anti-Zionism in itself to be a legitimate form of bigotry? @JGreenblattADL

8.4.2022 11:59am - (Replying to @Big_Worker @EliSennesh and 2 others) Evidently, some people take seriously what these 2 UN "experts" say. Some even quote them (Lynk and Rajagopal) on Tweeter, as if apologizing: "This is where we get our education on the ME conflict. Don't blame us for being so ignorant."

8.4.2022 10:00am - (Replying to @kareem_carr0 Why "think" when it is easy to test. The part that comes from the mental model is not available to a machine, so, lets give the same data to a machine and ask her questions about "what's going on". That's what we did mathematically and found: "No causes in, no understanding out".

8.4.2022 9:10am - (Replying to @waseemkiani and @MyNutritionSci) Let me assure you, and I am unbiased, in this case it's not "jut marketing." When you are done, I'll explain what makes "revolutionary" different from just "useful".

8.3.2022 11:02pm - I am willing to forgive @jstreetdotorg for their fight with AIPAC, if only they would change their insulting logo: "pro-Israel, pro-peace", as if Israel, if left alone, is not craving for peace; as if Israelis need J-street push and supreme wisdom to fulfil their lifelong dream.

8.3.2022 10:33pm - (Replying to @ESMDcan123 and @artistexyz) My point is that the study consists of two stops: The consenting subjects are first assigned randomly to treatment and control arms, and then subjected to forced intervention, depending on the assignment. This second step corresponds to the letter 'C' in RCT.

8.3.2022 10:13pm - @NaviPillay1 seems determined to bring the reputation of the UN Human Right Council to below zero.

8.3.2022 10:00pm - Who said Americans would give up their hard earned freedoms?

8.3.2022 6:43pm - I believe it came about to prevent people from assuming that you can get similar results by doing post-treatment randomization. IOW, to emphasize that an RCT entails two steps: randomization and forced intervention.

8.3.2022 6:34pm - Hold your head high, @Andy_Levin . You now have a choice: (1) Bask in @PeterBeinart admiration of your heroic stand, (2) Mull over a realistic assessment of your mistake: Palestinian freedom entails honest co-existence, a word detested by the squad members you have embraced.

8.3.2022 5:19pm - (Replying to @TheOtherSimo) The ideology that equates the survival of Israel with "hating Palestinians" is precisely the ideology that has prevented Palestinians from accepting co-existence with Israel (since 1948), which would have brought peace, independence and prosperity to both sides.

8.3.2022 3:28am - We are informed of an interesting discussion group on causality from Neurips viewpoint. It has Rung-3 of the ladder on their icon, so they must know where they are heading.

8.3.2022 3:05am - (Replying to @RabbiLindaGold1 and @JenChangeFL) I see great future for him, once he realizes that Jewish voters do care about Israel, and are no longer fooled by populist slogans J-Street style.

8.3.2022 2:49am - (Replying to @StaceyEBurke and @JenChangeFL0 Zionophobes can't tells lies from truth, their entire universe is built on the former.

8.3.2022 2:43am - (Replying to @JenChangeFL) It's 2022 and people are still touting: "There is a difference between anti-semitism & anti-Zionism", not realizing that voters deem the latter uglier and more dangerous than the former, and that the real issue in the case of Andy Levin was his embrace of the dangerous latter.

8.2.2022 9:05pm - (Replying to @jacobmbuckman @kareem_carr and 2 others) Here is the proof, @kareem that everyone, including deep learning folks are interested in "understanding", not just prediction. Do you still see such a big dichotomy between stats and ML? Isn't the latter of great value to the former?

8.2.2022 7:48pm - (Replying to @elderofziyon) I don't understand why good people try to minimize AIPAC's role in Levin's defeat. How else could people like me, a resident of California, express the danger they see in having another Squad sympathizer enter the Hall of US Congress, if not by openly supporting AIPAC?

8.2.2022 6:27pm - (Replying to @kareem_carr) I'm not sure what "stats" you have in mind for which it's fair to say: "In stats, understanding is almost always the goal." My definition of "understanding" makes stats a discipline for which "understanding is almost NEVER the goal". Where do we differ? On stats or understanding?

8.2.2022 4:54pm - The racist always walks around with a genetic test-tube, to test his neighbors linkage, especially racists lacking history, songs, language or holidays; they got nothing to base claims on, but soiled, polluted test-tubes.

8.2.2022 4:35pm - (Replying to @MilenaAlmagro) Worth asking: What Kept the Cowles Commission at Bay? See, p. 168

8.2.2022 4:17pm - I couldn't believe @RBReich actually said this, until I checked it myself, he used to think before speaking. @RBReich , Berkeley professorship does not absolve you from thinking. Do you really believe AIPAC's goal is to "take down progressives"? Not to "take down Zionophobes"?

8.2.2022 2:16pm - (Replying to @78tiger and @blakeflayton) The rumor I mentioned was, of course, poetic. But not too far off. To the best of my recollection, it was @IfNotNowOrg that said Kaddish for both "Israeli and Hamas victims" in the latest Gazza clash -- any distinction between good and evil makes them feel too "ethno-centric".

8.2.2022 11:07am - And if anyone objects to your definition of "indigeneity" ask him/her the questions that Lord Balfour asked professor Rashid Khalidi in this poetic encounter:

8.2.2022 10:48am - ‘I Solemnly Demand That Anti-Zionism Be Banned in France,’ Parliamentary Deputy Declares
8.2.2022 10:48am - The first Jewish leader to recognize that our current fight against antisemitism is misguided, futile, and only gives cover to our real enemies - the Zionophobes. @JGreenblattADL , @AJC , @almog_doron , @noatishby , @EinatWilf , @newzionists , @blakeflayton @SAFECUNY

8.2.2022 10:20am - Our neighbors always side with the oppressed freedom-fighters: Putin, China, Ayatollah, Isis, Taliban,... As the Mishnah says: Your most important choice = Good neighbor

8.2.2022 10:06am - Significant step in re-thinking religion's role in society.

8.2.2022 9:00am - Interesting details on Ayman al-Zawahiri's life, before last week's recall to answer his creator.

8.2.2022 2:09am - Sharing our weekly harvest of causal inference papers: Many of the titles look enticing, but I didn't have a chance to look at them yet.

8.1.2022 11:30pm - (1/2) Background: Here is a sweet letter USC @PresidentFolt sent to 45 Zionist professors who complained about the treatment of Rose Ritch Guess what word is missing! "Zionism" - the core of the complaint. Why? Because dealing with
8.1.2022 11:30pm - (2/2) "antisemitism" is so much safer; all the Jewish organizations praise you for inviting them to the "Advisory Committee on Jewish Life," and give you 5 years to philosophize and pretend you can't spell "Zionism," the toxic word you refuse to embrace in Jewish identity. Sad!

8.1.2022 11:00pm - Jewish leadership still hasn't learnt the futility of whining "antisemitism" and the power of indicting "antiZionism". See how they melt away at USC, and give @PresidentFolt the license she needs for inaction. @bandlersbanter , @ADL , @AJC , @noahfarkas

8.1.2022 9:36pm - (Replying to @blakeflayton) Rumor has it that @InNotNowOrg folks are saying Mourner's Kaddish tonight for Ayman al-Zawahiri, the greatest de-colonizer of all times, and the prophetic voice of "true Judaism."

8.1.2022 8:50pm - (Replying to @kevinsalwen and @MaxBoot) Thanks for writing, Kevin.

8.1.2022 8:11pm - I have a personal account to close with al-Qaeda, more so with al-Zawahiri the "ideologist", than with bin-Laden or Khalid Sheik Mohammed, the executioners of our son, Daniel. Today I feel an inch closer to closure. Angels in heaven tell me they feel that way too.

8.1.2022 7:39pm - (Replying to @HumzahAlkindi) Thanks for sharing the article in Haaretz, but I do not see any contradiction between Popper's view and mine. My position that a nation state can be both successful and open to criticism is attested in the form of Israeli society, which neo-Marxists feel obliged to delegitimize.

8.1.2022 7:09pm - This is not an endorsement (hate to lose my tax status) only my moral duty to let readers know the names of some election candidates who are not Zionophobic - a gift of fresh air in squad-ruled politics.

8.1.2022 6:49pm - (Replying to @HumzahAlkindi) Oh, I've confessed my views here when a reader asked: where does this moral deformity comes from?

8.1.2022 6:33pm - (Replying to @pdddavid1) Thanks.

8.1.2022 3:07pm - (1/2) Readers who inquired on foreign publications of #Bookofwhy, here is what our publisher tells us: 1. US Basic Books, 2. UK Penguin Books , 3. SPANISH Pasado y Presente, 4. HOLLAND Maven, 5. FINLAND Terra Cognita, 6. PORTUGAL Temas y Debates, 7. JAPAN Bungei Shunju,
8.1.2022 3:07pm - (2/2) 8. KOREA ScienceBooks, 9. TAIWAN The Walk Publishing, 10 CHINA CITIC, 11. ISRAEL Kinneret-Zmora Dvir Publishing, 12. RUSSIA AST, 13. POLAND Copernicus Center Press, 14. TURKEY Ginko Publishers You're right, no French publisher found thus far; amazingly behind!

8.1.2022 9:47am - (Replying to @dmkalman and @skdh) I'm generally unsatisfied with "trial and error" until we define what "error" is and how to choose the next "trial". Once we do that, bingo, we can systematize the process, hopefully even automate it.

8.1.2022 9:43am - (Replying to @sean_welsh77 and @David_Gunkel) So you see any impediments to emulate these functions on a computer? I don't. Especially "caring" "morality" and "curiosity", where we currently have the beginning of a thread to their secrets.

8.1.2022 9:35am - (Replying to @StillTr05207382 and @YiMaTweets) You have provided is a fairly accurate description of what has happened in #ML in the past 15 years: Build machines to do tasks, without defining what the "task" is, what success is, and what theoretical limitations are there on the success. e.g.,

8.1.2022 1:32am - (Replying to @sean_welsh77 and @David_Gunkel) From practical AI viewpoint, faking it would be sufficient and extremely useful (and dangerous). From a philosophical viewpoint, just saying "its more than just xyz" whenever xyz is less than an actual human being is not very interesting, unless we hear what function is missing.

7.31.2022 9:22pm - (Replying to @ProudUSJew) Daniel's Yahrzeit is February 1st. If you have a chance, ask your minyan tomorrow in shul to say: Avinu SheBashamayim, Tsur Israel V'Goalo, Barech et Medinat Israel, Reshit Tsmichat Geulatenu... Daniel would appreciate using his name to inspire Jewish pride, resilience and hope.

7.31.2022 7:46pm - (Replying to @_brohrer_) How can you tell a cult of complexity that is a cover up for lack of substance from complexity that arises when science matures? My answer: The latter is accompanied with a toy problem that illustrates the idea with 3-4 variables. I bemoan the lack of toy problems in #ML.

7.31.2022 6:57pm - Correction. Biden said "50 years", not "25 years". Indeed, his first trip to Israel was in 1973

7.31.2022 6:48pm - I would stick to "mimicking" as long as no one explains to me what she means by "ACTUAL phenomenal consciousness". (Recall, faking it is having it, because it's so hard to fake w/o having) And I wouldn't knock "symbol manipulation according to rules" - this is how my brain works.

7.31.2022 6:37pm - (Replying to @bonnieshouses and @Claire_V0ltaire) Thanks, I also like this article, but Wikipedia won't allow me to reference it -- Sister Linda is a protected species. Please try to cite it and see what happens. Perhaps it's only me that she fears.

7.31.2022 6:24pm - (Replying to @ramespaulus and @IfNotNowOrg) I'm going to ask @Twitter for reinstatement, present my Jewish credentials, and pretend I am a Jew of Discomfort. I can even put on an Arafat Yarmulka, chant Havdalah at noon like an idiot. As long as they don't ask me to support Rep. Andy Levin -- I cannot betray my people.

7.31.2022 5:45pm - There are many Americans who are great friends and allies to Israel. But Biden was unique in telling the hecklers so precisely why their screams "Occupation! Occupation!" sounded so idiotic in light of his 25 years search for a single Palestinian leader willing to accept Israel.

7.31.2022 4:03pm - My! My! I am blocked by
----Quote ----
You’re blocked You can’t follow or see @IfNotNowOrg ’s Tweets. Learn more
As Joe Biden noticed, only Jews who cut their historical roots can join these juvenile hecklers.

7.31.2022 3:18pm - (1/2) This should be the main topic of discussion in the next international conference of Clinical Psychiatrists. Some say it's morphed antisemitism - inability to stomach Jews, whose historical place is down below, suddenly act as sovereign equals. But I believe it's more a matter

7.31.2022 3:18pm - (2/2) of social ideology. Israel's success makes a mockery of neo-Marxist textbooks, according to which a nation state must be evil, corrupt and unjust. If it isn't, we must make it look like one. I wish some university would establish a separate department to study this pathology.

7.31.2022 2:42pm - (Replying to @Abel_TorresM @eliasbareinboim and @kerstingAIML) Can you share an example of a "well-known" model = ?

7.31.2022 1:14pm - The #Bookofwhy claims that data are dumb, namely, that they tell us little about the world generating the data. That was before seeing the UNHRC data. Anyone who thinks it doesn't tell us about the moral deformity of those behind these condemnations has not seen moral deformity.

7.31.2022 12:56pm - Emory University Student: I Have Been Attacked and Defamed for Visiting Israel
7.31.2022 12:56pm - I'm retweeting this piece for the hundreds, perhaps thousands of readers facing similar situations: "I will not be stepping down, I will in fact be stepping up and speaking out for all those facing harassment and discrimination on campus because of who they are." @newzionists

7.31.2022 12:20pm - (Replying to @chroma_infinity) I'm not surprised about the Israel issue, the anatomy of which I've studied inside out. But what about econometric in industry? It's hard for me to imagine even ONE such task, where an econometrician cannot benefit immensely from the tools provided by the causal revolution.

7.31.2022 12:08pm - Retweeting: How I wish he (Sewall Wright) was alive today. Reading the #ML literature and saying: they haven't learnt a thing, have they? What an enchanting fantasy!

7.31.2022 11:56am - (Replying to @grbradsk and @YiMaTweets) That would be understandable and forgivable. Trouble starts when these script jockeys decide to take over academia to gain respectability. They then hire hordes of script jockeys to teach and review+reject conference papers and scientific proposals which they cannot understand.

7.31.2022 11:42am - (Replying to @EricJGreenberg1) Any one of the tutorials on the #Bookofwhy page would do. See 3. Scientific Backgrounds

7.31.2022 11:37am - (Replying to @roberto_limongi) I'm more inclined to believe they do not KNOW the philosophical or scientific assumptions behind #ML, else they would be thrilled to teach them.

7.31.2022 9:48am - (Replying to @NicholasMatiasz and @skdh) I havn't given it sufficient thought, since it is only last week that I saw the word "sentience" for the first time.

7.31.2022 6:11am - Unfortunately, narrow-mindedness caused by "too busy writing" also afflict reviewers, so it becomes self-enforcing, hermetically sealing the field from new ideas.

7.31.2022 5:54am - (Replying to @letonyo and @YiMaTweets) How about writing: "The 10 milestones in the history of #AI every busy #ML student should know". Same with "history of science".

7.30.2022 11:05pm - Speaking about Linda Sarsour, did you know that Wikipedia considers her "protected minority"? I tried to cite my article about her - they won't allow it -- she will for ever remain beyond criticism, like a religious icon.

7.30.2022 10:53pm - I've just watched @skdh video on "What's the difference between a brain and a computer" and, I must confess, I remain a staunch believer in AGI, seeing no theoretical impediment to mimicking a conscious human agent on a digital machine.

7.30.2022 10:14pm - Joe Biden is the only Zionist who knows how to handle the @IfNotNow crowd. When they heckled him during the 2020 campaign, he said: "Show me one Palestinian leader willing to accept Israel, and I will talk to you about the "occupation"" (paraphrased from a friendly memory).

7.30.2022 6:53pm - Whenever I see Jaffa in the horizon I am reminded that my 14 yr old father landed there in 1924, with his entire family, hoping to start a new life, back in the country they left 2000 years earlier.

7.30.2022 6:33pm - I've just registered to a lecture that I've always wanted to hear: How to be an ethical computer scientist Knowing Moshe as a top formal-logic researcher, I am sure his take on ethics will be different from the usual social media preaching. @vardi

7.30.2022 4:53pm - Education Dept. to Investigate USC’s Handling of Antisemitic Harassment Against Rose Ritch via @jewishjournal
7.30.2022 4:53pm - (1/4) In as much as I dislike litigations clouding academic affairs, I recognize that they serve a useful function. They call public attention to what's going on, they force administrators to rethink past blunders, and they entice Jewish organizations to tweet: "We're here!"
7.30.2022 4:53pm - (2/4) The next step, once litigation runs its course is for university administrators to act meaningfully in the normative sphere: (1) Do NOT appoint a showcase "task force" to deal with antisemitism. (2) Spell out the word "Zionism" openly and explicitly. (3) Tell the campus
7.30.2022 4:53pm - (3/4) why denying a people's homeland is morally deplorable, why Israeli and Zionist students and faculty are welcome on campus, and what unique cultural and nation-building experience they carry that make them valuable contributors to campus life and academic excellence.
7.30.2022 4:53pm - (4/4) Zionist students and faculty are not asking for legal protection; they ask for rightly earned equity which is within your duty to sustain, by setting norms of respectful academic discourse, while protecting everyone freedom of speech. @bandlersbanter @ADL @newzionists

7.30.2022 8:39am - (1/2) There is a similarity of music between Simpson's paradox (bad for men, bad for women, good for people) and the two-slit experiment (classical pattern from left slit, same from the right, but interference when we don't know which slit it is). However, the two-slit findings
7.30.2022 8:39am - are observed empirically, whereas Simpson's miracle drug does not exist; it only appears to reverse effect direction due to our misinterpretation of what "bad for" and "good for" mean. No reversal can be observed once we properly interpret what we mean by "good" and "bad".

7.30.2022 8:03am - A congresswoman who does not grasp how her persistent dehumanization of Israel weakens, if not negates the credibility of her arguments for alleged dehumanization of Palestinians.

7.29.2022 10:14pm - (Replying to @BabakWork) Analogies to quantum physics may have their place, but not when we are trying to capture human intuition, which is at odds with the predictions of quantum physics.

7.29.2022 4:52pm - (Replying to @artistexyz @GunnarBlohm and @TRyanGregory) A parsimonious representation, from which one can deduce a compendium of facts is still called "knowledge". This is the convention in AI, and this is the reason why we need to distinguish "explicit" from "implicit" knowledge.

7.29.2022 3:21pm - (1/2) Good question, the answer to which is elaborated in #bookofwhy (p. 200) and wikipedia, but can be seen clearly from the fact that human intuition, across all cultures and persuasions, rejects the possibility of a drug that is bad for women, bad for men, but good for people.
7.29.2022 3:21pm - Such impossibility is a theorem in causal-calculus, but not in probability calculus. In other words, you cannot prove this impossibility using the language of probability theory alone, without invoking causal assumptions and causal logic. See

7.29.2022 1:49pm - (Replying to @CresaKay @GunnarBlohm and @TRyanGregory) I've been meddling in this playground for quite some time, and it hasn't mislead me even once. I would dare call it "safe".

7.29.2022 8:20am - (Replying to @MichellePearl6 and @EpiEllie) Don't worry, IT was a male dog at birth.

7.29.2022 7:14am - (Replying to @OlaSAhmad_) The good people at @IJCAIconf promised to post it.

7.29.2022 7:12am - (Replying to @juan_pigurina) When was it that you felt uncomfortable speaking the language of causality? Simpson's paradox proves that our mind is a causal machine, not a probability machine. If it weren't for the invention of statistics (Pearson, 1890's), we would never stop speaking cause and effect.

7.29.2022 5:53am - If I were the @USAmbUN , I would simply address what "cannot go unaddressed" and call Miloon Kothari to resign disgracefully, both for what he said, and for not demanding that a UN member State that threatens the destruction of another member State be stripped of its membership.

7.29.2022 5:08am - Swayed by your recommendation, I took another look. Boy, either this book improves each time I read it, or I am becoming more addicted to commonsense!

7.28.2022 4:56pm - (Replying to @FredrikHeintz) Can't causal graphs represent temporally dependent variables, sequential planning, etc? Here is an old paper

7.28.2022 4:50pm - (Replying to @soboleffspaces) How about just commonsense? We do not need mania to be sensible.

7.28.2022 4:47pm - The reason I believe the next revolution will be "more impactful" is that it will rally all the smart data science folks behind it, all 105,980 of them, not just the heretic few who are currently working on causal inference. See

7.28.2022 1:13pm - Overwhelmed by the friendly reception of my keynote talk @IJCAIconf today, in Vienna, I am happy to share my slides for our readers: A video of the 2nd part of my talk, separately recorded, will be available on @IJCAIconf website.

7.27.2022 1:02pm - (Replying to @dylanarmbruste3) It is not really an "idea" but the "heart" of causal inference, without which causality is reduced to statistics. That's why statisticians like Rosenbaum, Imbens and Rubin love it, at the cost of being unable to solve the toy problems I present here:,

7.27.2022 1:25am - (Replying to @PHuenermund) Yes. The problem Manski complains about is "missing data", not "measurement error", but his aversion to using graphs is universal, stronger even than Imbens and Rubin.

7.27.2022 12:32am - (Replying to @PHuenermund) Manski does not like Rubin's imputation, for reasons different than ours: But when he talks about "credible assumptions" you can tell he hasn't seen "missingness graphs" yet, without which one cannot judge the credibility of one's assumption. Try it!

7.27.2022 12:30am - Thanks for setting up the audio-visuals for my talk tomorrow, and for alerting me to update my slides. See you at @IJCAIconf 2022.

7.26.2022 7:05pm - (Replying to @erikbiz and @JewishJournal) Not only Arab countries are "threatened", every Palestinian person is personally threatened by the presence of a sovereign Jewish state, no matter how far, no matter how small, no matter how meek.

7.26.2022 6:21pm - Education Dept. to Investigate USC’s Handling of Antisemitic Harassment Against Rose Ritch via @jewishjournal
7.26.2022 6:21pm - An endless cycle: A university is investigated for harassing a Zionist student, and responds with "Look how great we are in talking against anti-Semitism." Moreover, Jewish leaders thank them for being invited to a "committee", instead of demanding they learn to spell "Zionism".

7.26.2022 10:04am - (Replying to @papa_cientifico) In a specific example? Or in a formal model?

7.26.2022 5:03am - (1/3) Yesterday's article by @DrMichaelOren mentions a comparison (made in @TheAtlantic ) between the death of #ShireenAbuAkle and the premeditated murder of my son, Daniel Pearl, in 2002. It compels me to reiterate my reaction to such outrageous comparisons:
7.26.2022 5:03am - (2/3) They highlight the difference between a professional journalist, who never forgets the human face behind the news, whatever the news, and a propagandist, who consistently dehumanizes one side in a conflict. I have humbly asked: In her 25 yrs of covering the Middle East,
7.26.2022 5:03am - (3) has Abu Akleh once described an Israeli child or a grandmother or any human feature of 8 million people trapped in a conflict? If the answer is NO then, regardless of the tragic circumstances of her death, describing her as a model journalist degrades my son's profession.

7.25.2022 6:28pm - This week's harvest of Causal Inference papers is listed here:, for your scrutiny, enjoyment and panoramic overview of where the field is going.

7.25.2022 10:49am - Why I am helping to start a new Jewish Studies Network? Adam Fuller speaks for me. Plus, I have seen 1st hand what happened to Jewish Studies at UCLA and neighboring campuses - our students deserve scholarly alternatives. @GillTroy @EinatWilf @blakeflayton

7.25.2022 9:41am - The celebration cannot be complete without Steve Zipperstein's in-depth analysis of the legal and historical forces at play:

7.25.2022 9:31am - Naïve me! And I was taught listening to the other side is a moral duty.

7.25.2022 6:57am - (Replying to @DKedmey) That's a tough one. Partly because I have zero experience with Times Square campaigns, and partly because passengers speak Rung 3 routinely, e.g., "Are you sure, doctor, that this surgery is worth my risk?" Nothing to be intrigued by.

7.25.2022 6:47am - (Replying to @artistexyz @maxkasy and @HarvardEcon) It's worse than geometry without figures. We can do geometry without geometrical figures because our assumption are defensible (eg. Euclide's axioms). In Potential Outcomes, the assumptions are conditional ignorability assertions, the plausibility of which humans cannot judge.

7.24.2022 10:23am - (Replying to @DavidDeutschOxf @artistexyz and @paul_snively) Confessing ignorance: Where can I find what Everett says about "Potential Outcomes"?

7.24.2022 9:22am - (Replying to @artistexyz @paul_snively and @DavidDeutschOxf) I once lamented thus: Probability function P is a powerful thing to have, it even tells us, on its own, how P ought to change when an observation X=x is received. Unfortunately, it does not have the power to tell us how P ought to change when an intervention do(X=x) is enacted.

7.24.2022 7:30am - (Replying to @NeilHudsonC_R @ToKTeacher and @DavidDeutschOxf) The notion of "intervention" is not defined in probability theory. If it were, then causality would be reducible to statistics and "correlation does not imply causation" would be false.

7.24.2022 5:40am - (Replying to @DavidDeutschOxf) A similar obstacle plagues the more modest endeavor of mimicking human thoughts (as opposed to seeking general theories). Subjective probabilities, when applied to explanatory theories, do not peak with growing evidence, so what's the point in being Bayesian? The notion of "intervention" is not defined in probability theory. If it were, then causality would be reducible to statistics and "correlation does not imply causation" would be false.

7.24.2022 5:03am - (Replying to @paul_snively and @DavidDeutschOxf) To the best of knowledge, Jeffreys/Cox/Jaynes never dealt with counterfactuals, hence, they could not distinguished Theory-1 from Theory-2 if the two gave rise to same probability. IOW, they missed the interesting part of scientific development.

7.23.2022 11:14pm - A harsh indictment of American higher education. @HusseinAboubak , how about sharing a few concrete incidents from your experience at GWU, to see if they match with mine or my students.

7.23.2022 1:23pm - How lovely are your tents, O Jacob; your encampments, O Israel! How safe is your journey, planet earth; Behold, your guardian angels from above.

7.23.2022 6:25am - @DavidDeutschOxf argues that probabilities may be harmful if our goal is to capture truth in the physical world. I am arguing that probabilities are insufficient if our goal is to capture how people think about the world. The latter defines what AI is trying to accomplish.

7.22.2022 10:23pm - (Replying to @soboleffspaces and @DavidDeutschOxf) I am referring to statements like (Pinker p.149):"the rule saying how much to revise our probabilities (change our minds) when we learn new fact." Equating "our minds" with "our probabilities" is seductive but wrong; much of our mind (eg, counterfactuals) lies beyond probability

7.22.2022 4:27pm - Have a great weekend everyone and, as the Shabbat is descending upon us, sing with me: Shalom Aleichem Malachei Hasharet. (Have peace upon you, angels of His message) Composed: Tsfat, Galilee, 16th Century

7.22.2022 11:29am - (Replying to @DeborahProudJew @AbigailShrier and @womenmarch) My heart goes for her - Saint Mother Teresa.

7.22.2022 9:57am - A youtube video of a superb talk by Steve Zipperstein (on one-hundred years to the Israel-Arab conflict) is now available here: I've vowed to memorize all the dates that Steve mentioned, before my next debate with the "new historians - BDS style".

7.21.2022 4:10am - (Replying to @stephensenn and @smueller) We do understand the problem intuitively, but when we speak about "calculus", we need to translate our understanding into formal specification. That's why I am interested in how Nedler specifies the two designs. BTW, would things change if we replace "centers" with "age-groups"?

7.21.2022 2:03am - If you are attending the ICML conference, hungry to learn about non-iid data, don't miss our paper: It will be presented by @zcccucla , at Session P15.

7.21.2022 1:52am - I haven't mentioned other gold mines, because I assumed RCT trialists are interested only in Average Treatment Effect, nothing else. For enlightened trialists interested in distinguishing "harm" from "benefit", we have a 3rd gold mine -- Rung-3 analysis:

7.21.2022 8:07pm - New Poll: Andy Levin was not very smart when he endorsed Rashida Tlaib and thus turned this election into a one-issue race: Whether she is fit to serve in the US Congress. He miscalculated how strongly voters feel against Zionophobic racists.

7.21.2022 2:15pm - (Replying to @GadSaad @JoeBiden and @KamalaHarris) As an American and an uncompromising atheist, I pray to the All-mighty! Oh God! Not @KamalaHarris . This country needs a Golda Meir!

7.21.2022 2:03pm - (Replying to @AndrewPGrieve) Well put! Carefully avoiding the phrase "the true value".

7.21.2022 1:56pm - (Replying to @exactsake) I don't know how to answer your question, because you are using probabilistic terms and, we know, they are orthogonal to causality. "No causal in, no causal out" (Cartwright)

7.21.2022 11:34am - (Replying to @mishtal) This was one land auction I always kill myself for not bidding on.

7.21.2022 7:34am - (Replying to @GedigMoritz) You start by defining what you want to "find out". Note that C is conditional dependent on A, given B. Note also, that this example illustrate the boundary between Rungs 2 and 3. (see Causality page 24)

7.21.2022 7:26am - You are right! It was July 21 2002, 20 years ago that Bernard Lewis wrote this Essay for our "I am Jewish" book. It may have been his influence that made me avoid the word "Judaism" in favor of "Jewishness" - the former is overly theological.

7.21.2022 7:34am - (Replying to @Meh3625555 and @PosenIzzy) My problems with EDI: (1) No list of "ethnic groups" is available. (2) "Jews" are not mentioned in ANY of their official documents. (3) If we demand Jewish representation, they will choose JVP, which is worse than no Jews. Campus discrimination targets Zionism, not Judaism.

7.21.2022 6:17am - (Replying to @RachelMiniK and @PosenIzzy) You will be surprised to know that ALL Diversity officials in the US think that way. I haven't met ONE (I mean ONE) that understands what "diversity" is, namely, that excelling & contributing despite ethnic discrimination is more inspiring than color-based statistics.

7.20.2022 9:23pm - Tomorrow morning, 9am Pacific, Steve Zipperstein will be giving a talk on the 100th anniversary of the British Mandate for Palestine (july 22, 1922). Here are the details, in case you're interested in a thorough understanding of the ME conflict

7.20.2022 7:27pm - Congratulations to @bschoelkopf on a well-deserved recognition and, more importantly, on a breakthrough 2012 paper that has built bridges between causality and machine learning.

7.20.2022 7:00pm - Engineers on the brink of extinction threaten entire tech ecosystems via @theregister
7.20.2022 7:00pm - As a PhD in electrical engineering, I feel sad to see this field on the brink of extinction. I've always attributed my love for science to my excellent EE training. But the show must go on. Are we about to see similar extinction in data-centric enterprises?

7.20.2022 6:37pm - @eliasbareinboim shares the slides of his ICML tutorial on fairness: It contains a critical appraisal of the fairness literature, and solutions based on the logic of SCMs: My Question: Is there a non-causal "fairness"? #datascience

7.20.2022 6:04pm - The instinctive commitment of Jewish leadership to the protection of religious minorities comes from centuries of history. The enigma is why it took them two decades to find out that organizations like CAIR, MAS and MPAC have nothing to do with the protection of minorities.

7.20.2022 1:57pm - Tsadik Katammar Yifrach **** The righteous blooms like a palm tree K'Errez Balvanon Yisgeh **** Like Cedar in Lebanon climbs he, LeHagid Ki Yashar Adonay **** To tell us how fair God must be Tsuri V'Lo Avlata Bo **** My rock, No wrong in him.

7.20.2022 6:21am - (Replying to @ewerlopes and @PHuenermund) Can you narrow down the "topic" to a twittable size? Else I'll have to send you all the papers I wrote since 1980. Here they are:

7.20.2022 6:15am - (Replying to @jonathanrichens @RWJE_BA and 9 others) If you take SCM to be "mechanistic," I'll agree, otherwise, I would love to see some example of "mechanistic assumptions" which are not captured in SCM.

7.20.2022 3:45am - These two gold mines could also rescue experimental economists, many of whom believe they don't need a model once they find an IV (eg, But, DAGs truly life-saving gold mines are recognized by confounding-challenged economists, in non-experimental setups.

7.19.2022 7:22pm - (Replying to @elderofziyon) They don't get it. It is so much easier to forgive an anti-Semite than a Zionophobe. The former evokes stories of unfortunate upbringing, religious prejudice, alcoholic parents etc. The latter are calculating eliminationists with no excusable circumstances -- totally despicable.

7.19.2022 4:37pm - Begging the question: Any reason they can't do the same with women's right to abortion?

7.19.2022 4:20pm - (Replying to @Dem_Bitz) Great painting. Thanks for posting.

7.19.2022 1:21pm - FYI, the harvest is actually richer than I thought. The link is the same,, but the list is now 8 instead of 5 pages long. Let us know (in our input-output language) if you think any of these papers is a breakthrough that should not be missed.

7.19.2022 9:24am - FYI, this week's harvest of causal inference papers: As you can see, it's richer than usual, presumably due to the large number of conferences this time of the year. Or, perhaps our friends in #datascience & #EconTwitter are sensing where the future lies?

7.19.2022 6:52am - The Modern Language Association, Antisemitism and Anti-Zionism –
7.19.2022 6:52am - Every academic scholar or educator should take careful notes of the techniques used by Zionophobes to take over academic institutions like MLA, and turn them into one-issue, foreign-policy mouthpieces. The main enabler: Faculty silence. Any members of MLA among our readers?

7.18.2022 11:46pm - (Replying to @Amol_Sharma) I heard they banned it in Iran.

7.18.2022 11:31pm - DAGs have two gold mines to offer RCTs trialists: (1) Cleaning up RCTs imperfections (eg, selection bias, residual confounding, placebo effects...) (2) Integrating data from multiple sources (eg. multiple RCTs, experimental and observational studies) #datascience #Bookofwhy

7.18.2022 10:21pm - (Replying to @good2epic) Strange, my biggest complaint in DS/ML in industry is almost opposite to yours; too many think probabilistically, at whatever level, and few if any think about the decisions they are about to make and how the data impact that decision.

7.18.2022 10:15pm - I've scanned the TOC and Introduction and I have strong suspicion that the course stays on Rung-1 of the Ladder, namely, limited to prediction and association. No harm in it, but students who are looking for Rung-2 and Rung-3 may be disappointed. #datascience

7.18.2022 4:52pm - And here another interesting paper on "harm" and "benefit", this time conditioned on variables that, in themselves, may be affected by the treatment: #datascience #economics #MachineLearning #Bookofwhy

7.18.2022 2:56pm - For readers interested in the epistemology of causal inference, I've revised my commentary on Dawid's "Decision Theoretic Foundations of Statistical Causality" and have posted a clean copy here

7.18.2022 2:47pm - (Replying to @tom4everitt @RWJE_BA and 8 others) Thanks for posting this paper about "harm". "Harm" and "benefit" have been the subjects of our latest papers on "personalized decisions" and Would be interesting to compare results.

7.18.2022 2:56pm -

7.18.2022 1:29pm - (Replying to @soboleffspaces and @IJCAIconf) Yes, but what is their motivation for relying ONLY on randomization and resisting any knowledge that the investigator may have acquired by experience, education and judgment.

7.18.2022 7:14am - (Replying to @GinoDelFerraro) Progressives do not count wounds, we count choices.

7.18.2022 7:05am - The ship that opened the world's eyes to British Govt treatment of Holocaust survivors, and swayed public opinion in favor the UN vote (33:13) for partition of Palestine (Nov 29, 1947) despite Arab's promises of "War of Extermination" (Azzam Pasha, Oct. 11, 1947).

7.18.2022 6:45am - (Replying to @MissDiagnosis and @IWF) Pres. Roosevelt died in 1945.

7.18.2022 6:39am - (Replying to @mishtal and @freemonotheist) Those who have no history must fake one, even at the cost of being caught.

7.18.2022 1:28am - (Replying to @maradonasattva and @BillDonovan1234) Judaism without Zionism is like Christianity without a Cross. Not entirely infeasible, yet entirely a-historical and pitifully crippled.

7.18.2022 1:17am - @POTUS main achievement: allaying fears that leaving #Afghanistan was phase one of leaving #MiddleEast. Saudi "no normalization before X" is actually hopeful, b/c X was phrased as "agreement" on 2-states, not implementation thereof.

7.17.2022 5:54pm - (Replying to @soboleffspaces and @IJCAIconf) Who are the algorithmists?

7.17.2022 10:16am - (Replying to @Wurmserscribit) I know that Biden knows who the oppressor is in the Israel-Palestinian conflict. I assume therefore that he chose this rhetoric as the only common denominator he could find with his audience, who, btw, also knew what he thinks of them.

7.17.2022 10:10am - Extremely interesting paper: BTW, what is the current status of the statistical marginal problem?

7.17.2022 7:40am - That's why we call it "data science" - a two-body enterprise - the data "suggest" and the science "reasons".

7.17.2022 7:05am - (Replying to @KPMDas) Many tell me: "OK, so the NYT has an obsession with Israel but, by and large, it is still "good journalism"". I say: I can only judge them in areas that I've some competency, Israel and AI, and if they are glaringly so anti-professional in the former, I can't trust them anywhere.

7.17.2022 6:41am - (Replying to @HillCitizen) "Saying" and "documenting" quantitatively are too different things. I have some fairly rational friends telling me: "I read NYT and haven't seen anti-Israel bias". Seeing the word statistics may convince them.

7.17.2022 6:37am - (Replying to @EvanHoffman) Unfortunately, terrorism has been normalized in "freedom fighters" land, so it carries zero weight in public discourse.

7.17.2022 6:33am - (Replying to @adamalmo) These two papers summarize the main thrust of my talk:,

7.17.2022 4:15am - Comparing Irish and Palestine? I don't recall Irish Catholics shouting "London is a stolen land!" or "Dismantle Great Britain". Have you?

7.17.2022 3:17am - To restore your reputation, you must remind the public what your values are, and to state explicitly that they include dignity and safety for Anna's brothers and sisters who survived the camps and have found hope and new life in Israel. Vagueness on values projects empty values.

7.17.2022 2:41am - I am delighted to hear that Michael Waldmann is hiring a postdoc for the project “Mechanisms, Capacities, and Dependencies: A New Theory of Causal Reasoning”, funded by the German Research Foundation (Koselleck program). See for more info.

7.17.2022 12:15am - Top Israeli Daily’s Exposé Paints Troubling Picture of New York Times’ Israel Coverage
7.17.2022 12:15am - The poetic power of statistics. Here are irritating facts that we (readers of NYT) encounter on a daily basis, but when quantified and compiled into statistical patterns, unveil a purposeful design.

7.16.2022 11:48pm - (Replying to @soboleffspaces PS/IPW has also been used in #MachineLearning as a disguise for doing statistical estimation and calling it "causal inference". #Bookofwhy, #DataScience

7.16.2022 5:21pm - Philosophers, linguists, and XAI practitioners would find this paper of interest. It deals with evidential and causal conditionals and proposes a system that handles both: I need to understand why my proposal in is inferior.

7.16.2022 12:40pm - The Daniel Pearl Foundation is not as powerful as the @AnneFrankTrust but, believe me, we screen our speakers very carefully. See

7.16.2022 12:12pm - This is precisely what fascinates me about Diplomacy. What you say or do does not represent your beliefs; it represents who you wish to please. "Peace" pleases Israelis, "UNRWA" pleases AOC, and both sound so noble.

7.16.2022 3:57pm - (1/3) The habit of assuming away the identification problem and treating matching and PS methods as if they reduce confounding bias comes from surveys like this one: (from ACM Trans. on Knowledge Discovery from Data) which is wedded to Potential Outcomes
7.16.2022 3:57pm - (2/3) and mentions SCM as an "equivalent approach". Missing is the warning that logical equivalence is not conceptual equivalence: In particular, that the PO approach "forces investigators to articulate assumptions whose meaning they do not comprehend,
7.16.2022 3:57pm - (3) whose plausibility they cannot judge, and whose statistical implications they cannot predict." See: One explanation for the prevalence of this habit is that scientists fear transparency of assumptions they must defend. Habits are preferred.#datascience

7.16.2022 11:55am - It's true, with one exception: The official title of my talk reads: “What is Causal Inference and Where is Data Science Going?" See you there. #Bookofwhy, #datascience, @IJCAIconf

7.16.2022 7:42am - @julius_adebayo sharply dissects the XAI enterprise to its core, but his conclusion: "explainable ML = model debugging" still leaves an elephant in the room: A model of what?" of the ML program? Or a model of reality?. Examples abound where the former doesn't match the latter.

7.16.2022 5:30am - (Replying to @yudapearl @roydanroy and @nickchk) To answer your question, Yes. David assumes away the identification problem (using the magical un-confoundedness assumption), which reduces causal inference tasks to statistical estimation, and gives ML folks the impression that "it's all in the data". Not advisable.

7.16.2022 5:13am - (Replying to @roydanroy) I am pointing to backdoor admissibility because it is necessary for deciding what features to match on, and which ones to exclude from matching. #Bookofwhy has example where matching on all features leads to wrong conclusion. @nickchk makes this point nicely in "The Effect".

7.15.2022 6:33pm - (Replying to @jeffreywatumull and @DavidDeutschOxf) My first reaction: Bacon, Hume and Popper tried to capture the way science works, this paper tries to capture the way machine learning works. I do not see the connection given how far machine learning is from emulating scientific thinking. (e.g., counterfactual thinking)

7.15.2022 1:54pm - (Replying to @GadSaad) Satirical or not, this outspoken protects the honor and credentials of a tenured UC professor. Hats off, outspoken!!!

7.15.2022 9:27am - In Bethlehem, Biden pledges to keep up efforts for Israel Palestine peace
7.15.2022 9:27am - Biden: "Even if the ground is not right at this moment to restart negotiations" It seems that I would never be a diplomat. I would never be able to control myself from telling Abbas "WHY not restart now?", and what he can do about it. Diplomats can ignore giant elephants.

7.15.2022 4:40am - Try reading some of the graphical causal inference literature and I bet your enthusiasm will increase 10 fold. E.g.,, Specifically, introducing matching and propensity score w/o backdoor admissibility is not advisable. #datascience

7.14.2022 4:01pm - (Replying to @Eve_Barlow) Here is the irony. Jews get their strength from history, and their history has culminated in Israel. The poor souls that reject Israel have lost the strength to stand up to their bullies -- a vicious circle. I empathize with their dilemma here:

7.14.2022 1:41pm - (Replying to @artistexyz and @soboleffspaces0 I tried to follow it, but I couldn't. Do you have a gentle tutorial to the method?

7.14.2022 12:04pm - (Replying to @DKedmey) I think it worked; the smearing guy stopped. I believe these guys simply do not see the difference between talking and smearing; typical of Al-Jazeera news coverage, where the facts/accusations ratio is below 10%.

7.14.2022 9:46am - (Replying to @WelshBrendan2) I understand. The question is whether a peace-seeker like yourself would be willing to accept the harsh facts as learned by another peace-seeker (myself) who has been searching for a moderate voice the past 75 years -- there isn't any, and it ain't "climate." Hard to swallow!

7.14.2022 9:33am - (Replying to @WelshBrendan2 and @CotlerWunsh) So you, personally, believe Israel "operates under an apartheid model". Am I right? You do not buy the possibility that the "apartheid model" was attached to her out of pure malice?" Am I right?

7.14.2022 9:12am - (Replying to @soboleffspaces @JohanDH2O and 2 others) For me, getting meaningful definitions of NIE and NDE was more important that preserving the additivity property, which I thought to be a disposable feature of linear systems.

7.14.2022 7:56am - (1/2) Certain breed of Jews, mainly those who believe that Israel is a colonial apartheid state, get extremely upset when called "Ex-Jewish". Some even call me "anti-Semite" for suggesting their exclusion from the Jewish fold. I'm willing to reconsider and call them:
7.14.2022 7:56am - "colonial apartheid Jews", except they want to be treated as mainstream. I once proposed "The New Marranos", and explained why they should not be given a podium at the Bimma - our universities. Today, I much prefer "Ex-Jewish", or "Yester-Jews".

7.14.2022 6:50am - Finally, the Jerusalem Declaration signed! The key sentence: "The United States stresses that integral to this pledge is the commitment NEVER TO ALLOW Iran to acquire a nuclear weapon, and that it is prepared to use ALL ELEMENTS OF ITS NATIONAL POWER to ensure that outcome."

7.14.2022 4:03am - Confession, I could not hold my tears when President Biden listened to these two ladies and, ignoring all the "time is up" hints from his entourage, including Blinken, continued to listen to what his visit meant to people who survived Auschwitz and have made their home in Israel.

7.14.2022 3:44am - (Replying to @tomgoldsteincs) This is a legitimate question for every acronym in machine learning. I've asked it on Stanford's "Foundation model" with no answer in sight. The problem is that acronyms are attached to successful programs, not to scientific principles. #datascience, #Bookofwhy

7.14.2022 3:30am - (Replying to @hari_hari) I am searching for the fundamental ideas, and all I can find are different labels of the same problem, coined by different groups in DARPA .

7.14.2022 3:18am - I don't know much about pregnancy, borders or obesity, but have had some fun in science, till yesterday, when I read that it's more racist than Trump, w/ all its equations named after white male adults; Galileo, Newton, Maxwell, Einstein -- incurably oppressive to most students!

7.13.2022 8:40pm - (Replying to @shagunsodhani) To those versed in causal inference, what you are describing seem to be minute differences of emphasis in several versions of "Data-Fusion" theory I wonder if lifelong folks looked into the solutions offered by Data Fusion.

7.13.2022 7:23pm - President Biden has given Jewish students their first line: "You don't need to be a Jew to be a Zionist." Their next line should be: "Dear Chancellor, You can't bring yourself to state that Zionist students are more than welcome on campus? Aren't you the one who sets norms here?"

7.13.2022 6:34pm - I've always wondered, but was afraid to ask: How does "lifelong learning" differ from “transfer learning,” or “domain adaptation." In other words, would there be a problem of "lifelong learning" if domains stayed constant? Or, won't LLL be solved if we solved "transfer learning?

7.13.2022 6:10pm - Congratulations to @tgjelten & @MomentMagazine on winning two first place AJPA awards, and for the excellent investigative story on the plight of the Uyghurs in China; a topic that ceased to interest social media activists since the start of the war in Ukraine. @NadineEpsteinDC

7.13.2022 5:31pm - (Replying to @malickheydar and @Harvard) Me too. I just gave a copy of #Bookofwhy to my dentist, dedicated with: "Easy on my wisdom tooth, doctor, it ain't dead."

7.13.2022 5:18pm - (1/2) Thanks for re-twitting this timely post. As more economists enter the world of DAGs, I sense their genuine appreciation of DAGs as a communication language, but insufficient appreciation of DAG's computational power. This post shows that even in
7.13.2022 5:18pm - simple IV models, you can't go very far without DAGs. Imbens's claim that economists indulgence in simple IV's absolves them of needing DAGs is grossly misleading. Same with Heckman's disposal of d-separation @causalinf , @nickchk

7.13.2022 3:54pm - (Replying to @RitchieTorres) I salute you @RitchieTorres for articulating so clearly what the nature of BDS is, and what your vision is of ME peace, based on "co-existence", a word that no BDS crony has ever uttered. You have my back.

7.13.2022 3:42pm - (Replying to @mynameisjro and @RitchieTorres) I salute @RitchieTorres for speaking for our people, because those EX-Jewish New Yorkers who believe that Israel is an apartheid state, and that BDS is not eliminationist, have abandoned us, in need of a non-Jewish Congressman to counter our predators in the US Congress.

7.13.2022 7:19am - I once put it this way: "The landscape has become a scripture". Lord Balfour put it even nicer: ".. intimately bound up with a petty territory wedged in between States more powerful far than it could ever be". See

7.13.2022 12:50am - (Replying to @GadSaad and @AyannaPressley) I do have the capacity of answering this question, but I'd rather keep mouth shut for fear that "my line of answering" would render me vulnerable to accusations of transphobia (or truthphobia) hence indirectly responsible for suicides among bald white men, a protected minority.

7.12.2022 4:25pm - (Replying to @adamalmo) These two papers summarize the main thrust of my talk:,

7.12.2022 4:20pm - You caught me off guard. I was going to jump out of my skin and shout: "Peace! finally!" before I saw the humble #WishfulHeadlines. Back to the real Mahmoud Abbas.

7.12.2022 4:15pm - This red carpet is longer than River Jordan It's purpose to generate media focus on the real issues of ME, keeping @AJEnglish and other noise makers out of the conversation. My question: Will Biden treat Palestinians as responsible agents, or just reinforce their passive posture.

7.12.2022 8:57am - Here are some facts that even Israel's worst enemies know to be true, and which they find to be just nasty obstacles in their heroic mission to prove the opposite.

7.12.2022 7:52am - (Replying to @richard_landes) A Palestinian never says or "I recognize Israel", it's always: "Arafat did, in 1990", forgetting that Israelis have learned something in the past 30 years, especially to demand a fresh Arabic copy of the PLO charter, w/ the words "right to exist" spelled out for children to read.

7.12.2022 6:56am - Glad the Academy of Management is about to enter the age of causation, through the very practical and omnipresent question of Good and Bad controls, a question that graph-averse economists have chosen to avoid rather than address.,

7.12.2022 6:31am - This paper, though written in 2014, should be read today in other fields undergoing methodological transformations: especially those laden with paradoxes or controversies (who isn't?)

7.12.2022 4:09am - The market is waiting to see how innovation economists circumvent the infinite inertia of mainstream economics.

7.12.2022 3:30am - Welcome to peace-dodging land! Here is a billboard that would be torn down immediately by our neighbors, yet is put up repeatedly as if we do not know; as if they never told us; as if pretending we are deft would make them change their dream.

7.12.2022 1:27am - How can one tell when Causal Inference (CI) goes mainstream? (1) When universities dedicate buildings as CI-Centers. (2) When recruiters demand CI-qualifications. (3) When US patent office grants patents to CI-ideas. Beware! The latter just occurred:

7.11.2022 6:45pm - For an academic organization (Association for Israel Studies (AIS)) that counts Zionophobes like @Ian_Lustick among its former presidents, breaking ties with MESA must be a traumatic catharsis. MESA can safely change its name now to IMESA - the Irrelevant side of ME Studies.

7.11.2022 1:21pm - FYI, this week's harvest of causal inference papers:

7.11.2022 9:35am - This is a fair comparison. Biden should meet the parents of Malki Roth in Jerusalem. They have been waiting 21 years for justice, and Malki, their slain daughter, was a symbol of goodness and joy.

7.11.2022 9:22am - Informative narrative of the history of Czech-Israel relations. Thanks

7.11.2022 7:29am - (Replying to @DerriereLaLune and @rcguo) Interesting! First time I see it. But I don't buy the argument: "The derivation does not use an intervention on the mediator, which in many situations would not be possible in reality." Derivational step need not be "possible in reality". Best example: Complex numbers in math.

7.11.2022 7:05am - This is all true. I'm honored to be speaking again at IJCAI (first time was in 1983 and to challenge the direction "big data" is taking. No, I'm not flying to Vienna, but will make my Zoom screen as big as possible. See you there.

7.10.2022 5:58pm - (Replying to @elderofziyon @BenWeinthal and @ShuratHaDin) I'm glad nevertheless that a serious complaint has been filed with the International Criminal Court against the PLO; people just don't get the idea that the evidence is not clear cut, until legal complaints are filed, each supporting one side of the story.

7.10.2022 2:12pm - (Replying to @HaimHabulai) Thanks for the correction, I didn't know it. Kol Halomed MeChavero Davar ....Rabbi

7.10.2022 8:15am - Israeli president to meet with Czech counterpart in Prague via @JNS_org
7.10.2022 8:15am - Whenever I see a Czech flag, I can't but remember Jan Masaryk, the Foreign Minister of Czechoslovakia, who agreed to send Israel 4,000 rifles, thus saving us from a genocidal attack in 1948. My basic training (1953) was still with Czech rifles. Masaryk Square now crowns Tel-Aviv.

7.10.2022 7:35am - Jewish father files lawsuit over discrimination at Los Angeles private school via @JNS_org
7.10.2022 7:35am - As much as I hate law suits, this one is absolutely necessary to encourage parents everywhere to standup to recent takeovers of our high schools by EDI priests, who do not value Jewish students as a distinct minority that can contribute positively to other minorities on campus.

7.10.2022 5:58am - (Replying to @elderofziyon @BenWeinthal and @ShuratHaDin) I'm glad nevertheless that a serious complaint has been filed with the International Criminal Court against the PLO; people just don't get the idea that the evidence is not clear cut, until legal complaints are filed, each supporting one side of the story.

7.10.2022 5:42am - Refreshing sight, even to a Californian like me. My mother used to take us (me and my younger brother) to this very beach every week. You see, she was Polish, and believed religiously that sun rays and sea waters is good for your health. Wow she was right!

7.10.2022 5:28am - (Replying to @kareem_carr) What about for the kind of data scientist that's more like a data-scientists?

7.10.2022 5:16am - Most economists use potential outcomes in conjunction with instrumental variables, not realizing that both are derivatives of SCM, Some begin to use DAGs but, to make them look home-grown, strip them of their main powers.

7.10.2022 4:53am - (Replying to @blakeflayton) We got here (partially) because even Zionist writers use the term "anti-Zionism" as if it was some sort of a legitimate political opinion, instead of using "Zionophobia", which highlights its racist, genocidal and pathological character.

7.10.2022 4:42am - (1/2) Our next heroic effort is (again) in economics. My attention is captivated by a BA thesis on the mediation fallacy and the analysis of police violence: Considering the graph-averse culture dominating economics, and the risk-averse, sheepish style of
7.10.2022 4:42am - most theses (including PhDs) it's both mature and heroic to assert: "DAGs’ main strengths are their accessibility, clarity, transparency, and testability, and the current skepticism by economists against their use is largely unfounded." Prediction: Johanna Barop is going places.

7.9.2022 11:44pm - (Replying to @EricJGreenberg1 @mishtal and @Ixar_Bargains) That was easy. I asked him to sing me a Philistine song, or tell me a Philistine story, or name a Philistine hero he would name his son after, or a Philistine holiday he enjoys celebrating, or, ... or.. It was easy.

7.9.2022 10:13pm - (Replying to @rickwsint and @skdh) Got it, thanks!

7.9.2022 10:11pm - (Replying to @rickwsint and @skdh) I didn't get the better source.

7.9.2022 8:42am - It's called science-centric paradigm:

7.9.2022 8:12am - (Replying to @Ostrov_A @Israel and 10 others0 The one on the bottom left reminds me of Shuck Hapishpeshim (the bed-bugs market) where I bought my first acoustic guitar, in 1976. Glad the place is still striving.

7.9.2022 7:57am - (Replying to @roydanroy) "Swallow", of course, by claim of ownership (called "inductive bias"), but not before ML folks learn to articulate what it is they claim to own.

7.9.2022 7:34am - (Replying to @rasbt) Jobless - not yet, but outdated, on the mark.

7.9.2022 7:31am - Surprise! Ali Khamenei Slams Western Values: A perfect candidate for Dean of Social Science at one of our finest US Universities. The salary and faculty enthusiasm would surely match his current employment benefits.

7.9.2022 6:40am - When ‘Racial Justice’ Means Antisemitism: A Takeover of the Yale Postdoctoral Association
7.9.2022 6:40am - Delighted to see faculty members at @Yale (my semi-alma) speaking out against the Zionophobic takeover of their Postdoctoral Association. Like USC before them, when faculty say "enough is enough", it means: "enough will prevail."

7.9.2022 6:07am - (Replying to @thinkyparts) "Reasoning" entails decision making under murky waters.

7.9.2022 6:04am - This paper by Grosz etal, "The Taboo Against Explicit Causal Inference in NonExperimental Psychology," explains much of the research environment in that field, and why I consider the works of @dingding_peng & associates to be so critical to psychotherapy.

7.9.2022 4:36am - Every country on the right side of reason is a fortress on the right side of humanity

7.9.2022 4:04am - (Replying to @blakeflayton) You are giving readers the impression that Zionism just "does not say they dont have a right". I truth, Zionism explicitly says they DO have this right "even if it compromises Zionism". Why not quote the sources, Ben Gurion, 1931

7.9.2022 3:51am - (Replying to @eIumax) Problematic is the mind that sees the sun, the moon, and living organisms as "problematic"

7.9.2022 3:18am - (Replying to @blakeflayton) Why? Has anyone mistaken it with the civil rights movement of another nation? French? Egypt?

7.8.2022 9:21am - (Replying to @JohanDH2O) Excuse my ignorance, but what ideology does Jordan Peterson represent??

7.8.2022 9:08am - (Replying to @tkaiser_science) Very encouraging, but what is the NAME of this journal? We need to publicize it and commend the editors on being forward looking. (unlike some fields I won't mention).

7.8.2022 8:53am - (1/2) Our next hero welcome goes to an economist (who would ever imagine). Nick Huntington-Klein has a new book which, according to its TOC, promises to deliver economics students from the graph-less torture that their professors went through, and which
7.8.2022 8:53am - (2/2) Heckman calls "Econometric Causality." It starts with research questions and identification, and then goes to the tool box, armed with backdoors, front-doors, perhaps even with do-calculus and the First Law. I've ordered a kindle copy and will let you know more.

7.8.2022 7:23am - (Replying to @MiriamElman @GeraldNGOM and 2 others) Miriam, I need no FYI's. Here is what I posted two days ago: and, you are right, Joe was on the UCLA committee that fought the Zionophobic racism of the Asian-American Studies department, and testified before the Regents. I know we will prevail.

7.8.2022 6:00am - (1/2) The next heroic effort is in psychology. Based on a 2018 paper by Julia Rohrer @dingding_peng My first reaction was: Who is this community of Psychologists who were not convinced by Julia's article? It's hard to believe
7.8.2022 6:00am - (2/2) that they can continue to function in 21st century without causal inference tools. The progress of a field invariably depends on one/two enlightened leaders, so, don't they have any? Kudos to Li Qian and Julia for trying to lift this field out of the stone age.

7.8.2022 5:10am - (Replying to @IllyseusDasein @SamHarrisOrg and 2 others) I do not understand what you are trying to insinuate, but I've known Joe Manson for 20 years. He ain't an oppressor, nor a "self-styled victim", nor any of the adjectives you associate with people you don't like. Please do not use this smear technique on this platform.

7.8.2022 4:48am - (1/2) I would like to dedicate the next few posts to heroic readers who are struggling to introduce causal clarity to fields where such clarity has been lacking. I will start with which focus on neuroscience. My only gripe: they push me into "statistics" and
7.8.2022 4:48am - (2/2) fail to mention that many (if not all) of the diverse types of causes they discuss are now unified through SCM. I never understood those who prefer pluralism on monotheism. See

7.8.2022 3:59am - This film taught me quire a few things I didn't know. But most of all, I liked the dedication to feminists like Linda Sarsour and other fake leaders of the Women March, who continue to peddle their dishonesty to the young and unsuspecting. @ZionessMovement @bariweiss

7.8.2022 3:10am - (Replying to @SAFECUNY) Great! As the first step. The next steps are: 1. CUNY's administrators learn to spell "Zionism" 2. They learn to pronounce it, loud, w/o watching if anyone is listening. 3. They tell the campus what it contributes to diversity and education.

7.8.2022 2:22am - Populist slogans are not supposed to mirror facts or figures; they are made to feed the ignorant and move the mob.

7.7.2022 11:30pm - The next take belongs to the @latimes , whom Californians trust to spot such stories and turn them into a public outcry: "Are we losing our universities?" They used to listen to me, before the woke takeover. Perhaps they would listen to @bariwies.

7.7.2022 11:05pm - (Replying to @KabirCreates @ProfLaurenRoss and 2 others) The beauty of Hume's counterfactual definition of "cause" is that, for 2 centuries, people thought it's final, and no one asked: "So how do you define counterfactuals?". Why? Because counterfactuals are so entrenched in our mind, we thought they're irreducible primitives. Wrong!

7.7.2022 3:34pm - (Replying to @ProfLaurenRoss @KordingLab and @shansiddiqi) I am still trying to understand what people want when they crave for "definition of cause". If they crave for a trick to help them think when they construct a model, then I have a good one: "who listens to whom?". If they seek reduction to probability, we know it's impossible.

7.7.2022 2:57pm - (Replying to @mishtal) Now you see why I play it safe, and call him a "dangerous Zionophobe." He can't sue me for what he is so proud of, though it is ten times more dangerous and certainly more despicable.

7.7.2022 8:33am - As Abdurrahman Wahid said (former President of Indonesia): "Countries that deny the Holocaust also tend to be backward and repressive." Why? "Because those who are dishonest when it comes to the past are hardly in a position to reckon honestly with the problems of the present."

7.7.2022 5:25am - Johnson's resignation speech starts 9:20 into this video Beautiful words. Gloomy future. Hail to Britannia -- God Save the Queen

7.7.2022 12:15am - I can't dive too well, but if anyone is swimming already in the Indian Ocean, I would be interested to examine the earliest counterfactual expression in ancient Indian literature. Similar to the one spoken by Abraham: "What if there were 50 righteous people in the wicked city?"

7.6.2022 7:34pm - A confession of a repentant Zionophobe.

7.6.2022 5:45pm - The New Zionist Congress - 2022 @blakeflayton @newzionists

7.6.2022 5:05pm - Photo of the leaders of Palestine and Hamas, very symbolic, but this will not improve the situation via @ePrimefeed
7.6.2022 5:05pm - "The New Zionist Congress - 2022"

7.6.2022 8:23am - An important page of history. The event that led to Israel's "defensive shield" campaign, which some regard (including me) as the only effective "war on terror" by any country, though still continuing.

7.6.2022 7:36am - (Replying to @EpiEllie @KordingLab and 2 others) Beautiful introduction to a Short Introduction, and very useful, for every CI scholar, to collect this arsenal of examples to communicate nuances and variants of causal thoughts. Thanks.

7.6.2022 3:26am - (Replying to @MechncOfMeaning) I'm only concerned about the last line. do(c) always appears under a probability expression. e.g., P(r|do(c)) = P(r)

7.6.2022 3:19am - (Replying to @intrinsic_motiv and @ylecun) What do you expect to get by implementing this strategy? I am asking, because a theory is now available that can tell you what the strategy can and cannot do. I would recommend that use the theory to guide your implementation. e.g.,, #Bookofwhy

7.6.2022 2:45am - (Replying to @JGreenblattADL @almog_doron and @JewishAgency) Congratulations, Almog, on assuming this leadership. You can count on our help.

7.6.2022 12:31am - Yes, you can get the Solution Manual if you write to my assistant

7.6.2022 12:03am - Can you help me climb this paywall? It tells me: "Access to this article via University of California Los Angeles is not available." Why? I think I've lived a clean life!

7.5.2022 11:52pm - Agree full-heartedly. Unfortunately ML folks are busy building Gothic Cathedrals for a handcuffed God, funding them, and convincing investors that Gods should know how to liberate themselves.

7.5.2022 11:37pm - I would recommend Primer which is available free (chapter by chapter) and logically coherent, starting with the First Law and ending with Tool Kits for Mediation and Attribution.

7.5.2022 11:30pm - (Replying to @giulioprisco @SMcfarnell and 12 others) Interesting conjecture which has skipped my #FreeWill optics. I am inclined to surmise that the philosophical debate is orthogonal to the political, with the latter misunderstanding the former.

7.5.2022 11:14pm - I was right then in predicting that Imbens writings and prizes will entice people to study causal inference the scientific way, starting with the First Law: @causalinf #Bookofwhy @VC31415

7.5.2022 10:46pm - (Replying to @animesh1977) Viewed as a threat, yes, but I don't recall the "clowns" era.

7.5.2022 10:41pm - For readers and teachers of Causality (2009), viewgraphs and exercises are available here: Alternatively, Primer is loaded with exercises:, & a solution manual is available. Please RT, to let others know, especially #Bookofwhy folks

7.5.2022 10:17pm - (Replying to @texan_marshall) I suspect if such a system turns successful, it would spawn a movement to "defund AI," which no AI researcher wishes to see happening.

7.5.2022 4:42pm - When an activist quotes @AlJazeera and says: "It truly appears to be XYZ" you can bet on two things: (1) He knows nothing about what happened, and (2) what happened is definitely not XYZ.

7.5.2022 4:09pm - Defeat Overture INT o2 PCUSA - Sign the Petition! via @Change
7.5.2022 4:09pm - I've signed this petition, because it hurts me to see my friends in the Presbyterian Church bullied by hard core Zionophobes who took over the PCUSA administration. Please join me in defeating overture INT 02.

7.5.2022 3:13pm - (Replying to @equilibriumcow) Yes, this is the page. I'll tell readers to go there. A more gentile technical introduction is Primer It has exercises and solutions.

7.5.2022 5:57am - (Replying to @equilibriumcow) Thanks for the comments, but which "causality" do you have in mind, and which website are you referring to?

7.5.2022 4:02am - I'll second your recommendation. I truly have no idea how many ML folks read #Bookofwhy. There are days when I think: "zero". And then, when I see them struggling to emulate causal thoughts, I think: "99%". Hard to tell.

7.5.2022 1:11AM - To complete my story about the Kielce Pogrom below is a picture of my aunt, Mania (an Auschwitz survivor), during the funeral, consoling families of the victims. (Courtesy of her son, now living in Israel, a proud graduate of my own college, The Technion.)

7.4.2022 9:19pm - (Replying to @Wetassprior and @KeurigBad) No, Z is a son of W, so, conditioning on Z opens the backdoor path X<--U2--->W<--U1-->Y.

7.4.2022 9:12pm - Dan Rather: "The U.S.A. is and always has been an idea more than a place" " Made me wonder: How many societies think like that about their countries? I know Israelis do. But I don't get it from my Canadian granddaughters. Switzerland? Egypt? Anyone knows?

7.4.2022 2:07pm - I've recently took the liberty of calling it "Ladder of Knowledge" rather than "Ladder of causation"; I've seen it govern areas that are not immediately associated with causation.

7.4.2022 1:43pm - On July 4th we realize that the words "defending freedom and democracy" are more than a broken record. It hits you that certain freedoms and certain democracies do need defending.

7.4.2022 1:23pm - This hits close to home. My mother's family is from Kielce. And her sister witnessed the pogrom in her own eyes, miraculously survived, and later eulogized the victims in the cemetery, as was recorded on movie camera. Only one year after liberation.

7.4.2022 11:12am - What a coincidence for July 4th holiday. An important addendum, for American Jews to remember. In 1997, Herzl said:“Zionism is a homecoming to the Jewish fold even before it becomes a homecoming to the Jewish land.” We can see it being incarnated today, in the US.

7.4.2022 10:30am - I'm still surprised that no one in the media raises the question of motive: Why would Israel aim to kill a journalist when she has everything to lose and nothing to gain from such act?

7.4.2022 9:27am - Happy 4th of July! On this day, it is meaningful and relevant to read aloud Longfellow's poem: "Thou, too, sail on, O Ship of State! Sail on, O Union, strong and great! Humanity with all its fears, With all the hopes of future years, Is hanging breathless on thy fate!" How true!

7.4.2022 12:44am - (Replying to @arbutuspointBC) You can tell from his logic that he ain't likely to be a fan of Israel or of other islands of commonsense. @EnglerYves

7.4.2022 12:34am - I have elaborated on these Eight Myths in "The Causal Foundations of Structural Equation Modeling", an updated Chapter for the upcoming R. H. Hoyle (Ed.), Handbook of Structural Equation Modeling. New York: Guilford Press,

7.4.2022 12:22am - (1/2) Toronto logic at its glory: "Irrespective of the veracity of accusations, rallying for the accuser is reasonable!" Imagine rallying against Ukrainian borscht at TO store called “Taste of Ukraine” because Putin called Zelensky Nazi and deemed Ukraine "stolen land."
7.4.2022 12:22am - (2/2) This is the new logic Toronto shoppers are encouraged to follow at a food store called "Taste of Israel". Thanks to @EnglerYves for unveiling this logic, and to @MelissaLantsman for exposing its dangers.

7.4.2022 4:10pm - (Replying to @robertclab) I doubt people who shun DAGs even for simple problems would fall in love with SWIGs when problems get harder.

7.3.2022 11:45pm - (Replying to @arbutuspointBC) Toronto logic at its glory: "Irrespective of the veracity of accusations, rallying for the accuser is reasonable!" Imagine rallying against Ukrainian borscht at TO store called “Taste of Ukraine” because Putin called Zelensky Nazi and deemed Ukraine stolen land. Grand-Logic!

7.3.2022 7:42am - (Replying to @anon61845948 @jrootham and @MelissaLantsman) Beg to differ. An ethnic group that equates its liberation, loudly and unequivably, with the elimination of another ethnic group is not equally deserving of support and respect. Please listen to how THEY define liberation, not how Westerners wish them to define it.

7.3.2022 1:37am - PM Lapid spoke for me when he said:" It was here before us, and be here long after us". I will vote for him, if I'm so lucky as to be in Israel November 1st.

7.2.2022 12:20pm - Every time a reader retweets this interview I am tempted to re-listen to it and then wonder how the interviewer, Guido Imbens, could possibly continue to work with Potential Outcomes instead of switching to graphs. An resolved mystery of scientific evolution.

7.2.2022 11:41am - Israel is the only country whose future depends on its neighbors textbooks more than it depends on their guns.

7.2.2022 7:42am - (Replying to @hdjnicklin) Your logic is right, but missing one fact: Palestinian riots of 1936-1939 forced the British Government to stop immigration from inferno Europe, thus sealing the fate of 6 million Jews who sought refuge in Palestine, the only country where they were welcome as brethren.

7.2.2022 7:30am - (Replying to @dizzdangles and @MelissaLantsman) You got a point, if it weren't for the death-threat encoded by the "Free Palestine" slogan, these "protester were shouting.

7.2.2022 7:17am - (Replying to @jrootham and @MelissaLantsman) Not sure if you seriously do not know what "Free Palestine" means to those who shout it.

7.2.2022 7:11am - @MelissaLantsman , you are talking as if anti-Zionism is an excuse for this form of protest. Is it? Suppose it was purely anti-Zionist, soaked with angelic love for non-Zionist Jews. Would it be less racist? Less morally deranged? Less despicable?

7.2.2022 6:46am - COVID-19 hero: Pfizer CEO Albert Bourla reflects on the past, future

7.2.2022 12:53am - Only a Zionophobe like @KenRoth would retweet this misleading statement. It was not the Presbyterian Church that voted, but its lopsided Committee on International Engagement. The overture will be voted on by the entire membership next week. Hoping Presbyterians defeat INT 02.

7.2.2022 12:25am - I wish left extremist would be as generous and accuse us of things we are proud of.

7.2.2022 12:15am - Jewish Groups Criticize CUNY Chancellor for Not Appearing at NYC City Council Antisemitism Hearing via @jewishjournal
7.2.2022 12:15am - Glad the conversation at CUNY is converging toward the real issue: Welcoming Zionist students as a valuable minority group. Sadly, CUNY adm. has not passed the litmus test for open-mindedness: Spelling the word Zionism.

7.1.2022 1:18pm - (Replying to @stephensenn) By "asymptotic" I mean "as the number of samples in each center increases indefinitely". Isn't ACE an asymptotic concept? (assuming the 'A' in ACE stands for 'average', not sample 'mean.'

7.1.2022 1:09pm - I welcome the emergence of the new Jewish Studies Zionist Networks. From now on, anti-Zionist professors will not be able to say "We represent THE Association for Jewish Studies" without specifying WHICH ONE; the old and irrelevant, or the new, vibrant and organic? I am joining!

7.1.2022 12:31pm - (Replying to @kareem_carr) Great! But didn't they rule last year that machine learning is statistics on steroids?

7.1.2022 2:16am - (Replying to @elderofziyon) We call it "The Tectonic shift of John Kerry."

7.1.2022 1:06am - Replying to @elderofziyon We call it "The Tectonic shift of John Kerry."

6.30.2022 4:51pm - As a holder of a dual citizenship, Israeli+American, I am proud to introduce readers to the new Prime Minister of Israel, Yair Lapid, a man of vision and integrity, who just took office today, and is being congratulated by my President Joe Biden. To their partnership!- LeChayim!

6.30.2022 4:22pm - (Replying to @AvrahamAdler @mishtal and @benandjerry) The malice comes from Anuradha Mittal. But @benandjerry remind me of so many Jewish American uncles, who just want to please everyone (including Mittal) as long as you dont ask them to study more carefully what's going on in ME, or what the Palestinians say about Israel's future.

6.30.2022 1:44p - (Replying to @mishtal) You are too harsh on innocent boycotters like @benandjerrys who simply wish to signal virtue, peace, love and ice-cream and, not knowing any better, picked up the most populist slogan on the internet, "occupation of Palestinian land", and made it their banner of thoughtlessness.

6.30.2022 5:40am - (Replying to @LekhtNaya) And I want to be an "oppressor" - teach me!

6.30.2022 2:16am - (Replying to @PMinervini and @kerstingAIML) I would ask it more provocatively: I have a large model that someone found on the pavement outside. We know nothing about it except that it is "large", in fact huge, humongous. Is it a "Foundation Model" in Stanford sense"? In anyone's sense?

6.30.2022 1:26am - (Replying to @kerstingAIML) A calculus of intuitive causation would be great to have. Any aspirations as to its eventual structure? Here is an intuitive causal inference: If a drug helps men and helps women it must help people. Can "Foundation Models" reach this conclusion?

6.30.2022 1:14am - (Replying to @kerstingAIML) What is LLM? Couldn't find it in your paper, nor in Wikipedia.

6.30.2022 12:58am - Orwellian. But I don't believe President Pollack, a no-nonsense lady, would have approved of such stupidity. It tells more of how Wokism works: Some looney complains, a clerk gets scared of being accused of insensitivity, another hurries to his protection -- soon Lincoln is gone.

6.29.2022 11:39pm - My list of recent articles on causal inference: One paper catching my attention is "Can Foundation Models Talk Causality" a topic discussed on our platform which made me wonder: How can one ask "Can X do Y?" when X is undefined?

6.29.2022 10:56pm - @MESA_1966 decision to boycott #Israel academic institutions amounts to stripping the Middle East from the one country that has ushered the region to modernity-- in art, science, business and democracy -- thus committing the region to another century of hopelessness.

6.29.2022 4:10pm - This betrayal thrives on disinformation: Donors to "Israel Studies Centers" are kept ignorant of what leaders of those Centers teach or advocate. I therefore link here to the original statement, to help current & potential donors identify hijacked Centers.

6.29.2022 1:00pm - An amazing building! They mushroom on so many campuses. Naughty me, I once called them "Gothic Cathedrals for a handcuffed God." I still do, while hoping that the ratio of "science" to "data" would justify the hype.

6.29.2022 8:14am - (1/2) I have signed this petition, and I urge other readers to do so. However, adopting the IHRA definition of antisemitism is no replacement for pro-actively nourishing a Zionist-friendly culture on CUNY's campus. First, by explicitly welcoming Zionist students and faculty for
6.29.2022 8:14am - their unique contributions to campus life and, second, by declaring anti-Zionism incompatible with the educational goals, the principles of diversity and the norms of discourse the administration is striving to establish.

6.29.2022 4:27am - (Replying to @stephensenn and @snoble) Agree, and promise to reciprocate. Moreover, no need to compress "everything", just TWO different ways of doing RCT would do, and if it takes 3 x 280 character, so be it.

6.29.2022 4:09am - Congratulations to our colleague @Adi_Schwartz , co-author of "The War of Return" (with @EinatWilf), on receiving a doctorate degree from Bar Illan University.

6.29.2022 4:03am - (Replying to @Ostrov_A) What is there for a journalist to cover in Israel when you know in advance what you are about to see?

6.29.2022 3:34am - (Replying to @artistexyz and @AleksanderMolak) I see a "formulae" there, but unless I see the assumptions that lead to that formulae I cannot call it a "definition". I understand the assumptions that go into the sequential backdoor (as in Pearl and Robins 1995), but not the ones in WhatIf page 261, my weakness.

6.29.2022 3:27am - (Replying to @DrBobGoldberg) When and how did she make it clear? Has anyone listened to her? Is she on Twitter?

6.29.2022 3:08am - A clear video description of the Zionophobic culture permitted to incubate @CUNY . Hoping the City Hall protest tomorrow ends with actionable items, not merely protest. First item: apology for inviting a speaker who criminalized the identity of thousands of students and faculty.

6.29.2022 1:44am - More on dynamic systems. The best I've seen so far is: Bongers, Stephan, Tineke Blom, and Joris M. Mooij (2022). “Causal Modeling of Dynamical Systems”. Preprint. Available at arXiv:1803.08784v3 [http://cs.AI]. More in Bongers PhD dissertation, forthcoming.

6.29.2022 1:06am - (Replying to @artistexyz and @AleksanderMolak) I confess that I've never understood what g-formula is. Is it any truncated product of conditional probabilities? Or one that emanates from a specific graph (as in the do-operator). If the latter, how can you define it w/o a graph or an SCM? This I havn't been able to figure out.

6.28.2022 11:32pm - Losing your mind? Why? Fatima Salah and her linguistic fertility is just another example of Western scholars refusing to treat Palestinian mentality as an epiphenomenon deserving its own logic and psychology. An elephant in the room of rejectionism:

6.28.2022 10:22pm - (1/3) This is a gross misunderstanding of Palestinian rejectionism. Not a single refugee existed in 1936 (when I was born), no "occupation" nor "ethnic cleansing", yet rejectionism was at its highest. "Not even the size of a postage stamp" screamed Mufti Amin al-Husseini, when
6.28.2022 10:22pm - (2/3) 9 million Jews (including my grandparents) were trapped in Europe, seeking refuge. Not a single Arab refugee existed in 1945-7, when Holocaust survivors lingered in DP camp, seeking a home, and Arab rejectionism said NO to partition. Arab rejectionism is DNA deep, as
6.28.2022 10:22pm - (3/3) articulated clearly by every Arab historian (eg Rashid Khalidi) but ignored by Western analysts, who refuse to accept the absence of a quick "fix" they can engineer after taking a class in "conflict resolution". It remains the longest ignored elephant in the room of history.

6.28.2022 5:36pm - (Replying to @razingarizona @Vandalay_Inc and 3 others) I see nothing wrong with a 'Two-State Solution Act" if only it correctly identifies the main obstacle to such solution, and conditions every policy item on removing that obstacle - Palestinian rejectionism.

6.28.2022 5:23pm - (Replying to @blakeflayton) They blocked you because you represent the thing they fear most: A secured "Pro-Israel Pro-Peace" ideology rooted in the understanding that the latter follows organically from the former, and that it is wrong therefore to trumpet the latter as if it doesn't.

6.28.2022 2:56pm - (Replying to @BrettWilkinsSF @impsardo and 2 others) An interesting new logic. Israel is guiltier because she is hated more viciously, by more creative gang of predators, able to stockpile a richer arsenal of populist slogans, fabricated more craftily for Western minds to fall for. The more battered, the guiltier.

6.28.2022 9:00am - (Replying to @StaceyEBurke @Vandalay_Inc and 2 others) This may be the reason why JStreet, as well as @NewIsraelFund , spend their time attacking other Jewish organizations - impressing their donors with some action and uniqueness. I once wrote an oped about the Jews of Discomfort: Still relevant today.

6.28.2022 7:37am - Samantha, those who harassed you expect you to beg for protection. Try to address them as "Zionophobic racists" and see how they lose balance. Example: Zionophobes are shocked when their ideology is examined as a moral deformity, by facts & principles.

6.28.2022 6:57am - (Replying to @AleksanderMolak) The closest I got to dynamic systems was in a paper with James Robins ". "Probabilistic Evaluation of Sequential Plans" which computes the effect of time varying actions, in discrete times, each depending on the outcomes of its predecessors.

6.28.2022 4:33am - Thoughtful analysis of the Boston Mapping Project, especially on why Jewish groups have a hard time naming the ideological roots of progressive antisemitism.

6.28.2022 1:30am - (Replying to @GadSaad and @KamalaHarris) Finally, someone surpassing @AOC understanding of the Middle East.

6.28.2022 1:05am - (Replying to @stephensenn and @snoble) It's not a general rule. It's specific to this particular problem which we have cycled dozens of times, with zero progress, despite my honest attempts to penetrate the verbal curtains of RCT's discourse. It should also be helpful for trialists to see their words cast in 2 Eqs.

6.28.2022 12:53am - (Replying to @Martin_Kramer @MESA_1966 and @DovWaxman) Hard to believe my letter pushed UCLA to rescind its membership. I assume there were stronger oppressive and colonial forces behind the process, and no trumpets, to avoid backlash. That should only encourage Israel Studies Centers at other universities to try.

6.28.2022 12:14am - Illuminating article and fascinating demonstration

6.27.2022 11:50am - I became friendly with the Presbyterian Church when they asked me to go to Alabama to debate BDS cronies. Today they sent me an interesting video which I would like to share here. It's an interview with Palestinian HR activist Bassem Eid: Worthy watching.

6.27.2022 10:18am - The next step is direct flight, Mississippi - TelAviv, paid for by confiscated Iranian bank accounts.

6.27.2022 10:04am - And after the awards comes the yield: List of last week's articles on @causalinference : The one paper that caught my eyes was the thesis by Nick Pawlowski on Probabilistic and Causal Inference in Imaging

6.27.2022 9:17am - It has been good year for @causalinference . Congratulations to Laureates James Robins, @_MiguelHernan , Thomas Richardson, @RotnitzkyAndrea and Eric Tchetgen Tchetgen for the King Baudouin Award. All that remains to do now is to agree on what @causalinference is about!!!

6.27.2022 8:33am - Sorry to see @sharon_nazarian leaving @ADL at this crucial time, and wishing her good luck and much success and satisfaction in her next endeavor.

6.27.2022 8:07am - The FDA was formed in 1906. I wonder when was it that they allow counterfactual vocabulary to enter, perhaps even dominate their official documents?

6.27.2022 3:20am - UN: Look at what you have done with a truly noble idea! How could you let it slide to such lows?

6.27.2022 1:45am - Speaking of Human-Centered AI, I can't think of a more human-centered project than this.

6.27.2022 12:12am - (Replying to @rory_watts_ and @cauchyfriend) I would paraphrase your observation a bit more crisply: Potential outcomes without the First Law are 2 decades behind the times, and can't begin to do the pseudo-experiments @cauchyfriend is advocating without a causal model:

6.26.2022 11:44am - (Replying to @snoble) This paper might help in further shaking the supremacy of RCT

6.26.2022 8:35am - (Replying to @sbc111 and @aleixrvr) Sorry, I can't grasp the message of this paper. Safety and effectiveness are causal notions, and require causal analysis. Why burden ML systems with tasks they cannot handle?

6.26.2022 8:22am - Western minds cannot grasp one hard fact which holds the key to peace: Palestinians, to the last man, are light years away from accepting Israel as permanent, in any border. Wishing away this hard fact is so convenient, but also so delusionary and costly.

6.26.2022 7:53am - Reminds me: @benandjerrys have somehow been excluded from the "mapping project" hit list -- this renders them "redeemable-Jews". I'm wondering though what got @jstreetdotorg and @NewIsraelFund on that honorable list, must be sloppy scrutiny.

6.26.2022 7:15am - They fought and died so that we can say:" They were there" ("Anachnu Po").

6.26.2022 7:07am - (Replying to @chroma_infinity) Why? I am seeing several "straight lines" in that building meeting at more than one "point", so I jumped to conclusion. And thanks for commenting favorably on my Israel-inspired perspective. It pains me to see that commonsense requires "activism"

6.26.2022 5:13am - (Replying to @michaeldickson) It does not occur to @falastiniyat that the intersection of "reproduction justice" and "anti-Zionism" is the empty set, also called "logical contradiction".

6.26.2022 4:51am - Amazing! I bet the people working in these twisted offices believe, like most of us, in non-Euclidean geometry.

6.26.2022 4:35am - (Replying to @KordingLab @lessc0de and 4 others) As an amateur historian, I am grateful for the pointer to Ibn Sina.

6.26.2022 4:08am - Last I checked, @JvpAction and @codepink are not on the "mapping project" hit list; they got a "Hechsher" for doing more damage to their people if left alone (for the time being). You made it, @ArielElyseGold

6.26.2022 3:37am - (Replying to @amadea_de and @ArielElyseGold) No, the "mapping project" does an excellent job of exposing the mentality of those leading that struggle.

6.26.2022 3:22am - (Replying to @richard_landes) It's even funnier to see both talking about "facts on the ground" without asking Israelis (presumably still there, on the ground) whether they are willing to submit to such experiment.

6.26.2022 3:11am - (Replying to @mishtal and @WAFANewsEnglish) There is great value in letting Leila Haled preach about "freedom fighters" "occupation" and "apartheid" so that these words will obtain their intended interpretations when spoken by BDS cronies. Spread widely!

6.26.2022 2:52am - (Replying to @aobeid_1) The Amalekites disappeared not because King Saul killed them, but because they contributed nothing of value to human civilization, not even one legend their descendants found worthy of preserving from generation to generation.

6.26.2022 2:28am - (Replying to @DrCharlesJacobs @MarjorieKaufman and 19 others) Excellent analysis and blasting indictment of higher education and the enterprise of "human intellect". If thinking organisms like Crimson's Editors fall easy prey to populist slogans and tictok conspiratorial theories, who is protecting our species?

6.26.2022 2:08am - Retweeted:
StandWithUs @StandWithUs Today is #Holocaust Survivor Day. Perhaps you know a Shoah survivor or have heard one's story. We invite you to keep this person in your heart today, to put a face & name to the horrors of the Holocaust & to ponder the resiliency of the human spirit, even in the darkest of times.

6.26.2022 12:34am - In particular, Rev Dr. Pillay should not parrot vacuous calls "for an end to the occupation" without addressing the root causes that make the occupation an existential necessity. The world expect a more enlightened perspective from the new head of WCC.

6.25.2022 11:24pm - Are the intellectual assets of this @tictok nymph truly different from those of Harvard students who bark: "Racism, Settler Colonialism, White Supremacy, Apartheid?" then graduate Law Schools to become @hrw officers all over a planet called "Earth"? See

6.25.2022 1:40pm - Israel & Abortion
6.25.2022 1:40pm - As we are trying to recuperate from the Roe-Wade depression, let's lift our eyes to a brighter spot on the planet, a spot which both anti-choice and anti-Zionist minds prefer to ignore: Israel and Abortion.

6.25.2022 1:22pm - We are informed of the resurrection of the UAI probabilistic inference competition, to which a new machine learning component is now introduced. For details, see the competition website:

6.25.2022 7:13am - I assume this is what they mean by "apartheid", separating tolerance from intolerance.

6.25.2022 6:41am - This quote is eternal, but became extra-relevant in the past two years, when the last line "(hate) just their nation state" became openly normative, especially among academics, even among Jewish colleagues. Always making me wonder: Do they truly grasp how history will judge them?

6.25.2022 4:24am - (Replying to @JordiTorresAI @FundacionBBVA and 2 others) Thanks. Google is amazing. And Barcelona is inspiring. (My grandson lives there).

6.25.2022 4:10am - To our lioness sisters at @ZionionessMovement. This is the time for all men of conscience to join your march, for you represent the two most noble revolutions of the 20th century, Zionism and women rights; we can't let the 21st century betray your two achievements.

6.25.2022 3:32am - In a sane world, the prolongers of a conflict would be embarrassed to admit how long they have been prolonging.

6.25.2022 3:00am - (1/2) A new organization in town: @IfNotTomorrow Its slogan: Jews Against Drinking Christian Blood. Weird? Not really! Observe the slogan of @IfNotNowOrg below: Jews Against Apartheid. How about: @IfNotYesterday. Slogan: "Jews against all horrible things Jews are blamed for"
6.25.2022 3:00am - (2/ ) If you're a decent American, innocently presuming that Jews are just ordinary folks, @IfNotNow recruiters now tell you: "Wrong! A typical Jew, unless he/she joins our select camp, is by default guilty of those horrible things." To join them is to indict your parents. I won't.

6.25.2022 1:55am - At the request of curious readers, I am re-posting a video of my acceptance speech in Bilbao: Upon re-listening, I realize that I agree with every sentence! Strange for poems that compress four decades of research into 900 words.

6.24.2022 4:26pm - (Replying to @GnomeSortingHat) I would march it to its obvious conclusion: "Breathing has no effect on death" And, to sound statistically savvy, I would add: "On the average".

6.24.2022 7:44am - A significant by rarely mentioned fact: The first book to be printed in Western Asia was printed in the city of Tsfat (upper Galilee)

6.24.2022 7:16am - (1/3) This is a major legal victory against BDS, agree, but now we need to move to the moral dimension and explain to people concerned with First Amendment why boycotting Israel differs from maligning Israel, or even lying about Israel, both are protected speech. I believe the
6.24.2022 7:16am - (2/3) the rationale for opposing boycott (putting aside the whimsical and baseless nature of the allegations raised)
6.24.2022 7:16am - (3/3) with the functional nature of those services, undistorted by political or ideological considerations. I would be curious to hear from readers if other considerations exist; why running a business is different from shouting in the park.

6.24.2022 5:54am - This reminds me of the possibility of demonstrating "Reverse Regression" and the argument whether we should compare salaries of equally qualified men and women, or, rather, compare qualifications of equally paid men and women. See

6.24.2022 5:16am - (Replying to @le_lummy) Sure, but I couldn't cite all Biblical/Mishnaic cities Tzfat deserves a special mention, I agree. Why? Because the first book printed in Western Asia was in Tzfat, in Hebrew, in 1577. Thus, it was the Hebrew language that ushered the ME to modernity, not Greek, Latin or Arabic.

6.24.2022 2:59am - Investing in UNRWA means investing in self-sustaining bureaucrats whose jobs, relevance and mission is to perpetuate the conflict, and providing them with the means to do it: Education to Passivity and Hate.

6.24.2022 2:40am - Good morning, Tel Aviv! The first Hebrew speaking town in the entire universe (barring ancient Jerusalem, Beer Sheva and, of course, my own home town of Bnei-Brak)

6.24.2022 1:52am - (Replying to @grbradsk and @IfNotNowOrg) I think it is more of a "battered wife" syndrome in the case of @IfNotNowOrg and @NewIsraelFund ; "Something must be wrong with us if the bullies make us feel so uncomfortable. We can't control the abusers, but we can work on ourselves to feel like we are in control."

6.24.2022 1:30am - (Replying to @soboleffspaces and @MatthewFe94) Good paper, agree. And just in case there are changes, here is the revised, published version of this paper

6.24.2022 12:59am - (1/3) I could accept that kind of "controlling" behavior from reviewers if they themselves felt more comfortable with Imbens-Rubin notation. But they don't, b/c no mortal does. So what could be their motivation, aside from the cultish preservation of the Imbens-Rubin torture?
6.24.2022 12:59am - (2/3) I like it: "The preservation of torture", a new force in the social dynamics of science. I don't believe Thomas Kuhn considered this inertial force in his "The Structure of Scientific Revolution", but it's very real: "I suffered in my student days, you better suffer too".
6.24.2022 12:59am - (3/3) "There must be some wisdom in suffering, Otherwise I've suffered in vain, which is hard to swallow. So, suffer with me, make your readers suffer too and, in this way, we will all touch some of the wisdom that continues to inspire author after author in this journal" Love it.

6.23.2022 11:55pm - (Replying to @MairavZ) The days of such accusations are over. Today we label "Zionophobes" those who rush to blame Israel with hearsay evidence, and who refuse to grasp that Israel has everything to lose and nothing to gain from targeting journalists. @pressfreedom

6.23.2022 11:30pm - What an honest confession of the confused, misinformed, rootless and spineless culture that spawns converts to @IfNotNowOrg . What an indictment of Jewish education in America.

6.23.2022 10:39pm - (Replying to @soboleffspaces) Where do you find statisticians "controlling"? Those I know are exhausted trying to explain to their students what "unconfoundedness" is - an insurmountable, super-human task according to Imbens and Rubin (Cambridge, 2015)

6.23.2022 5:14pm - Those of us who adopted the First Law of Causal Inference as a starting point for formulating theories, models and empirical methods will see no problem satisfying @ethanbdm quest for bridging the gaps noted, especially the testability gap.

6.23.2022 4:20pm - A reader complained yesterday that "55 years of occupation is unjustified". I complained too, but, in contrast, I show my awareness of what the main force is that prolongs the conflict. Here it is: Hard to fight, hence easy for @NewIsraelFund to ignore.

6.23.2022 8:01am - (Replying to @EinatWilf) Trial balloon, yes. But, realistically, I can't see Israelis going through the civil strife of uprooting X number of settlements without a more meaningful ideological commitment to "end of claims" on the part of the Palestinians.

6.23.2022 7:26am - (Replying to @TheAcsMan @BDSBoston and @BDSmovement) The same can be said about some Jewish organizations who discovered BDS only when their names appeared on the "Boston Mapping Project." Two decades of BDS delegitimization of Israel only made them yawn; their name on the "Map" and they scream: Gewalt!

6.23.2022 7:16am - (Replying to @DKedmey) Both Wingate and Patterson are inspirational! Perhaps my poem will inspire someone in the 22nd Century to salvage the remanences of our great civilization and rebuild the walls of Rome.

6.23.2022 7:05am - (Replying to @neuroviana and @FundacionBBVA) I've nothing but praises for this book - enjoy!

6.23.2022 1:01am - (1/2) Poor @BDSBoston , they simply tried to explicate what mother @BDSmovement has been telling them for years; who the real enemies of justice are; not the good Jews of JVP, not the anti-Zionist reform synagogues, just the evil Jewish organizations that push for co-existence.
6.23.2022 1:01am - (2/2) Now mother @BDSmovement is telling them: Drop it. You know that we count on the world not to take seriously what we say. You spoiled it. Americans are too sensitive to "mappings", so pretend the map is only for ordering Kosher food, and go back to advocate

6.22.2022 11:38pm - (Replying to @patrickmesana @ylecun and @GaryMarcus) I would like to believe that that mysterious life force is called "the scientific method".

6.22.2022 11:24pm Begging to differ. If the only reason Israel should take existential risks is the moral burden of "55 years of occupation is unjustifiable," then I would venture to suggest that 74 years of denying a neighbor's right to exist is many many times more unjustifiable.

6.22.2022 10:50pm - (1/2) Derek Penslar, unsurprisingly, gets it totally wrong, as would any author of the Van Leer "definition" of antisemitism. What's the point of wasting energy on scholastic distinctions between "anti-Zionism" and "antisemitism" when the real issue (which the Van Leer "scholars"
6.22.2022 10:50pm - and @NewIsraelFund ignore) lies in exposing the racist character and eliminationist aims of organizations such as BDS and JVP who pose as "Palestinian rights advocates". Hail to the Boston Map for forcing the Van Leer "scholars" to halt their knit picking and take responsibility.

6.22.2022 8:44pm - Thanks for the reminder. I've lost my poetic muse lately; this piece was written when I still believed that Western culture has the wisdom to see its virtues and the will to protect itself.

6.22.2022 8:22pm - (1/2) (Replying to @slowhoneybee and @blakeflayton) You are analyzing the Forward article as if it was a piece of news when, in effect, it is written to drum up sympathy for the poor semi-Jews of Boston who are being forced to examine their identity, something they haven't done since they coined the term "pro-Palestinian"
6.22.2022 8:22pm - (2/2) (Replying to @yudapearl @slowhoneybee and @blakeflayton) to replace the term "anti-Israel", which allowed them to ignore the anti-coexistence aims of the so called "pro-Palestinian activists", eg, the nice guys from @JVPBoston and @IfNotNowOrg , who are not antisemitic, heaven forbids, just Kosher eliminationists.

6.22.2022 9:10am - (Replying to @blakeflayton) The laws of arithmetic are a Zionist invention, created to justify Jewish aggression. Palestinian science has different rules. It predicts about 100 million refugees in 2030, and says that the longer one refuses to recognize its neighbors the shorter the conflict would last.

6.22.2022 8:39am - (Replying to @kareem_carr) Apropos, Why I am only a Half-Bayesian

6.22.2022 8:10am - (Replying to @AndrewPGrieve) Congratulations! For my curiosity, can you share a simple problem the Hybrid method helps solve, which a half-Bayesian like me would find difficult to solve?

6.22.2022 8:04am - (Replying to @SAFECUNY and @ChancellorCUNY) I don't get it. Your Chancellor declares anti-Zionism a legitimate form of campus hate and you applaud him?

6.22.2022 7:54am - The Boston Map is just another "pound of flesh" anti-Israel "progressives" are requested to give for social acceptance, awakening some of them to notice that there is no more flesh to give -- dry bones is all that's left. Maps have virtues after all - happy awakening.

6.22.2022 6:10am - (1/2) One last remark on my experience in Spain: If you have the option, try to have your next Award given in Bilbao. In addition to the Basque dances and Presidential reception that you see below, this town transformed itself into an "Ode To Science" for three whole days, with
6.22.2022 6:10am - (2/2) photos of all 14 laureates hanging from balconies and tree tops, children running after you for an autograph and high school students asking what books they should read. Man, even my grandchildren began to respect my profession by the 3rd day. Don't hesitate, choose Bilboa.

6.22.2022 2:25am - (1/2) A.B. Yehoshua, one of Israel?\200\231s most revered authors, died on Tuesday last week at the age of 85. In lieu of an obituary, I've reprinted his essay "The Definition of the Jew" -of-freedom/ I wish some of my esteemed colleagues
6.22.2022 2:25am - would ask themselves "what is a Jew?" before jumping to define "antisemitism." Doing so, would have spared their people sloppy proclamations and immensely harmful documents, such as the Van Leer Institute and the @USCAnnenberg Nexus "definitions" of antisemitism.

6.22.2022 2:19am - Lot's of new papers on Causal Inference this week. Here are some that crossed my screen: I wish I can read them all, but I have only two eyes to try and sort the lions from the rest.

6.21.2022 10:07pm - As promised, here are a few links received from Bilbao, Spain: 1. Text of my acceptance speech: nce_speech_Frontiers_Award_14th_edition.pdf 2. Video of acceptance speech: 3. Dancing at the Award Ceremony nowledge-awards/ 4. Profile page

6.21.2022 5:09am - I've just returned from Bilbao, Spain, where they staged quite a show for the Laurates of the BBVA Frontiers of Knowledge Award. See, and I will post a transcript of my acceptance speech later.

6.21.2022 4:48am - Please note, all Jew-haters deny that they are "Jew-haters" (even Linda Sarsour), while no Zionopnobe has ever denied having this uglier variant of the virus, thus making it easier to detect and irradicate.

6.21.2022 3:03AM - Rapists always argue that a woman's short skirt is responsible for her rape. Our panelists at Van Leer Institute are now claiming that its not her skirt, but her very body. They're right: If Jews were not what they are at heart- lovers of Israel - there would be no antisemitism.

6.21.2022 2:35am - This smallpox example was my first exposure to the "dark world of causation", as related here:

6.21.2022 12:30am - (Replying to @HenMazzig) If over 90% of Jews view Israel as the homeland of the Jewish people, what's the point in whining "antisemitism" instead of protesting an uglier form of racism: Zionophobia.

6.21.2022 12:12am - @Unesco own site on Masada is also worth visiting: It was inscribed as a World Heritage site in 2001, and has not been taken down yet, after Unesco decided (2016) to erase Jewish connections to the land. Hurry!

6.19.2022 8:29am - Fighting antisemitism: 'Pound of flesh' in 2022 is never enough - opinion
6.19.2022 8:29am - I would not confer academic credibility to people who signed the Van Leer definition of antisemitism. Let history judge their thoughtlessness.

6.19.2022 7:08am - (Replying to @blakeflayton) A culture that has produced no knowledge should not be mocked for trying to produce something.

6.19.2022 8:29am -

6.15.2022 11:36am - Retweeted Concierto extraordinario en homenaje a los galardonados en los XIV #PremiosFronteras del Conocimiento.

6.12.2022 2:37am - May his prediction match his knowledge of history. "Make believe" about the future is cute kids game, "make believe" about the past is hardly cute.

6.11.2022 1:13pm - (Replying to @charleswangb) Every universal machine, be it cellular or no cellular, network or no-network, can simulate the do-operator if given the information about the underlying DAG. The issue is getting the DAG structure and knowing what to do with it when asked a question about actions and policies.

6.11.2022 12:57pm - (Replying to @BifarinTheFifth) I love the beads on your hand, each singing a hymn to good taste!

6.11.2022 6:00am - (Replying to @terrible_archer @arya_amsha and @ruchirsharma_1) I wasn't aware of such anti-Hindu sentiments in America. But I've only been here 62 years.

6.10.2022 10:38pm - (Replying to @artistexyz @rspadim and 3 others) Do you think I would be able to understand what "transformer" is, given my limited vocabulary? You know what my limitations are, I start with a research question, I must know what information is provided in addition to the data, & finally, what performance guarantees r needed.

6.10.2022 2:39pm - @AlJazeera : This is another Zionist pink-washing trick, to divert attention from Israel's blood drinking ceremony a few miles away.

6.10.2022 2:08pm - A rare visit by Pakistanis to Israel has caused furor in Islamabad
6.10.2022 2:08pm - An unbelievable visit, of historical significance. I wish I could say I had a part in conspiring to arrange it. Israel is the modern catalyst of co-existence.

6.10.2022 1:47pm - Agree! Congressmen need love too. Especially for the courage it takes to remember Jerusalem in a Squad infected Congress.

6.10.2022 12:31pm - (Replying to @GaryMarcus @slatestarcodex and 8 others) Can we have a link to the main arena, where the punches are thrown among ideas?

6.10.2022 5:52 - (Replying to @mariotelfig @philipcball and @newscientist) It seems that this is the example I was looking for, many thanks . Now I need to swallow and digest.

6.10.2022 5:38am - (Replying to @_fernando_rosas) I'm purposely trying to avoid man-made interventions which will open the door to all kind of "free-will" speculations.

6.10.2022 3:26am - (Replying to @IntuitMachine) Yes, this is the kind of example I am looking for, starting with ideal gas and proving that the atmospheric pressure (a macro variable) "causes" the barometer reading (also macro variable) and not the other way around. Here "causes" is defined by counting configurations.

6.10.2022 3:15am - (Replying to @philipcball) The reservations cited in your article merely express my inability to follow the arguments when all measures of causal strength are expressed as probabilities. My comments now beg for an example that goes from symmetrical '=' to asymmetrical '--->', say earthquake to tsunami.

6.10.2022 1:32am - (1/3) Few comments on @philipcball article @newscientist . (1) That causation is an emergent property is indisputable. Both classical and quantum mechanics are governed by the symmetric equality sign '=' at the microscopic level, hence could not account for causal asymmetries
6.10.2022 1:32am - (2/3) we observe at the macroscopic level of discourse. (2) To demonstrate emergence, we need to show how a system of particles governed by '=' leads to an asymmetric relation ---> among macro variables defined in terms of micro states. (3) The demonstrations I've found in this
6.10.2022 1:32am - (3/3) and related articles @ericphoel are too complex for me to follow, and are articulated in terms of causal vocabulary which should be avoided until the asymmetry emerges. (4) Would appreciate the simplest demonstration available, hopefully using classical deterministic systems.

6.9.2022 12:05pm - University of Toronto to Withhold Thousands in Graduate Student Union Fees Over Israel Boycott
6.9.2022 12:05pm - Any U. of Toronto affiliates among our readers? As a fellow alumni (honorary Doctor, 2007) please convey to your Provost my appreciation for protecting the reputation of your/our university.

6.9.2022 11:52am - (Replying to @HumzahAlkindi) The journalists Hippocratic oath is not "being found credible by both sides" but "showing the human face of both sides". Since the latter is a prerequisite to "normalization", it is resisted by all anti-normalization forces, e.g. @AlJazeera , @en_emamkhomeini , @RashidaTlaib

6.9.2022 4:08am - This linguistic transition has not been linear but circular, returning to the vocabulary of the Mufti in the 1920's who stated openly "Not even the size of a postage stamp". It was socially unacceptable for a while (after the Holocaust) but regained respectability lately.

6.9.2022 3:09am - A serious question indeed. We can joke about the mentality of @khamenei_ir and his ilk but, considering that he has 900K followers, each a potential killer, Twitter should seriously consider suspension. No one would be allowed to make such remarks on say Mexican nationals.

6.9.2022 2:28am - A second list of newly published paper touching on causal inference is compiled here: I hope to continue on a weekly basis.

6.9.2022 12:52am - (Replying to @rjherber and @memarkelliottme) I am not a Rabbi, true, but I preach that Judaism is "peoplehood first and religion second". As Ruth said: Amech Ami V'Elohayich Elohay (Your people is my people and you God is my God). Namely, once I become part of your people, accepting your G-d will be easy.

6.9.2022 12:46am - (Replying to @memarkelliottme and @razingarizona) Real estate is the least disputed commodity in the ME conflict. The peculiar issue has always been to reconcile a conflict in which one side says "Let's share" and the other: "Not even the size of a postage stamp".

6.8.2022 9:20pm - (Replying to @rjherber and @memarkelliottme) This is because most Jews, even most Jewish organizations, do not insist on correcting the mistake. Some enjoy the ambivalence: pleading "tolerate us like any other religion" with their leftist friends, and demanding "two-states for two peoples" in international arena.

6.8.2022 9:10pm - (1/2) I will be taking a few days off from Tweeting - need to go to Spain, soon, to pickup the BBVA award I told you about: If you happen to be at Bilbao, Spain, next Thursday, please drop by the presentation ceremony, 17:30pm at the Euscalduna Conf Center.
6.8.2022 9:10pm - (2/2) If it's out of your way, don't worry, my acceptance speech will be recorded (in English, Spanish and Hebrew) and made available on this channel. It will focus on the Ladder of Causation and why it helps us understand who we are.

6.8.2022 7:20pm - (1/2) Glad you brought up this question, because I know that many people think Jews are a religion, as opposed to a "people", bonded by common history, attachment to one land, and common destiny. With this in mind, I hope you join the Jewish aspiration: Two state for two people,
6.8.2022 7:20pm - Some people end here. I don't. I continue: "Two states for two peoples, Equally legitimate and equally indigenous" Why the second line? Because it is easy to cheat on the 1st, impossible to cheat on the 2nd.

6.8.2022 6:54pm - 100 Years Ago This Month: When Congress Embraced Zionism—Unanimously via @jewishjournal
6.8.2022 6:54pm - A chapter of history I did not know, though I should have, given the striking similarities to debates that still rock the Halls of US Congress.

6.8.2022 5:06pm - (Replying to @AnalytiCringe) Your life will change, I'm sure. Mine was changed twice. First when I discovered I could do things I've always wanted, but didn't know how. Then when I discovered that so many folks (eg economists) still don't know how, but are afraid to ask.

6.8.2022 12:50pm - (Replying to @aleixrvr and @ccaballeroh10) In Wikipedia I introduced p(y|do(x)) conceptually, not as an operation on graph, but as a distribution created once we make sure X=x is enforced. I believe even high-school kids would understand the difference btwn "the price listed was X=100" and "I raised the price to X=100."

6.8.2022 8:06am - I felt like this post was coming from a UCLA colleague and is talking directly to me, about me, and about the new plague I have been scheming against the forces of light and goodness.

6.8.2022 7:44am - The sea that saw the birth of "thinking" never ceases to impress us with its depth.

6.8.2022 7:33am - Speaking about de-confounding confounding, I recently cleaned up the entry for "confounding" on Wikipedia Sections: Definition, Control and History are fixed, while the rest were left in their original state of confusion. Hard to fight Wiki editors.

6.8.2022 3:07am - (Replying to @aleixrvr) From your experience as a "data scientist" (who isn't?) what did you find to be the most important Causal Inference tool your peers can exploit to benefit?

6.8.2022 1:11am - Delighted to welcome a new book on Causal Inference for Data Science, which is based on DAGs, doesn't shy from basic issues, and promises to usher data science towards modernity. Looking forward to see an electronic copy.

6.8.2022 12:34am - (Replying to @stefanharmeling and @shalom123211) You are right. This is a simple situation to which Rule 3 applies (to ignore an action) but not the succession of Rule 2 (exchanging action by observation) followed by Rule 1 (ignoring an observation).

6.7.2022 11:47am - What a song! One would bet it was written yesterday, or at least two weeks ago, when @CNN came out with their Kangaroo verdict as to WHO killed Abu Akleh: "His enemies they all chant he's killing their journalists, He has nothing better to do, but to kill and to kill."

6.7.2022 4:08am - As a carrier of a prize given in his honor, I am posting my Alan Turing lecture of 2012: The works of other Turing laurates is described here:,for%20inference%20in%20these%20models. (Click on the photo)

6.6.2022 9:52pm - I remember D-Day. No, I wasn't on a boat, I was only 8 yr old. But I remember my father rushing home with a newspaper screaming: HaPlisha Hechela (The Invasion Started), and everyone gathered around the newspaper which had a huge headline: Haplisha Hechela. All I knew: it's BIG.

6.6.2022 8:12pm - (Replying to @ishapiro and @GeorgetownLaw) Hard to read the small print. Do you have a summary of the issue you had at @GeorgetownLaw ?

6.6.2022 7:33pm - (Replying to @AvivaKlompas and @AyaIsleemEn) I don't understand why everybody is condemning anti-Israel @AyalsleemEn for pretending this olive tree is in Palestine. History is a life-sustaining commodity, like water and oxygen; those who have much of it should share with those who have none.

6.6.2022 6:25pm - (1/2) I am starting a new service on this educational channel: A weekly list (not endorsements) of newly published papers touching on causal inference. First installment: I believe it should help readers obtain a panoramic view of where the field is going.
6.6.2022 6:25pm - (2/2) It should also help our readers disseminate their papers among like-minded researchers and, in some small measure, propagate commonsense among the infidels.

6.6.2022 9:51am - (Replying to @CureT1Diabetes) Thanks, Ellen, I can't get enough of them. I'm in love.

6.6.2022 9:06am - (1/2) Those of us tuned to Palestinian social media often read serious articles on what ought to be done with the Jews once they overrun Israel. Who will live and who will die, who will be deported and who be allowed to stay, etc. Like in 1967, they are grossly underestimating
6.6.2022 9:06am - what forces can be unleashed when you have everything to lose and nowhere else to go. They just don't get it. They are brought up on the Crusaders model, and no model of an indigenous nation returning from exile -- there isn't any. And no one to warn them of the difference.

6.6.2022 8:19am - (Replying to @mel_blue and @EinatWilf) Thanks for the correction.

6.6.2022 4:31am - (Replying to @rina31420183 @NYT and @DrMichaelOren) Good to know. Let's see how they celebrate the day that changed the dynamics of the Middle East.

6.6.2022 3:19am - (1/2) This article is a must for anyone seeking a concise, 3-page history of the Israel-Palestinian conflict: Note that @DrMichaelOren set the criterion for peace as: Stating that they favor “two states for two peoples”. I think Israelis will demand a more
6.6.2022 3:19am - (2/2) explicit statement nowadays: "Two states for two people, equally legitimate and equally indigenous". As I've explained in several tweets before, a commitment to equal indigeneity status is of crucial importance, making one's commitment to peace truly hard to fake.

6.6.2022 12:44am - I almost fell from my chair in disbelief: Can @NYT print a piece celebrating 1967? I was about to send a Kudo to @DrMichaelOren , only to discover that it was a 2017 article, not 2022. The content is as relevant today as it was in 2017 but, boy, the @NYT has changed.

6.6.2022 12:19am - (Replying to @olivier_klein) Wrong. These attitudes are as "ingrained" as mother milk and morning prayers. I have lived with them since 1936 and 1947, before refugees, before "occupation" and before restrictions. Please read R. Khalidi or Said. Why ignore what they are saying? It is ingrained and universal.

6.5.2022 11:57pm - (Replying to @olivier_klein) You can get any result you want from a pole, depending how you phrase the questions. What most Palestinians mean by "two-states" is a temporary arrangement for them to prepare for the next assault. You'd get opposite results asking: Does Israel have the RIGHT to exist. Zero.

6.5.2022 11:46pm - (Replying to @WahbaRachel @Ostrov_A and @bobjcarr) There is one difference, of great significance. I don't know an antisemite who admits to being one. I don't know a Zionophobe who denies being one. They consider it a duty and an honor to deny Jews a homeland.

6.5.2022 9:45pm - (Replying to @olivier_klein) There is no danger of antagonizing those who don't match this description because, and this is the cardinal mistake of most Western observers, analysts and peace makers, the number of exceptions is infinitesimally miniscule. Standing to be corrected, all evidence admissible.

6.5.2022 8:55pm - The West Bank is still under occupation because since 1936, Palestinians view "co-existence" as a defeat.

6.5.2022 8:49pm - I would put it even stronger, in present tense: “More than the Palestinians want a state for themselves, they want to destroy and deny a state to the Jewish people.” Moreover, they would rather give up all hope for the former to retain some hope of the latter.

6.5.2022 7:21pm - What have you done to me? I am listening to Isaac Y Nora again and again I can't quit, and I can't work. I don't want to work, I want to sing with them, more and more! VEINTE AÑOS, Caballo Viejo! Que Nadie Sepa Mi Sufrir! Amazing!

6.5.2022 6:54pm - June 6th marks 74 years to the 6 day war. Some readers ask: "how did it start?" Many more ask: How come the West Bank is still under Israeli Rule? @EinatWilf has a simple answer, with which I fully agree, it's a necessary temporary control till the Arabs say: "Equally permanent"

6.5.2022 6:42pm - (Replying to @KhaledAbuToameh) May their forecast of the future match their knowledge of the past. Amen!

6.5.2022 4:33pm - (Replying to @artistexyz and @RWJE_BA) Gee, I love this song, the music, and this family. I love them so much, that I can't see how it is related to EBM, unless EBM no longer stands for Evidence Based Medicine . Is it?

6.5.2022 1:58pm - (Replying to @DKedmey @sapinker and 5 others) This "projection project" is of crucial importance in the fight against Israel delegitimization. Most BDS followers are the gullible and misinformed, who go for "'progressives' cannot all be wrong". Only a credible explanation of WHY they ARE wrong would sway back the gullible.

6.5.2022 8:41am - Not bad at all! Reminds me of Omar Khayyam: Here, with a book of verse beneath the Bough, A Jug of Wine, A Loaf of Bread—and Thou Beside me singing in the Wilderness— Oh, Wilderness is a Paradise enow! #Bookofwhy

6.5.2022 8:24am - (1/2) (Replying to @databacked101) I love this comparison, because Putin's declaration that Ukraine is not a legitimate country matches Palestinian declaration that Israel is not legitimate. But forgetting this parallel, a journalist stationed in Russia and wouldn't show faces of Russian mothers who lost
6.5.2022 8:24am - (2/2) (Replying to @yudapearl and @databacked101) their sons in the war, nor faces of orphans of fallen Russian soldiers for fear of raising sympathy for Putin is not doing his/her duty as a journalist.

6.5.2022 1:38am - (1/2) Today is the Jewish holiday of Shavuot, the day the Israelites received the Torah on Mt. Sinai. We also read the Book of Ruth for a variety of reasons: It's a story about the harvest time, and it is a story about conversion to Judaism. But for me,
6.5.2022 1:38am - (2/2) taking liberty to interpret the story of Ruth in the context of Mt. Sinai I would say: Ruth's quote: "your people is my people and your God is my God" embodies the essence of Judaism. We are people first and religion second. This is also the meaning of Jewishness in Israel.

6.4.2022 9:25pm - There is no law against mapping. The danger of mapping is operationalized by ideologists such as Rashida Tlaib and Ilhan Omar, whom BDS's lunatics view as authorities on who deserves Allah's wrath.

6.4.2022 8:33pm - (Replying to @yudapearl and @kareem_carr) No one guessed. My two words are "curve fitting". But, since this would also make some statisticians cry, which @kareem would hate to do, I would modify it to read: "Theory-free curve fitting", thus sparing statisticians' feeling.

6.4.2022 8:21pm - I can't get over @NYT platitude: “Journalists cannot do their jobs if they are targeted with impunity by any side in a conflict.” Of course! But would they accept the corollary: “Nor can Journalists do their jobs if they consistently target with impunity one side in a conflict.”

6.4.2022 7:08pm - (Replying to @kareem_carr) I can do it in two, does it count? Guess!

6.4.2022 6:54pm - (Replying to @FJnyc @nytopinion and @NYT) There is nothing wrong in being "state run media" if it remains committed to the Hippocratic oath of the profession. The trouble with Al Jazeera is that has become the Mecca of journalistic oath-breaking.

6.4.2022 6:39pm - I am proposing this slogan as a prerequisite qualification for every reporter sent to cover the Middle East.

6.4.2022 3:29pm - I have humbly asked: In her 25 yrs of covering the Middle East, has Abu Aqleh once described the human face of Israeli society? How many times has she portrayed an Israeli child or a grandmother or any human feature of 8 million people trapped in that conflict?

6.4.2022 2:41pm - I think it is only reasonable to ask my esteem colleagues at the @NYT to spell out a few litmus tests that would help naïve readers distinguish an "honest reporter" from a "dishonest reporter". I've formulated one such test, but would be curious to hear from @NYT .

6.4.2022 1:01pm - (Replying to @ben_shore) I am not familiar with the "scaling hypothesis". Curious.

6.4.2022 12:04pm - (Replying to @amiralex @HaShemsWife and @Doc_Muh) I would add that, having given this phenomenon a name: "projection" does not absolve us from continuing an in-depth, psychopathological study of pseudo-progressives' animosity towards Israel. "Projections" come in different shades and may emerge from other pathologies.

6.4.2022 11:53am - (Replying to @MerkinMuffley5 and @SpclDelivery) The reason such a term is not coined is that the world has not seen another case of a beleaguered nation enduring a 74 of war and finding it impossible to defend herself except by controlling the territory from which her enemies wish to operate, encouraged by Western observers.

6.4.2022 11:15am - I concur. This was one of my best interviews. It is so rare to find an accomplished scientist who hasn't already formed a fixated opinion about causation, willing to explore its mysteries from fresh perspectives. Highly recommended.

6.4.2022 12:27am - UCLA Students Urge UC Regents to Take a Stronger Stance Against Antisemitism via @jewishjournal
6.4.2022 12:27am - This is my university, the fate of which will be decided by the Regents; will they overrule those who should have acted and haven't? Two students reminded the Regents of their 2016 Principles Against Intolerance The students are probably not aware
6.4.2022 12:27am - of the fact that, to implement those Principle, a task force was formed by former EDI Vice Chancellor Jerry Kang, the conclusions of which are (probably) still in Kang's drawer. The lesson to other universities: A Task Force is a license for inaction, do not bless that license.

6.3.2022 11:17pm - (Replying to @MerkinMuffley5 and @SpclDelivery) And I have no idea why people use the word "occupation" to label a "temporary control". It is "temporary" until the attacker agrees to end the war of 1948, and "control" to distinguish it from "annexation" which would have made Gaza and WB "part of Israel".

6.3.2022 10:45pm - (Replying to @shalom123211 and @stefanharmeling) The question is whether there are graphs where Rule 3 applies and in which you can't remove the action in two step, Rule 2 followed by Rule 1.

6.3.2022 8:33pm - Have mercy on the poor guy. Did you give him a chance to repair the damage? Did you draft the statement you would like him to issue campus-wide. How about a list of items he can act upon next week? Most administrators are bewildered by the BDS Zoo. They never encountered fascists

6.3.2022 2:32pm - (Replying to @hyounpark) Interesting. The people who are making AI-driven business decisions, what kind of literature do they read? Perhaps we can submit a business-minded version to the top magazine they read?

6.3.2022 12:27pm - (Replying to @TobiasNavarro) Welcome to the causal revolution.

6.3.2022 10:13am - (Replying to @MerkinMuffley5) Not needed? The fact the Israelis can fly to Dubai did not deter Diaa Hamarsheh from killing 5 people in my home town, Bnei Brak. IDF presence in Jenin (Diaa's home) is needed to detect and prevent future attacks. Until when? Until you advise Palestinians to accept co-existence.

6.2.2022 11:03pm - Again, there is no need for #ADL to display its credentials (as "good" Jews) by the trivial slogan "one can reasonably object or criticize Israeli policies." We count on #ADL to inspire unapologetic, secured Jews. The image is libelous on its on de-merit. @JGreenblattADL

6.2.2022 10:51pm - I don't see why one must be #antisemitic to be dangerous, repulsive and stupid. Suppose this image points exclusively to the Israeli-Palestinian conflict, and blood-libels never existed in Jewish history, shouldn't everyone be appalled by this image? Shouldn't #ADL condemn it?

6.2.2022 9:59pm - (Replying to @MerkinMuffley5) I'm responding to the practicality issue (not to populist slogans). Permanent occupation does not solve a thing, but temporary control, awaiting your enemy to accept co-existence is a necessary measure for survival, and might just lead to peace, if Westerners focus on the hurdle.

6.2.2022 9:35pm - Brilliant! This is the most convincing explanation I've heard. It does not lean on "antisemitism", not even a latent one, and it explains why Asian-American departments are the first to slander Israel as the source of absolute evil. We need to study the anatomy of "projection".

6.2.2022 7:50pm - Well put, with one addition: 4. Western Anti-Israel NGO's. Moreover, terrorist groups have ceased to exist; they have been replace by "freedom fighters."

6.2.2022 7:30pm - (Replying to @yudapearl and @MerkinMuffley5) As to mental illness, I don't know you personally, but I honestly believe there is something profoundly wrong in the mentality of Israel de-legitimizers who do not pause even once to remind Palestinians how they can get freedom and dignity, if this is indeed their goal.

6.2.2022 7:10pm - (Replying to @MerkinMuffley5 Let's analyze it a bit. Assume for the sake of argument that Palestinians are dead-sworn to eradicate Israel no matter what Israel does. What human rights should they enjoy while Israel is preventing them from fulfilling their dreams? Voting? Importing weapons? Come and go?

6.2.2022 5:19pm - (Replying to @MerkinMuffley5) This is one way of putting it. But progressives are enlightened, right? So they shouldn't ignore an alternative theory, according to which the "occupation" is a temporary control of a territory from which millions of people declare their sworn intention to destroy a neighbor.

6.2.2022 5:09pm - Well said. Did I say it?

6.2.2022 4:54pm - (Replying to @FJnyc) I think her sign is very very meaningful: A society that lacks history and lacks distinct culture would resort to any trick, even the age of its elderly, to fake a semblance of "culture".

6.2.2022 4:38pm - (Replying to @MerkinMuffley5) The pathology resides not in condemning one policy or another but in obsessively negating the legitimacy a people's homeland and in willfully suppressing the possibility that at least some of those policies are forced upon Israel by its neighbors' actions and declared intentions.

6.2.2022 1:19pm - For educators, instructors and bystanders who are waiting for gentle material for teaching causal inference in AI and ML curriculum, here is a paper titled "Causal Inference in AI Education", which has just been accepted for publication by JCI:
6.2.2022 1:19pm - The course material is a product of high school classes, taught by Forney and Mueller, and reflects both: (1) students enthusiasm in those classes and (2) the dire need to enrich the causality-shy curriculum of standard data-science programs.

6.2.2022 1:24pm - (Replying to @MerkinMuffley5) If you have found a non-pathological explanation for this obsession, please share. Readers are seriously puzzled.

6.2.2022 12:44pm - Looking forward to you book. I hope it doesn't limit its coverage to the absurdities of modern Zionophobia, but includes a serious psychopathological study of this obsession. It defies every scientific explanation I could find, thus contributing to its longevity.

6.2.2022 11:03am - (Replying to @soboleffspaces) It takes philosophers 1-2 generations to rid themselves of "problems" that older generations have concocted. This is what happened, for example, to "evidential decision theory."

6.2.2022 10:49am - How I wish you were right @mishtal . How I wish Islamists were our only obstacle to peace. In truth: the overwhelming majority (99%) of our neighbors refuse to accept Jewish self-rule; from toddlers to grandmas, from teachers to Imams, from politicians to entertainers, from ...

6.2.2022 10:11am - The first logical rationale for destroying a state: YOUTH. And they call it "culture".

6.2.2022 9:57am - Scientifically speaking, progressives animosity towards Israel is the most challenging, and barely studied psychopathological puzzle of our time.

6.2.2022 9:32am - (Replying to @soboleffspaces) Backtracking cf doesn't solve any problem. But some philosophers insist that the utterance: "Had the soldier shot, the captain would have given a signal" is legitimate, rather than an incorrect distortion of "if the soldier shot, we can conclude that the captain signaled."

6.1.2022 11:27pm - (Replying to @GnomeSortingHat) Trying

6.1.2022 11:19pm - The philosophy of "sensitivity analysis" has a totally different flavor in causal analysis, where the crucial assumptions are causal, not statistical. See @ParpiaSmeer

6.1.2022 11:04pm - Philosophy-trained readers should find this dissertation a refreshing departure from philosophical theories of causation, in its attempt to operationalize "closest world semantics". I remain unconvinced that backtracking counterfactuals are imperative.

6.1.2022 10:10pm - June is graduation time, which always evokes the best of my inspirations. This "Love Letter to the Jewish Class of 2022," by Mara Kleinberg, evokes heartwarming themes from my 2019 graduation speech at UCLA,, which I now re-dedicate to the Class of 2022.

6.1.2022 5:52pm - (Replying to @MarkusSchacher) The article conveys useful information about CI implementations. But the technical portions could stand polishing.

6.1.2022 2:49pm - June 1, 1941, is the day of the Farhud pogrom in Baghdad My late wife was 6 yr old when the looting and killing broke up in her street. This is how she described Jewish Baghdad in her own words:

6.1.2022 1:11am - The violence comes from the pepper spray gun the "old lady" is holding. Watch the guy with the blue hat crunching after being sprayed. He was interviewed on Israeli TV, to explain how one could possibly be less than brotherly to an old Palestinian lady.

6.1.2022 12:24am - (1/2) Our persistent clinging to good old probability theory is one of the main obstacles to the understanding of causality. Here is a recent Royal Society paper falling into this rabbit hole: I was happy nevertheless to find a reference to Papoulis (1965),
6.1.2022 12:24am - the book from which I learned probability and stochastic processes. Athanasios Papoulis (1921-2002) was also on my PhD thesis committed in Brooklyn Poly, 1965.

6.1.2022 12:04am - UN names Palestinian media program for slain Al Jazeera reporter via @timesofisrael
6.1.2022 12:04am - In judging the wisdom of this UN media program, permit me to mention an alternative program: The Daniel Pearl International Journalism Institute which I believe stands up more faithfully with the Hippocratic oath of the journalistic profession. @Ostrov_A

5.31.2022 10:35pm - (1/2) Best analysis by far of the death of Abu Akleh. Related in some fundamental way to the discussion we had about the importance of explanatory theories in Popperian epistemology. The idea that Israeli soldiers have everything to lose and nothing to gain
5.31.2022 10:35pm - from shooting a journalist never occurred to the reporting team - all baked in @AlJazeera oven. But now that public embarrassment is surfacing on a daily basis, would @CNN retract their premature accusations? It takes leadership, from another oven, and I'm not sure CNN has it.

5.31.2022 7:48pm - If Jewish leadership failed us on Jewish Heritage Month, @newzionists should fill the gap, since Israel is the culmination of Jewish heritage. How about extending it by one week, till June 7, and make up for this sloppiness by celebrating seven pillars of Heritage, one each day.

5.31.2022 5:07pm - Hats off to @emrek and @amt_shrma for the development of DoWhy and its new PyWhy platform. Next step is for @MSFTResearch to help starving academia re-educate the thousands ML folks who are stuck on curve fitting and will soon be needed in the workforce.

5.31.2022 2:41pm - (Replying to @memosisland and @MSFTResearch) Thanks for alerting. I just retweeted to cheerful readers.

5.31.2022 2:37pm - DoWhy evolves to independent PyWhy model to help causal inference grow
5.31.2022 2:37pm - Gee, this looks like one of our slides, cheering us to know that at least one company, Microsoft, is on the right track. Makes me wonder what keeps the other ML companies in business ... I know, shortsighted funders. @erichorvitz

5.31.2022 4:59am - Irwin Cotler, former Minister of Justice and Attorney General of Canada, is the most knowledgeable Human Right scholar I've met. Watch his video below, regarding Zionophobia and its subliminal process of spreading.

5.31.2022 4:38am - And I am amazed that the author seems totally unaware of modern Causal Inference, a science that formalizes and operationalizes the interplay between mental theories and experimental data. He certainly has not read #Bookofwhy or modern critics of RCT

5.31.2022 4:04am - Print Issue: Uncovering a Curriculum | May 27, 2022 via @jewishjournal
5.31.2022 1:19am - If they agree to host the next AI-ML conference in Riyadh, I'd volunteer to organize a Causal-AI workshop, and submit a paper on Middle East Peace - Necessary and Sufficient Conditions.

5.30.2022 11:59pm - (Replying to @TriciaRoth18 and @deborahlipstadt) Thanks, Tricia. I've just asked Alvin what affiliation exists between his Institute and the new Center. Aren't you surprised by the absence of "Israel" in the Center's charter? For me, it's a warning sign of reluctance to face issues, or reluctance to rethink stale paradigms.

5.30.2022 11:18pm - Not bad for a politician

5.30.2022 8:25pm - Can data reveal the flow of ideas? Here is an interesting paper "ON THE EVOLUTION OF A.I. AND MACHINE LEARNING: TOWARDS MEASURING AND UNDERSTANDING IMPACT, INFLUENCE, AND LEADERSHIP AT PREMIER A.I. CONFERENCES" But who influenced whom? & Where is AI going?

5.30.2022 6:51pm - Seeing a new Jewish Center always cheers my heart, but seeing one that does not explicitly address the pressing issue of our time - Zionophobia - makes me worry: Isn't it another waste of Jewish philanthropic resource, or another license for University inaction? @deborahlipstadt

5.30.2022 6:11pm - Socrates would say it more poetically: If my culture, language and holidays offend your religious sensibilities, you better switch religion.

5.30.2022 3:40pm - Funds Cut From City Program for CUNY Law After Faculty Backs Israel Boycott
5.30.2022 3:40pm - I beg to differ. Zionophobia is the only politically acceptable form of racism which exists, not anti-Semitism. Council member @InnaVernikov decision to cut funds from @CUNYLaw will be more effective if aimed at the right virus. @SAFECUNY

5.30.2022 1:59pm - Reminds me how our son, Daniel, got lost on this very beach when he was 6 - abruptly swallowed by the crowd. We found him 3 hours later, two miles away, drinking soda with the life guards. "Why were you worried" he asked.

5.30.2022 1:15pm - China's nervousness about what Israelis think of her is a good sign. Next they might be nervous about what their own citizens think, and that's democracy live.

5.30.2022 12:58pm - (Replying to @CotlerWunsh) Must be a brain child of @KenRoth . We can forgive them for the blood libels, but we can't forgive them for staining the words "Human Right"

5.30.2022 12:42pm - (Replying to @cubic_logic) We are light years away from each other. To know how to continue, please tell me if you've read #Bookofwhy. My view is diametrically opposite: What you want and what is possible CANNOT be dictated by statistical properties of the data. See many examples in

5.30.2022 7:38am - It was about a year ago that Stanford came up with the term "Foundation Models". Our discussions did not help me understand what it is, and I thought the term has died. Today I got this message: So I'm wondering: Anyone knows what it is? @machinelearnflx

5.30.2022 7:23am - (Replying to @RDub2 @sadneurons and 2 others) So the "spuriousness" was not in the kind of regression you aimed to estimate, but in the method you used to estimate it? I would not call the latter "spurious correlations" because what you got is not the correlation but some surrogate of it.

5.30.2022 3:34am - (1/2) The authors of JDA make their living by showing how creative they can be in generating nuances and sub-nuances of definitions. What history will never forgive them for is doing their acrobatics on anti-Semitism, calling it "Declaration", and appropriating the name Jerusalem,
5.30.2022 3:34am - thus depriving our students of the little protection they can derive from the label "anti-Semitism" and the general revulsion it evokes. History will judge these authors by the consequences of their action. @EinatWilf @GilTroy @havivrettiggur @BenMFreeman @GeraldNGOM @blakeflayton

5.30.2022 1:35am - (Replying to @hosamindeed) Joy is in the eyes of the beholder. These girls seem very joyful to me: perhaps because I know their song. An Al-Jazeera watcher will see no joy in anything expressed by an Israeli, unless it can be used to vilify Israel. Yes, Zionophobia is a disease.

5.30.2022 1:05am - (Replying to @cubic_logic) We need a description of what you want -- thus far missing from our discussion & we can't continue without one.

5.30.2022 12:51am - (Replying to @cubic_logic @gottfriedmath and 3 others) "Fooled by attempting" requires some de-cryption. Do you mean "you can't estimate them" or, "even if you estimate them, you won't get what you want."? If the latter, we need a description of what you want -- thus far missing from our discussion & we can't continue without one.

5.30.2022 12:29am - (Replying to @stephensenn @AdanZBecerra1 and 3 others) My serious point is that we should separate definitions from estimation. One can discuss whether Pearson's "spurious correlation shock" was justified without detailing how he estimated those correlations.

5.30.2022 12:18am - The authors of the "Jerusalem Declaration of Antisemitism" will have to answer history for the consequences of their action. They have given hard anti-Semites a powerful new argument: "You see, even Jews do not agree on what antisemitism is." @deborahlipstadt @JGreenblattADL

5.29.2022 11:42pm - (Replying to @nirhasson) Why don't we see any clips of the dancing girls in the west. Can you share some?

5.29.2022 11:32pm - (Replying to @bianca_aguglia) Many papers, books and other resources can be found on my website Works by other teams are well referenced, but only if they are relevant, clear and correct.

5.29.2022 9:48pm - The question below lured me into re-reading the first 6 pages of Wow, I'm so enchanted by how clearly and beautifully it's written, that I can't resist but invite you all to enjoy it. Please compare to what we have been seeing lately in ML, Stat and Econ.

5.29.2022 9:15pm - (Replying to @cubic_logic) Disagree. What Pearson lacked in 1899 had nothing to do with the way correlation was estimated, but with the way it was interpreted. See Stigler "History of Statistics" or @Bookofwhy p.79. “To those who persist in looking upon all correlations as cause and effect ... a shock.”

5.29.2022 8:58pm - (Replying to @cubic_logic) "Spurious relations that may come up by the tool itself" should not be called "correlations", but "wrongly measured correlations". "Spurious correlations" are still correlations, defined by the probability distribution governing the data, not by the tool chosen to estimate them.

5.29.2022 8:52pm - (Replying to @cubic_logic @gottfriedmath and 3 others) If you are trying to estimate correlations and you are estimating something else, you dont call it "spurious correlations". I would call it "wrongly estimated correlations".

5.29.2022 8:46pm - (Replying to @EliSennesh and @Eradicator_NYC) The problem is not "being more aggressive" but knowing to what to ask that would be both effective and in compliance with academic free speech. They think they know, but end up praising administrators for objecting to boycott, instead of demanding an explicit mention of Zionism.

5.29.2022 8:35pm - That is what the Jewish establishment said after Munich Beer Hall Party (1923). It's wrong! Hate must be nipped in its bud, and the bud is @CUNYLaw . @deborahlipstadt @JGreenblattADL @SAFECUNY

5.29.2022 8:19pm - Very good question. Ans. Causality textbooks all use statistics as a tool, enriched with causal relations and causal notation. See for a good example. Causal notation is what Pearson lacked in 1899, and what modern stat textbooks still lack in 2022.

5.29.2022 8:06pm - (Replying to @UweSiebert9 @sadneurons and 2 others) If "spurious correlations" may stand for anything that is useless, then we should not use the term "correlations". If I want to predict Y given X and I wrongly measure the correlation between Z and W, is that "spurious correlations" too? How about if I wrongly divide by 100?

5.29.2022 7:57pm - (Replying to @cubic_logic @gottfriedmath and 3 others) Is the problem with the pre-smoothing? Or with the correlations themselves?

5.29.2022 7:49pm - (Replying to @stephensenn @AdanZBecerra1 and 3 others) Causation was needed for defining "spurious correlations". We are now discussing @f2harrell 's "reliable association" vs. ""cherry-picked correlations". Are the latter a species of "spurious"??

5.29.2022 7:33pm - (Replying to @Wetassprior and @KeurigBad) That's the first reason I heard that makes some sense.

5.29.2022 7:23pm - CUNY Law Faculty Endorse Israel Boycott Resolution
5.29.2022 7:23pm - As I have been writing, again and again, CUNY will become a snake pit for Jewish students unless Jewish organizations get their act together and call on the Chancellor to explicitly welcome Israeli and Zionist students on campus, for their unique contributions, not as a favor.

5.29.2022 3:20pm - Amazingly joyful celebration of Jerusalem Day! With unexpectedly high number of participants and, despite Hamas threats, and contrary to Al-Jazeera "reporting", a successfully managed and peaceful celebration with no major problems. Hats off to Israeli police and Israeli public.

5.29.2022 2:46pm - (Replying to @JeremyBenAmi) Teacher: Happy birthday, Jenny. Jenny: Thank you, teacher Teacher: Charlie, what are you hitting Jenny for? Charlie: I wish her dead, I must spoil her birthday. Jenny: No way! It's my birthday! @JeremyBenAmi : Jenny, Do you have no critique of what your birthday turned into?

5.29.2022 2:00pm - (Replying to @stephensenn @AdanZBecerra1 and 3 others) I don't know why you are in such a hurry to talk causal inference. We were talking about correlations. Some were labeled "reliable", some "cherry-picked", and I was humbly trying to understand the difference.

5.29.2022 1:50pm - (Replying to @GaryMarcus and @rogerkmoore) I am hopeful that when my litmus test of "deep understanding" becomes standard,, people will stop confusing "deep learning" with "deep understanding".

5.29.2022 12:44pm - (Replying to @stephensenn @f2harrell and 2 others) I am still trying to understand what makes "cherry-picked correlations" less reliable than "reliable measures of association". To define "reliable" we need to agree on the research question and see which of the two gives us a better answer to the question asked. Puzzled.

5.29.2022 11:13am - Interesting, I was not aware of this struggle. But if you aim to distinguish between "cherry-picked correlations" and "reliable measure of association", I wonder what makes the former "unreliable", and whether we have a mathematical expression for the latter. Do we?

5.29.2022 12:46am - (Replying to @DavidDeutschOxf) Profoundly concise summary, agree. But I have yet to see an illegitimate use of power which is not justified as a prevention of its illegitimate use. And that includes Bin Laden justification of 9/11, the Arab armies attack in 1948, and Putin's invasion of Ukraine.

5.29.2022 12:20am - (Replying to @StillTr05207382) Good question and, indeed, the Model-based paradigm advocates "model-assisted" learning @slashML

5.29.2022 12:15am - I am watching images from the roof of Al Aqsa Mosque, the third holiest place to our neighbors, on which a nervous warrior is doing his best to protect the holiness of the place.

5.28.2022 11:52pm - As usual, the most thoughtful and inspirational expression of Yom Yerushalayim comes from @YKleinHalevi . Note his last remark on why we all need to be thankful. She'Hecheyanu!

5.28.2022 10:24pm - It would be amusing to read how @AlJazeera reports on these images, whether the rocks were not thrown by the police towards the peaceful prayers, and whether Al Aqsa is still in imminent danger.

5.28.2022 10:11pm - CUNY's president will claim, like all the others, that he is against divestment. That does not exonerate him from his duty to state clearly and explicitly why Israeli and Zionist students and faculty are most welcome to @CUNYLaw .

5.28.2022 9:50pm - Tell me if it doesn't sound more punchy: "Those greedy, thieving, Jewish supremacist, apartheid-loving Nazi Zionists keep trying to say that I am a Zionophobe"

5.28.2022 8:40pm - The #MachineLearning enterprise seems to be splitting. While some insist on going from "Data-Centric to Model-Centric" paradigms, others recommend going backward, from model-centric to data-centric: Can #BeADeepLearner post this?

5.28.2022 8:19pm - (Replying to @MaccormickIan @JessicaHullman and 3 others) I couldn't find much commonality between the #MachineLearning and #psychology papers considered. The former are data-centric. I am also wondering what kind of integration methods are available to those who wish to integrate predictive models & causal inference?

5.28.2022 7:52pm - (Replying to @martin_garcia_a) Yes, precisely. I got to know it by being on the examining board.

5.28.2022 7:37pm - Another slip of tongue in my lecture yesterday was the date when Pearson discovered "spurious correlations". It was 1899, not 1999 as I wrongly uttered. Please forgive, and lookup @Bookofwhy how surprised and confused Pearson was.

5.28.2022 7:20pm - (Replying to @erikbiz @brianstelter and @CNN) As I said, the team overseeing CNN editorials has changed, and is now dominated by Ex Al-Jazeera pseudo journalists. Actually, it needs not be "dominated" by this ilk, enough for it to be "seeded" with one or two loudmouths, for CNN to go down the slope of Al-Jazeera

5.28.2022 5:32pm - The best way to access the literature for each of the Seven Wisdoms of Causal Inference is to go to our website and search for the key words in the titles. The references in each of these articles will lead you to relevant works done by other teams.

5.28.2022 5:20pm - 80 years passed since the first time I visited Jerusalem with my mother. 50 years, since my son Daniel had his Bar-Mitzva at this place. But time does not seem to dim the special light that shines from this city upon her lovers, poets, and dreamers. Katonti Mi Tsair Bannayich.

5.28.2022 9:01am - This slide provides a taxonomy of the gifts that Causal Inference has given to science. Once a research problem is classified into one of these seven compartments, the relevant literature, algorithms and software packages can be harnessed to walk you towards a solution

5.28.2022 8:37am - (Replying to @erikphoel @philipcball and 2 others) I'm referring to PS, PN, and PNS which, although they involve probabilities, cannot be expressed by the language of probabilities alone (Rung-1), without counterfactuals, which are Rung-3. It is the counterfactual component that tells you where they emerge from.

5.28.2022 8:05am - Thank you Ross. Here it is, my talk of yesterday already made into a video on youtube, accompanied by the slides, ready to be used for fun, profit and science. "What is Causal Inference all About?"

5.28.2022 12:51am - No, No, @DerbyChrisW is not an antisemite - some antisemites are rational organisms. He is suffering from Zionophobia - a total collapse of one's moral compass. No treatment is known, though some research is in progress. Imagine UK under Corbyn. @mishtal

5.27.2022 10:48pm - (Replying to @sadneurons @f2harrell and @DrewLevy) I wonder indeed how authors of modern statistics texts are defining "spurious correlations". Are they still trying? They obviously cannot do it w/o causation but, since causation is still taboo-ish, it is always amusing to see how they try.

5.27.2022 10:26pm - (Replying to @DmitriGallow) You may be right. Hume tries to pass these two definitions off as one, by means of his innocent interjection “in other words." I am still surprised that 200 years passed and no philosopher screamed: Wrong! The two definitions are fundamentally different!

5.27.2022 10:03pm - Glad Plato's cave made an impact on you. I was considering deleting this slide, not to offend statisticians in the audience. But I think it is generally agreed by now that 20th century statistics has been a calculus of 2-dimensional shadows on the walls of reality.

5.27.2022 9:51pm - It was a pleasure answering these important questions which are raised and re-raised so many times on our Twitter account. I'm only sorry that the name Hubert Blalock escaped me, and my answer on causation in SEM came out shaky. For full account, see

5.27.2022 9:31pm - Excellent discovery. I am still waiting for a philosopher to scream: Hey! How come Hume changed his mind in nine years, from Rung-1 to Rung-3.

5.27.2022 4:00pm - (Replying to @mishtal) Can you link us to his wisdom?

5.27.2022 11:30am - (Replying to @HarrySurden and @AlexJohnLondon) Thanks for affirming the purpose #Bookofwhy

5.27.2022 4:55am - Paraphrasing Gillerman, it's a tiny country that ha contributed to the world more than all those who wish her dead.

5.27.2022 4:49am - I am shocked! When did Iraqis acquire Palestinian mentality? This is serious matter. If any of our readers is of Iraqi nationality, he/she might be hanged for tweeting on this account, since I have dual, Israeli-American nationality. Watch out!!!

5.27.2022 12:55am - Almost forgotten. A devastating fall into 1st place in # of Turing Prize winners per capita: Partially guilty of this fall, I dare make a personal appeal to our neighbors: Join us in falling, just say "equally indigenous" and things will fall into place.

5.27.2022 12:27am - As a scientist, I must note another terrible fall, straight into 11th place in # of Nobel Prize winners per capita Falling and falling...

5.26.2022 11:48pm - And for those who can't attend all the upcoming meetings on causal inference and graphical models, here is a competitions to consider: I am not sure what the rules are, but if you have a powerful problem solver, it may be the right arena to demonstrate.

5.26.2022 11:35pm - I am in receipt of an announcement for another interesting meeting: First Workshop Causal Representation Learning at UAI 2022: 5 August 2022, Eindhoven, The Netherlands, hybrid Submission deadline: June 6, 2022, 23:59 AoE

5.26.2022 9:30pm - (Replying to @TheAcsMan @hearNtalk and 2 others) Disagree. Editorial standards in every news outlet were instituted to prevent glaring flukes like that from happening. The fact that it did happen indicates that basic standards were not adhered to, namely that it was not a "fluke" but systemic change in the overseeing management

5.26.2022 5:52pm - (Replying to @TheAcsMan @hearNtalk and 2 others) CNN has a better track record, agree. That's why everybody is surprised by the amateurish way they reported on Shireen's death. My point: It's not sloppiness, but the result of a process where graduates of Al-Jazeera and Bir-Zeit have taken over CNN coverage of this conflict.

5.26.2022 5:43pm - Good news for inquiring readers. I've just been assured by the organizers that my talk tomorrow will be recorded and will eventually be made available on Youtube. How come so many readers want to know "Where data science is going" ?

5.26.2022 3:47pm - (Replying to @Tiff__Annie_ @EBRheum and 2 others) Why don't you take the lead and show them the way to say something meaningful, and expand on what I've started?

5.26.2022 1:18pm - (Replying to @humanai4ron and @HarrySurden) And I was hoping the first chapter alone would reconfigure readers neural architecture. Challenge for the 2nd edition.

5.26.2022 6:30am - (Replying to @hearNtalk @TheAcsMan and 2 others) Good point. People often ask me: How can you judge a network, or a politician, or a colleague by what they think about Israel - a small corner in the global scene? My answer: It's small, but it's a corner I know very well, allowing me to tell where they get their knowledge from.

5.26.2022 1:39am - In 2001, @CNN did not join the "Dancing Israelis" conspiracy theory. Why? It's editors and reporting staff were not so heavily dominated by graduates of the @AlJazeera school of journalism, where Qatari thinking is prerequisite for Qatari salaries.

5.26.2022 12:49am - Thanks for sharing. I see it for the first time, and Gee, had I known they would turn it into a movie, I would have acted more professor-like, as if I knew what's good for PhD students. Great movie.

5.25.2022 8:10pm - (Replying to @OJPAC @nyspolice and 4 others) Too early to tell if they are David Duke disciples or @AOC -inspired?

5.25.2022 7:14pm - I visited the place in 1941, which was a dark alley, hardly 3 meters wide. Enough to say a prayer, but not enough to grasp its historical significance.

5.25.2022 5:47pm - Fighting Jew-Hatred As Winners, Not Victims via @jewishjournal
5.25.2022 5:47pm - Suissa, as usual, has it right conceptually. However, the word "antisemitism," by its history and its very nature, connotes "begging for protection" -- unbefitting to winners. Winners' fighting word is Zionophobia. See @DavidSuissaJJ

5.25.2022 3:38am - (Replying to @mishtal @Ostrov_A and @noatishby) They can do it! CNN used to have such teams before they started hiring graduates of @AJEnglish school of journalism = Muslim Brotherhood in secular cloaks, British accents and Qatari salaries.

5.25.2022 3:18am - For CNN to restore its reputation as a journalistic news outlet, its teams must include at least one reporter who, at least once in his/her career, has unveiled, at least one human aspect of the 8 million human beings that make up Israeli society. @Ostrov_A @noatishby @mishtal

5.25.2022 3:16am - (Replying to @Ostrov_A) When Al-Jazeera "can prove" things, we must summon the world's greatest logicians, lest the logic of right and wrong be in danger of extinction.

5.25.2022 3:04am - Ruth Wisse, the only General who understands the war on US campuses: "Don't let the war of words ever be fought about Israel's nature, let it be fought about why you can't accept Israel." If only @Hillel and @ADL would listen.

5.24.2022 11:03pm - (1/2) The reliability of a "reporter", especially in the Middle East, is measured by the ratio of facts to accusations in the writings of that "reporter". Those interviewed by CNN barely reach a 1:9 ratio each. A combined report may obtain a higher level of plausibility, if those
5.24.2022 11:03pm - interviewed were on both sides of conflict, to cancel biases and extract facts. Unfortunately, CNN have selected witnesses who all share the same target of vilification. That makes the combined report less than 1:9 reliable, especially counting the glaring bias of the "combiner."

5.24.2022 1:07pm - I'm reminded that I'll be giving a talk on Friday on "What is Causal Inference and Where is Data Science Going" If you know the answer to one of the questions, the answer to the second is not far off.

5.23.2022 5:59pm - I bet if you ask an @AlJazeera reporter she won't know either.

5.23.2022 5:19pm - (Replying to @firoozye) I can tell you haven't read my son's stories from Iran, Pakistan, and North Africa. There is a different kind of journalism in the world, and @AlJazeera is its antonym.

5.23.2022 4:36pm - (Replying to @wayneholmes @AJEnglish and @ajimran) For @AJEnglish to be taken seriously, they would have to raise the ratio of facts to accusations from 1:9 to 9:1. For a rational listener not to revolt the stench of accusations takes a real stomach. My admiration.

5.22.2022 6:02pm - To readers requesting good books on the history of the Yishuv, I am recommending Anita Shapira "Israel: A history" (2012, Brandeis U.) Shapira is a first class, authoritative historian, who understands the pulse of history through its sources. @noatishby

5.22.2022 3:53pm - Speaking about Fisher and Wright, I chose the latter as my hero. Whereas Fisher talks about "causes," it was Wright who realized that Science had been negligent in denying this beast a mathematical symbol, and gathered the courage to create one (eg., an arrow) when none existed.

5.22.2022 1:32pm - It's hard to summarize the week that was better than @noatishby , except perhaps to explicitly mention @AOC as an icon of the hate mob. Why her? Because our tax dollars pay for her podium.

5.22.2022 1:03pm - There is a tiny difference. Nazi Germany did not say: "Some Jews are good Jews". The new haters are more sneaky, hence much more dangerous. They say or imply: "Nor all Jews are sub-humans, only Israel supporters." Even @AOC has learned the mantra: "My Jewish brothers and sisters"

5.22.2022 11:30am - (Replying to @instrumenthull @chdausgaard and @analisereal) I couldn't understand the setup. Must wait for @analisereal to translate to structural language. Glad we still have translators among the living.

5.22.2022 11:05am - Anyone knows of a serious study to assess whether the "ingrained habits of successful scientists" have become more "ingrained" in the age of internet and social media?

5.22.2022 10:55am - (Replying to @LomaahhMore and @oac) Thanks for the correction.

5.22.2022 3:55am - (Replying to @LomaahhMore) It's a mistake of many Westerners. They assume that Al-Jazeera venomous style turns readers off or, at least, will make readers question their reliability. It doesn't. Readers, even Westerners drink that poison like honey, and become like @OAC , true believers in Israel's evil.

5.22.2022 1:17am - You and I find it easy to ignore those sprinkles of hate, but their authors know better. Those sprinkles work subliminally and very effectively. Some of my rational colleagues asked me last week:"Why does Israel restrict Muslim worship?" as if there was some truth to it.

5.22.2022 12:58am - (Replying to @LomaahhMore @JoshuaBernardNY and 4 others) Triangulation, yes, as long as it excludes Al-Jazeera and its satellites, to maintain some axioms of geometry.

5.22.2022 12:52am - (Replying to @JoshuaBernardNY @LomaahhMore and 4 others) One dragon at a time.

5.22.2022 12:51am - (Replying to @JoshuaBernardNY @LomaahhMore and 4 others) Didn't you find Shmuel Rosner thoughtful, objective and accurate?

5.22.2022 12:40:am - (Replying to @aynumazi @AOC and @deborahlipstadt) The causal chain is: AOC-rants --->Audience concludes Israelis are inherently evils ---> Lunatics in audience angry at Jews being or supporting evil -->Lunatics translate anger to deeds.

5.21.2022 11:18pm - (Replying to @JoshuaBernardNY @aynumazi and 3 others) 'Productively' requires that readers be already involved and interested in your constructive solutions. I've been using the Jewish Press, knowing that my criticism of the Israeli Gov. will not be interpreted improperly.

5.21.2022 8:12pm - (Replying to @JoshuaBernardNY @aynumazi and 3 others) As much as I value your candid discussion, I believe you grossly underestimate the depth of @AOC animosity towards Israel and the imminent dangers of her demonization agenda. I will change my opinion if she once says: "equally indigenous" or "I met an Israeli grandmother".

5.21.2022 7:47pm - This article is too radical for me to endorse, partly for lack of domain-specific knowledge, partly for ideological disagreement. I can only opine that EBM is not scientifically defined, and the slow penetration of causal inference to EBM ranks smells from some form of tyranny.

5.21.2022 7:28pm - (Replying to @TheAcsMan and @strwberrybrain) The reason Israelis worship antiquities lies in the content they hope to find there. The reason their enemies destroy antiquities lies in the void they fear to find there.

5.21.2022 6:29am - Here is a thoughtful analysis of the killing of Shireen Abu Akleh As a US Representative, @AOC should be held doubly accountable for jumping to conclusions, and for indicting a whole nation for an imaginary crime that "our tax dollars are part of."

5.21.2022 5:47am - (Replying to @aynumazi @AOC and @deborahlipstadt) The right word is an inquiry into the source of those "disgusting theories". Someone with public stature is choreographing those theories for terrorists to find them explanatory. @AOC is NOT talking accountability, she KNOWS who the culprit is and demands collective punishment.

5.21.2022 5:24am - (Replying to @DorotheaBaur) Gee, I wasn't aware of the one on the right. I hope they don't charge me with plagiarism.

5.21.2022 5:05am - (Replying to @mishtal) I'll try to make them available.

5.21.2022 5:02am - (Replying to @AdamMilstein) Protection is not the answer. Jews should not be forced into American-made Ghettos. Hate should be nipped in its bud, and the bud is in the Squad. @deborahlipstadt @JGreenblattADL

5.21.2022 3:43am - (Replying to @aynumazi @AOC and @deborahlipstadt) Don't you agree that @AOC conspiratorial rants about Israel can trigger Buffalo type of madness? She even identified the soldier who killed Shireen Abu Akleh to make sure her audience hate the right people, for a good reason, and with no hesitation. Our tax pays for her podium.

5.21.2022 3:23am - Remember our Crash Course in Good and Bad Control? Well, it has been revised again (to pacify another reviewer) and it's now close to perfect, with references to everything that was ever done on "control", "adjustment" "covariate selection," OVB, Enjoy!

5.21.2022 3:06am - (Replying to @rakotesh @AngeloDalli and @artistexyz) I couldn't see where shadow networks are used or defined. The missing data problem addressed in this paper is treated using missingness graphs in Are the results different? More powerful?

5.20.2022 6:58pm - (Replying to @benj_kerstein and @BernieSanders @SenSanders) can't swallow the fact that his endorsement has become a liability.

5.20.2022 6:48pm - Some cultures (e.g., @strwberrybrain ) can't grasp that others may have respect for the past; they have roots there, not weapons.

5.20.2022 6:30pm - That's a good point. I've read dozens of articles on how conspiracy theories endanger lives, democracy and our sense of right and wrong. No one has called out @AOC for fueling such theories when it comes to Israel, not even @deborahlipstadt on NPR. Is @AOC not a Buffalo enabler?

5.20.2022 5:54am - (Replying to @mishtal and @CAMERAorg)
Shared: 5.20.2022 3:47am to start the conversation, I invite readers to envision how a "mutual recognition of past wrongs" might look like. Here is my contribution -- a 2008 piece in Haaretz titled: "Israel and Palestinians should apologize to each other". It's amazing how brutally relevant it is today.

5.20.2022 5:34am - (Replying to @mishtal and @CAMERAorg) Congratulations!

5.20.2022 3:47am - (1/4) A recent call for Palestinians to abolish 'Nakba Day' and rescind their "right of return" stands strikingly opposite to counter calls for Israelis to recognize 'Nakba Day' and to understand that such Palestinian compromise is infeasible, for it's as
5.20.2022 3:47am - (2/4) grotesque as asking Jews to give up their dream of 'return'. A third option claims that recognition of past wrongs should set aside political stalemates, for it may actually disentangle those stalemates. I believe this third option deserves serious consideration and,
5.20.2022 3:47am - to start the conversation, I invite readers to envision how a "mutual recognition of past wrongs" might look like. Here is my contribution -- a 2008 piece in Haaretz titled: "Israel and Palestinians should apologize to each other". It's amazing how brutally relevant it is today.
5.20.2022 3:47am - (3/4)
5.20.2022 3:47am - (4/4) @EinatWilf @blakeflayton @GeraldNGOM @YKleinHalevi @BenMFreeman @Eve_Barlow @GilTroy @CotlerWunsh @JGreenblattADL @bariweiss @newzionists @AndrewPessin @noatishby @rosnersdomain @ShMMor @Scholars4Peace

5.20.2022 12:27am - She is so helpless, so confused and so desperate that I feel all my empathy genes rushing to rescue her. Still, from the depth of this confusion emerges profound scientific knowledge: She KNOWS who killed Shireen Abu AQLA, not "maybe", not "most likely", confused @AOC just knows.

5.19.2022 11:42pm - (Replying to @Dam_Nuwen) Before dwelling into History books, which can be as bad as their authors, here is an article just published by a Palestinian HR activist, calling for correcting the Palestinian narrative and abolishing the Nakba Day: @EinatWilf @StandWithUs @blakeflayton

5.19.2022 10:50am - (Replying to @soboleffspaces) Are you asking about the Crash Course ??, If so, I know it was accepted, I'll ask for estimated publication day.

5.19.2022 10:48pm - In DAGs, counterfactual nodes Y_x exist in the mutilated models, M_x, as Fig. 44 To relate them to the original graph we construct a twin network, or ask: What exogeneous variables U* (in M_x) are responsible for variations in Y. Y_x has same stat as U*.

5.19.2022 10:21pm - (Replying to @Dam_Nuwen) I am glad we talked, because we are now seeing with our own eyes how some (perhaps most) people see the conflict, as told to them by Arabs. Behold: when Arab attack its "there were riots" and "war came" (from Mt. Olympus), when Jews defend themselves its "resorted to terrorism".

5.19.2022 9:56pm - (Replying to @artistexyz) Not clear why you dont agree with Fig. 16(b). Can you explain why w/o going to other books? (Y_0, Y_1) is simply U_y.

5.19.2022 6:17pm - Thanks, David, for resurrecting this video. It's amazing how little has changed on campus. With two exceptions: (1) BDS got much bolder, they no longer hide their eliminationist agenda, and (2) Bruins for Israel is much stronger (within Hillel), they fight Zionophobia directly.

5.19.2022 4:45pm - (Replying to @einsley23) The way to test Ignorability conditions is indeed tricky, even with graphs (let alone w/o graphs). I simplify it here:; figure 8. What your argument is missing is the term U_z, a second parent of Z, which is part of Y_x,

5.19.2022 1:33pm - (Replying to @Dam_Nuwen) "enlightened antisemitism: is not "hating Jews because they're Jews". It is placing every collective in its historical place. For example, placing Jews in their historical place as protected and highly respected minority, yet not in control over their own destiny.

5.19.2022 10:50am - (Replying to @soboleffspaces) Are you asking about the Crash Course ??, If so, I know it was accepted, I'll ask for estimated publication day.

5.19.2022 3:42am - This is not Gazza, nor Kabul. This is Birzeit University whose graduates apply for faculty positions at US colleges. Once hired, they teach "international relations" and pose as "human right activists".

5.19.2022 2:47am - The challenge for Republicans to eradicate the replacement theorists from their ranks is just as daunting as the challenge for Democrats to eradicate Zionophobes like @RashidaTlaib , @IlhanMN , @AOC and perhaps @marwilliamson too.

5.19.2022 1:45am - Bob Dylan’s Bohemian Zionism via @jewishjournal

5.19.2022 1:45am - This song should be part of the Passover Haggada.
5.19.2022 1:45am - The greatness of a song is in its longevity. I can't think of more fitting lyrics to current events in Israel and Western hysteria over them? "Well, the NEIGHBORHOOD BULLY, he is just one man. His enemies say he is on their land." Thank you for giving voice to a lonely bully.

5.19.2022 12:21am - (Replying to @Dam_Nuwen) The subjective perceptions of "weak" and "first" are part of the enlightened antisemitism. By objective standards, living 74 years under real threats of annihilation is a terrible weakness, despite the surface appearance of a "military superpower".

5.18.2022 3:05am - (Replying to @Foreman1David and @TIMO_32paths) It still does not explain why progressives, usually of above average intelligence, would fall for such infantile propaganda, regardless of the billions that fund it. We need a causal explanation for the foolability of good progressive minds.

5.18.2022 2:52am - This Tweet should be illuminated by the interview I did with David Hand: "What Statisticians Want to Know about Causal Inference" If you are a statistician, please tell us if Hand's questions captured yours.

5.18.2022 2:28am - (1/2) Difficult question. I'm oscillating btwn two theories: (1) Enlightened antisemitism - Jews should be protected, even respected, but not equal. It dosn't fit their historical role and our benevolent mold. (2) Zionophobic dissonance - Textbooks say that a nation-state based on
5.18.2022 2:28am - (1/2) peoplehood and historical lore must be evil and repressive. Israel proves all textbooks wrong, by being successful, democratic and pluralistic -- it will never be forgiven. See I'd like to examine more theories, because it's indeed a perplexing puzzle.

5.18.2022 12:44am - (Replying to @EliLake) It was not intentional, agree. But the clarification could have been less on what we didn't understand, and more on what she understands the leading role of Jews to be in bigotry-fighting America.

5.18.2022 12:30am - You should qualify it and admit that it was not a complete failure - they are still trying. @RepRashida struggles to do with words what bullets started in 1948.

5.17.2022 11:54pm - Terrible confusion. But please note that Cabal "Replacement theories" are not confined to white supremacists. "Jews will not replace us" was the first song my grandfather heard when he landed in Jaffa, 1924, and Arab workers thought he was not indigenous to the land.

5.17.2022 11:46pm - "Jews will not replace us" was the first song my grandfather heard when he landed in Jaffa, 1924, and Arab workers thought he was not indigenous.

5.17.2022 11:07pm - Shae will be badly missed in Bernie's camp.

5.17.2022 11:04am - (Replying to @TheAcsMan) Sure, the capital and knowhow brought by the Zionist pioneers created thousands of jobs, and attracted thousands of Muslim immigrants from Egypt and Syria. We called the latter Horranim, from Horran, and I recall seeing them in Tel Aviv vying for day-jobs.

5.17.2022 10:51pm - Reminds me of Oprah's quote:" If someone tells you who he is, take his word." Only Western analysts say "He doesn't really mean it" or "That's how Arabs start the bargaining process"

5.17.2022 10:45am - (Replying to @TheAcsMan) My theory of indigeneity makes both Jews and Palestinians "equally indigenous" to the land, the former by historical attachment, the latter by physical presence.

5.17.2022 10:33am - Thanks for resurrecting. This indeed was one of my most influential papers among statisticians, written in the heat of finishing the 2nd Edition of Causality, I packed everything into it.

5.17.2022 9:48pm - (Replying to @PeterBeinart and @blakeflayton) Two weeks of Pinsker may do it.

5.17.2022 9:24pm - (Replying to @Vandalay_Inc and @PeterBeinart) Palestinians' rejection of the partition plan can be understood, given some effort. What transformed the nature of the conflict, and left its traumatic marks to this day was Azaam Pasha's proclamation: "a war of extermination and momentous massacre" (Cairo, Oct 1947). I was 11.

5.17.2022 3:12pm - I have just posted a new blog, titled: "What statisticians mean by `Causal Inference': Is Gelman's blog representative?" It should help, I hope, to define key terms. #causalinference, #datascience, #statistics, #ML,#Bookofwhy

5.17.2022 2:45pm - (Replying to @adamalmo and @blakeflayton) It was politically useful for Shabtai Zvi too, since his followers were all Jews. But the Rabbis of Constantinople decided (1650) Enough is Enough & excommunicated him. Our Rabbis don't have the guts nor authority. Shabtai got the mesg and converted to Islam. Will @PeterBeinart ?

5.17.2022 5:12am - I keep on maintaining it's a stain on the record of each one of CUNY's faculty members not to come to Guntz support and not to reclaim @CUNYLaw as part of the civilized world. Spineless faculty makes bullies bolder.

5.17.2022 3:35am - (Replying to @blakeflayton) The transformation is not complete yet. Like Shabtai Tzvi in 1650, it's time for Peter Beinart to officially convert and declare himself an Ex-Jew.

5.17.2022 3:03am - Say what you will about symbols, but I am sentimental about flags, and this picture makes me almost spiritual: Israel and Ukraine, two beleaguered countries, whose neighbors decided do not have the right to be.

5.17.2022 12:59am - If this was a robot, not Kamala Harris, people would be complaining the AI is going nowhere, and that, despite what futurists are telling us, AGI will never be achieved.

5.16.2022 11:37pm - (Replying to @Tim_Akintayo) Yes, "experimental" for me means RCTs, or any method that you strongly trust to faithfully imitate an RCT. Namely, one that delivers P(y|do(x)).

5.16.2022 8:58pm - HMM, David Rothkopf blocked me, so let's start with basic logic: What Binds American White Supremacists and Palestinians Brutal Assaults on Israelis? "Great replacement theory". Namely, the theory that Jews are alien to the land, hence can be shipped off someplace and replaced.

5.16.2022 8:28pm - (Replying to @HenMazzig and @CAMERAorg) Congratulations!

5.16.2022 8:20pm - Here is my version: Bernie Sander did Buffalo! How? In the Israeli-Palestinian context, "great replacement theory" is expressed as opposition to Israelis right to a homeland, which treats Jews as aliens to the land. Sander advisors constantly declare support for this notion.

5.16.2022 7:13pm - (Replying to @LenGrunstein @MikeNew19512115 and 15 others) Same, thanks.

5.16.2022 7:10pm - While we are arguing, back and forth, the cultural and mental effects of the internet, the Ayatollahs found a solution, they cut it off entirely, to prevent people from writing about bread shortage. It didn't occur to any of us.

5.16.2022 3:36pm - (1/3) Just posted another comment on Gelman's blog: Individual effects are well captured within the framework of causal inference, because every SCM (structural causal model) can be regarded as “mechanistic model” and gives you, for every individual u, the counterfactual Y(x,u),
5.16.2022 3:36pm - (2/3) for every X and Y. At the same time, Gelman’s statement that you need the mechanistic model to get individual effect is wrong. See Causality chapter 9, and 11.9.1. Also, see recent conceptual paper on personalized decision making Note: No model at all
5.16.2022 3:36pm - (3/3) is needed, just experimental and observational data from the same population. Gelman's post re-enforces my earlier conjecture that behind every statistician there is a philosopher of science, namely, someone who believes that statistics has all the answers.

5.16.2022 1:53pm - A rare, half-way agreement with @StatModeling who says: "statisticians use the term “causal inference” when we’re not trying to model the process." True, they give up. Causal Inference folks use this term when they ARE trying to model the process, and succeed.

5.16.2022 12:19pm - (Replying to @ahtouati) If you know the source, where was it taken from? wasn't it violent? Wasn't the funeral violent? Does it negate my conclusions that she had a Royal funeral and that she was a propagandist more than a journalist?

5.16.2022 11:52am - (Replying to @DluLevine @digitald0ctor and 2 others) This is also the story I heard from my dialogue partner Dr. Ahmed from Pakistan, who witnessed the events first hand. The transfer was mighty bloody on both sides.

5.16.2022 3:40am - Correcting a tiny glitch in this post. My UCLA talk will take place on May 27, 2022 @ 12:00 pm - 1:00 pm, not May 26. Giving me one extra day to think: What is Causal Inference all about?

5.16.2022 2:43am - @IfNotNowOrg is the brain child of @NewIsraelFund , created to help Jewish youngsters offer their first Pound of Flesh on the altar of social acceptance. And to end up like @PeterBeinart , the Messianic Prophet (Shabtai Tzvi) of a totally flesh-less pack.

5.15.2022 11:29pm - (Replying to @erikbiz) The more educated they are, the more freedom they feel to express their inner eliminationist sentiments. Look at @4noura at Rutgers and her crew, she has no qualm at all committing Israelis to non-existence, like vermin, laud and clear, no fear of being heard.

5.15.2022 11:12pm - (Replying to @kailashbuki @icmlconf and 2 others) Thanks for this interesting paper. I am still puzzled by how outliers were analyzed in the pre-causal world of statistics proper. What questions were asked? Was there an analysis to speak of, or just collections of funny anecdotes?

5.15.2022 11:02pm - Behold! Palestinian culture at its best - a cesspool of deceit. They think people do not see through it, and they fail to see the damage it does to Palestinian credibility.

5.15.2022 10:52pm - (Replying to @AmbDaniDayan @ofirdayan94 and 2 others) Mazal Tov! Ofir and Dani. May this degree give you the credential to amplify the inspiring works you have been doing for your people.

5.15.2022 10:28pm - I agree with everything @AmbAsafZamir says in this @CNN interview, except one: Who says #ShireenAbuAkleh was an "excellent journalist"? Not by my criterion: She has violated journalism Hippocratic Oath by de-humanizing a nation of 8 million refugees.

5.15.2022 10:06pm - (Replying to @MaartenvSmeden) Just "weight of evidence"

5.15.2022 10:03pm - The key difference, indeed, is the treatment of flags. More than any other symbolic act, this asymmetry encapsulates the core of the conflict: Coexistence on one side, vs. eliminationism on the other. Many good people died for each flag; disrespect reveals both aims & mentality.

5.15.2022 5:25pm - My comparison of my son, Daniel Pearl, and #ShireenAbuAkleh was meant to highlight the difference between a true journalist, who never forgets the human face behind the news, whatever the news, and a propagandist, who consistently dehumanizes one side in a conflict.

5.15.2022 5:04pm - (Replying to @tsuleiman03) Readers who send me questions, please be specific. I am not sure Al-Jazeera definition of oppress/oppressor coincides with mine. I view the 74 yrs denial of normalcy to Israelis as an act of cruelty and oppression. Also, does "killing" refer to the 20 Israelis killed last week???

5.15.2022 11:05am - (Replying to @satisficer @Citizen22668787 and @LibsterSergey) It does not make sense to me too, unless of course the Israeli version is correct, and these "pallbearers" snatched the coffin from the church and from the family to conduct their own show.

5.15.2022 2:46am - A cheerful news from friends in Israel: The #Bookofwhy will soon be translated into Hebrew - I can't wait to see a copy in my high school's library, discussed in the same language that God tested Adam's causal thinking: Hey, did you eat from that tree that I told you not to?

5.15.2022 2:13am - (1/2) #ShireenAbuAkleh has received the most lavish royal funeral than any journalist since the Queen of Sheva. As a father of a slain journalist, and as a quantitative scientist, I dare question her journalistic credentials. I ask: In her 25 yrs of covering the Middle East, has
5.15.2022 2:13am - (2/2) she once described the human face of Israeli society? How many times has she portrayed an Israeli child or a grandmother or any human feature of 8 million people trapped in that conflict? Is a persistent demonization of a people a form of journalism? Or is it sheer propaganda?

5.15.2022 1:28am - We can understand and empathize with Palestinians who believed in 1948 that five armies and genocidal rhetoric could prevent a nation from returning to its historical homeland. We cannot forgive them for maintaining this same belief today, Western rhetoric notwithstanding.

5.15.2022 12:58am - Here is a typical Twitter tempting me to mock the flag of Israel's neighbors, which I will refuse to do, as long as there is some chance they would reciprocate.

5.15.2022 12:40am - The Five Miracles of Israel via @jewishjournal
5.15.2022 12:40am - (1/2) For me, May 14-15 will always be that Friday afternoon, Tel-Aviv Museum, when David Ben-Gurion declared the establishment of Israel, the dancing all that night, followed by a fearful morning under bombing of Egyptian war planes. Suissa's five miracles could be supplemented
5.15.2022 12:40am - (2/2) by at least 74 others, but I would add one that pertains to recent events in Israel, which lured every thoughtless 'who is who' in the media to make a nasty accusation of Israel. It's the miracle of remaining united behind the noble dream of Zionism, w/o a bubble of approval.

5.14.2022 10:25pm - One quote of @GloriaSteinem strongly resonates in my ears: "Feminism includes all women. Or it isn't feminism." It reaffirms my conclusion that Linda Sarsour (and other Women's March leaders) are fake feminists cannily excluding Zionist and Israeli women.

5.14.2022 5:57pm - Speaking about music, a friend sent me an old video that I didn't know exists: Arioso for spontaneous trumpet and bass:

5.14.2022 5:10pm - (Replying to @TaliaRinger and @plain_simon) I have zero knowledge of cubical/homotopy type theory. Any chance I could get motivated by a simple example, say the firing squad?

5.14.2022 5:02pm - To all fans of Primer Corrected copies of the book should now be available from all booksellers. I am holing a physical copy in my hand. Warning: Watch for the statement "Reprinted with Revisions 2021" on page iv. Enjoy! Return if you get a wrong page iv.

5.14.2022 3:04am - My next talk will be on May 26, at UCLA, I believe every CI researcher bears a responsibility to tell campus-wide audience what Causal Inference is about, and why it holds the key to the future of data science.
5.14.2022 3:04am - Driven by the responsibility to tell my UCLA colleagues what Causal Inference is about, and why it holds the key to the future of data science, I'll be giving a seminar

5.13.2022 6:56pm - When reports differ and inputs conflict, epistemology advises us to use logic, plain logic, and ask ourselves: Who has greater incentive to ensure a peaceful, civilized and dignified funeral procession? Israeli police? Or Hamas operatives?

5.13.2022 2:05am - (Replying to @SusanSarandon) How long before we say "enough" to the culture of deceit and the credulity of Westerners like @SusanSarandon

5.13.2022 1:05am - A timely reminder of the deadly consequences of decipt. Watch the second episode of the fabricated martyrdom of Muhamad Al Dura

5.13.2022 12:34am - I do remember Muhammad Al-Dura. His picture was projected behind my son, Daniel Pearl, in the last video taken of him, ostensibly to justify his murder. How many people will now be killed due to Al-Jazeera's fabrications? When would this deceitful culture be told: LIES KILL!

5.12.2022 9:44pm - How evil and despicable of @MariamBarghouti to speak like that of a people's homeland. We have a dog on our street, he might have rabies, who thinks cars are going to stop, just because he barks "apartheid! apartheid!".

5.12.2022 5:53pm - (Replying to @JewishSpaceLazr) I concede. Yet there is another difference, not less importance. One that most Americans miss, including most Jews. The obsession consumes not only the leadership, not only the majority, but an entire society, educators, clerics, TV anchors, NO EXCEPTION worth mentioning - nada.

5.12.2022 4:06pm - There is a difference. For classical antisemites, Jews were irritants, deserving a pogrom here and a massacre there, whenever mood permits. For Palestinians the elimination of Israel is a consumptive obsession, prioritized over everything else, freedom, bread, family, everything.

5.12.2022 11:08am - (Replying to @questionsin2014 @SethAMandel and 2 others) Love it!

5.12.2022 10:59am - (Replying to @JlaorMD @BibimbapNom and @sunwins888) No, No, they do love Jews. But only "Good Jews", namely, stripped of sovereignty. As Al-Jazeera's philosopher @mehdirhasan used to say: "Not every nation deserves a state".

5.12.2022 10:12am - Why is the PA hiding the bullet that killed Shireen Abu Akleh? See

5.12.2022 9:41am - When was the last time a Palestinian official talked about anything at all, from climate to Covid to abortions, without casting blame upon Israel? They haven't read Leo Pinsker's "Auto-emancipation" (1881) and never understood elementary Zionism.

5.12.2022 9:20am - (Replying to @noranta4 @TacoCohen and 8 others) I always look at SCM as a computational embodiment of Popper's philosophy, in which falsifiability and, in particular "explanation" (rung-3) are given concrete, algorithmic interpretations. Is there a Popperian desiderata that is not achievable in SCM?

5.12.2022 3:32am - (1/2) The genie is out. Every student of counterfactuals can conclude now that #ShireenAbutAkleh was killed by Palestinian fire. How? A Palestinian pathologist holds the bullet and refuses to produce it. Had they verified that it came from Israeli weapon they would have celebrated
5.12.2022 3:32am - such a finding on every news outlet in the cosmos. We don't see that jubilee or that bullet. Done. Biden should demand that they produce the bullet since, after all, she was an American citizen, and Governments owe their citizens an in-depth investigation of cause of death.

5.12.2022 1:47am - Another brilliant piece by @EinatWilf (Paraphrsing) It's impossible to bully a person who doesn't seek approval. In other words, the only response to anti-Zionism, is secured Zionism.

5.11.2022 10:49pm - (Replying to @thinkyparts) Some skepticism and lots of admiration.

5.11.2022 10:36pm - We, in the West, fail to realize the mentality of Al-Jazeera 8M listeners. Here is a sample, asking "Who else could have killed her"? As if she was covering a girl-scout picknick, Kumbaya, and soldiers had nothing else on their mind that pastorale day but her. Who else?

5.11.2022 9:41pm - God forbids! Next they are going to be Shocked that Al-Jazeera lies, or that @RashidaTlaib imagines things.

5.11.2022 8:04pm - Why not state things you do not know as if you were an eye witness? Someone will surely buy it! Al-Jazeera mentality in the Hall of US Congress.

5.11.2022 7:39pm - (Replying to @TacoCohen @saramagliacane and 7 others) In my mind, the CRL intuition (hence math def.) is fairly clear: changing environment acts like an intervention. Done. The rest is DAG mathematics to find out under what conditions we can extract structure from the change, given that we do not know where the change occurred.

5.11.2022 5:32pm - Thank you for resurrecting this note. But just to make sure, let me post the published version I always get a kick reading Will Roger's: "When the Okies left Oklahoma and moved to California, they raised the average intelligence level in both states"

5.11.2022 3:51pm - (Replying to @overlordayn and @OpenAI) Interesting! Did you feed it with the functional SCM, or with the DAG only, or with data only?

5.11.2022 3:44pm - (Replying to @TacoCohen @overlordayn and @trunghlt) I think it's feasible, but not easy, considering that million years of evolution have not produced it in owls or even chimps. The key, I believe, rests in having an innate, domain-independent template that makes babies restless until they reach: "understanding", ie, domain-SCM.

5.11.2022 11:36am - (Replying to @mattshomepage @tdietterich and @TacoCohen) Same happened to me. First excitement, then disappointment and then a much greater excitement: a proof that you MUST have a DAG, or, better yet, that you can figure out what is the minimal thing you must assume before you can get an answer to your question. No longer disappointed

5.11.2022 10:47am - (Replying to @emilesilvis and @seanmcarroll) Would it make a difference?

5.11.2022 12:03am - (Replying to @DaniLubin) Would you target a journalist knowing that the world image of your country is at stake? If you would, then you never met people who love and care for their country. I have. And I know how super-careful they are to avoid any resemblance of "targeting" the innocent.

5.10.2022 11:29pm - The credibility quiz: Who should we believe? Israel soldiers, who know that each of their moves is photographed and scrutinized by a blame-happy world press? Or @AJENews , a Qatar based organization that has yet to report a single news from Israel that is not blame-laden? Who?

5.10.2022 6:17pm - Clouds over higher education. Sharing a thoughtful and gloomy article by Cary Nelson, pertaining specifically to UC and Illinois, The Anti-Israel Politicisation of the US Academy: The Next Phase is Happening at California and Illinois

5.10.2022 6:12pm - (Replying to @stephensenn and @eliasbareinboim) I noticed there are no formulas describing the "treatment assignment". Is this because you wonder if such formulas exist? Or because you just didn't think it was necessary?

5.10.2022 6:01pm - (Replying to @amalec and @seanmcarroll) Tiny fix. That theory could be true in reality, but it cannot explain the uniformity and consistency of human thoughts. And I am aftr the latter, which accounts for our ability to communicate.

5.10.2022 4:30pm - Dozens of Crimson Alumni, Harvard Faculty Join Outcry at Paper’s Support of BDS
5.10.2022 4:30pm - I like @MiriamElman ' solution: "Retract! It is an adult thing to do,”

5.10.2022 3:40am - (Replying to @bano_sanjeeda and @AJENews) I can tell that you have never met an Israeli soldier. I have met many; all my (grand) nephews and nieces are. Their humanity and care for human life, including Palestinians, far exceed anything I have seen from those who demonize them as you did. Far, far exceed!!

5.10.2022 2:47am - The history, as told by its players.

5.10.2022 12:23am - Unmissable, agree.

5.9.2022 11:35pm - It's not what Israel does that makes the Jackals scream "apartheid, apartheid"! but what Israel IS. @KenRoth and @4noura will continue their howling even if Israel anoints a Muslim a Chief Rabbi. Can't wait to prove it.

5.9.2022 11:03pm - One day, someone with scientific eyeglasses and causal language would go over your narrative of the history of "experimental design" and organize it by "tasks" "assumptions" "guarantees" etc. very much like @eliasbareinboim has done here:

5.9.2022 8:11pm - Great lecture by @eliasbareinboim on what Data Science is all about, and the role of Causal Inference in that enterprise. Elias calls it a "computer science" perspective; I would rather call it the "Scientific Perspective", for its universality and unification capacity.

5.9.2022 7:59pm - Readers are invited to sign this open letter, to dispel the rumor that Harvard faculty just don't care about their students in times of madness.

5.9.2022 7:17pm - Rumor has it that a member of Jewish Voice for Peace (JVP) was allowed to take the flight. How? He produced his latest Tweet, the Captain took one look and said: No Jew.

5.9.2022 2:59pm - What a pleasure it is to have a conversation with someone with a different perspective on things, not just "someone," but a true scientist with genuine curiosity into the nature of things. Thank you Sean for having me on your show, and loading me with new questions and insights.

5.9.2022 2:53am - (Replying to @stephensenn @MariaGlymour and 4 others) DAGs will stop you right there and ask you to define the problem before proceding. What are we trying to estimate? What are the assumptions? Are the centers recruitments the same? etc etc.

5.9.2022 2:45am - A renowned Palestinian journalist complained this week that the media fails to notice the humanity of the murderers. Note, he said, how they refrain from targeting women and children, following their religious principles. I'm humbled and awestricken.

5.9.2022 1:22am - The same holds for two other statisticians notable for philosophizing about statistical modeling, William Cochran and Leo Breiman. In and I laud their seminal contributions but note the inadequacy of the models in their disposal.

5.8.2022 10:04pm - (Replying to @stephensenn and @soboleffspaces) No contest intended of the accomplishments of pre-1990 statisticians. Just a reminder that when they philosophize about models, they have statistical models in mind, not models that could help them answer their research questions. And that includes Sam Karlin (#Bookofwhy p.87)

5.8.2022 7:48pm - (Replying to @shulmaven) I wish this was true of all economists. Among the ones I know, some do not appreciate the power of their own models, others hide their models to appear credible.

5.8.2022 5:35pm - I continue to blame faculty apathy for this culture. This would not take place at USC after the "faculty uprising" of Dec. 1 when the word Zionophobia reminded the administration that Zioinists on campus deserve at least as much respect as Muslims.

5.8.2022 5:17pm - (Replying to @elderofziyon @Ostrov_A and @TheElders) Impunity starts by tolerating the de-legitimization of a neighboring country, be it by Putin or by Palestinian leaders.

5.8.2022 5:11pm - Peter Beinart was born in 1971, after Yasser Arafat decided to Westernize his struggle into "equality and freedom." I was 11 yrs old in 1947 when Azzam Pasha (Arab League) echoed Hamas, and honestly called the Palestinian struggle: "a war of extermination and momentous massacre."

5.8.2022 4:46pm - This is a great slide. I love Picaso's quote, because it implies that if you can't tell a dumb computer what your problem is, you don't really understand what your problem is, no matter how fancy the math you are using.

5.8.2022 1:10pm - I felt so grateful, that I forgot whom I should thank, the Goddess of justice? The All Mighty?

5.8.2022 12:55pm - (Replying to @stephensenn @hmatejx and @RWJE_BA) If you have a calculus then, surely, you can write down formally what quantity you want to estimate ideally, and what quantity we are actually estimating due to study imperfections. I am eager to see these two formulas with my own eyes.

5.8.2022 12:46pm - (Replying to @stephensenn @hmatejx and @RWJE_BA) SCM can tell us how to analyze any data, if trialists only tell us formally how study1 and study 2 were conducted, what exactly was compared, and what the purpose is of the "analysis". So far, one can only guess.

5.8.2022 12:13pm - This Bill will have only one effect, falsely portraying Israel supporters as opposing the two-state solution as an ideal end goal. I'm inclined to believe it is written for this purpose only. Because the bill's deficiencies would force any peace-lover to oppose it as written.

5.8.2022 11:47am - (Replying to @stephensenn @hmatejx and @RWJE_BA) We have universal languages to describe differences and similarities among experiment. Am I right to assume that (1) is P(y|do(x)) and (2) is P(y|x) ???? The only way for me to find out if an assumption is reasonable is see evidence from researchers who understand the assumption.

5.8.2022 10:29am - (Replying to @TAH_Sci @MaartenvSmeden and @ManuelRMondal) I am not sure Chomsky's innate structure had domain-specific component.

5.8.2022 10:25am - (Replying to @stephensenn @hmatejx and @RWJE_BA) The effects of any experiment, good or bad, can be described in term of SCM, sometimes the DAG itself would do. What CI experts cannot do is guess what the experiment is that trialists describe in their jargon. I gave up.

5.8.2022 10:07am - (Replying to @stephensenn @hmatejx and @RWJE_BA) I can guess what the difference is between "randomisation" and "random sampling" but, since it is not defined in terms of Nature (eg counterfactuals) I can't be sure, and hate to be told again "this is NOT what trialists are doing". The burden of description is with trialists.

5.8.2022 4:56am - (Replying to @stephensenn @hmatejx and @RWJE_BA) Why can't we express what RCTs involve in the language of counterfactuals, so as to prevent any ambiguity or misunderstanding.??,

5.8.2022 4:50am - (Replying to @AdanZBecerra1 @_MiguelHernan and 3 others) It's page 117, sorry. It says: In the case where the node A has a parent, A has to “listen” to its parent before deciding on its own state. "Listening" is one metaphor, another is: "The sources of variations in A are its parents." A third invokes hypothetical interventions.

5.8.2022 4:36am - An incredible success story - 130 million students. I can't help but thinking: Perhaps that's the way of teaching causal inference?

5.8.2022 3:56am - (Replying to @AdanZBecerra1 @_MiguelHernan and 3 others) I will let high ranking professors take a first crack at it, but I would re-phrase your question to read: What the heck does the absence of an arrow mean?

5.8.2022 3:34am - I bumped into this post, which pokes fun of stat, DS, ML and AI, but I believe is wrong when it comes to AI, especially Causal Inference (CI). Curve fitting will never get you answers to reasoning problems that CI can solve nowadays. See 7 such problems:

5.8.2022 1:29am - (Replying to @stephensenn) Those applying CI to RCT should do two things: (1) Examine how RCT's SHOULD BE RUN given new data sources and new logic, thus far untried. (2) Learn how trials are run nowadays, once translated to an appropriate language, capable of handling multiple data sources and new logics.

5.7.2022 10:57pm - It is a good time to map our discussion into a constructive decision problem: What is the right thing for post-1990's statisticians to do these days? Defend the honor of pre-1990's statisticians? or rejoice, acquire, use and teach the new algebra of causation? I'm w/ the latter.

5.7.2022 10:49am - A goldmine for a PhD student in math-stat: How many of those assumptions that escaped math analysis in the 1990's can now be mathematized (using the language of causal inference) and rescued from the mercy of intuition?

5.7.2022 8:53pm - It's thousands of miles away, I know, but how far is it really from the unsafe abortions and prison terms that thousands of American women will soon be facing if things go the way our fundamentalists wish them to go.

5.7.2022 8:18pm - (Replying to @JadePinkSameera) 1990's statisticians were not exposed to "causal math". E.g., they could not say (in math): Z confounds X and Y. Instead, they said: Z is "related" to both X and Y, and got it wrong. See #BookOfWhy Ch. 4 how badly they erred. They used heavy math, but the wrong one @kareem_carr

5.7.2022 6:12pm - I'm not only listening, but also putting what I hear into math, ask questions about it and coming up with a solution (when exists), something I could not do without the math. Here's a solution to the "selection bias problem" you raised: No math No solution

5.7.2022 5:46pm - When you read this piece you will be constantly asking yourself: Where did Yasser Arafat and Mahmoud Abbas learn Russian? Aka @4noura , Omar Barghouti and Yahya Sinwar ?

5.7.2022 1:36pm - "We have to listen to Nature"- beautifully put! But if Natures speaks Swahili and we barely speak English, how can we listen to Nature? The Nature I know speaks cause-effect, the statisticians of 1990's (including Box) spoke probability. How did they expect to listen? Now we can.

5.7.2022 11:05am - Just imagine how grotesque @4noura would be, once we strip her of the "conflating" defense", stop accusing her of antisemitism, and charge her with an uglier form of racism - Zionophobia which she cannot deny.

5.7.2022 10:10am - (Replying to @TheAcsMan and @4noura) Zionophobes like @4noura claim that they love Jews, and that the violence we see is done by a mob who doesn't distinguish good Jews from bad ones -- Zionists, like me. To deflate her rhetoric we simply show that her hate of bad Jews is racism of a more despicable degree.

5.7.2022 3:02am - (Replying to @Ostrov_A) Beinart took my words and reversed the players. I've been arguing for years that the media covers surface breakouts of violence by one side or the other, but keeps silent about the systemic overriding violence contemplated by the Palestinians against Israel - eliminationism.

5.7.2022 12:30am - Read the fine print: He is speaking at "University of Jordan". Universities in this part of the world are not there to teach, but (1) to distort & malign and (2) to grant academic degrees, so that their graduates get hired by American universities and continue to distort & malign

5.6.2022 11:47pm - You must be reading yesterdays news, @4noura . Times they are changing. No one is conflating anymore, and no one would ever accuse you of antisemitism. You are accused of a much uglier form of racism, Zionophobia, of which you cannot exonerate yourself, for it is part of you.

5.6.2022 11:32pm - (Replying to @Ostrov_A) Did they say "keep us safe"? Which "us" do they have in mind?

5.6.2022 11:25pm - For readers not aware of the latest transition, Ken Roth is leaving his post as head of @hrw and, having imprinted his distinct mark on the history of Israel maligners, is available for new leadership position in "human right" outfit, where "human" stands for "only some humans"

5.6.2022 6:30pm - The Harvard Crimson Normalizes Growing Campus Antisemitism
5.6.2022 6:30pm - If we needed another evidence for the thoughtlessness, racism, and shady practices of Crimson's Editorial Board, here is one by Alan Dershowitz, a Harvard's faculty and author of "The case for Israel". Glad to help defeat Crimson's censorship by retweeting this article.

5.6.2022 5:38pm - I've tried to get the signatories names, but was told the list is still growing. This means that all Harvard faculty among our readers can still add their names. Don't miss! Think, how often do we scientists get a chance to take direct action against this ugly form of racism?

5.6.2022 1:24pm - I hope the C2E2 people are open to #precisionMadicine

5.6.2022 5:49am - (Replying to @ShMMor and @AP) You can't talk like that. "Belief" is the most protected commodity in the modern era. The cannibal's "belief" is as respect-deserving as that of its victims.

5.6.2022 5:31am - I'm fascinated by the joke that a "discrete prayer" of on person can "anger" another, and shake a modern Kingdom, and that the @ap writers did not think it deserves a funny remark, or at least a pause.

5.6.2022 3:43am - (Replying to @zaffama) Philosophers are drawn to statistics by the thesis that all knowledge comes from observations, therefore, if you wish to understand how we know what we know, you better study the science of observations - statistics. But the bulk of what we know is in the ropes, not in the data.

5.6.2022 3:32am - (Replying to @RWerpachowski) Was I wrong in this observation? I've got a few examples.

5.6.2022 1:58am - (Replying to @whysomanykaykay) Never too late.

5.6.2022 1:39am - I've always claimed that every statistician is a philosopher at heart. That's why they get so disappointed when shown that statistics alone is still not philosophy. They were so hopeful that it was.

5.5.2022 11:40pm - I am not sure it's wise to call them "sick, barbaric society", because it implies that up until seeing their celebration we thought they act as "healthy, humane and reasonable" society. If we thought that way, why couldn't Israel make peace with these humane and rational people?

5.5.2022 10:56pm - I consider myself quite knowledgeable in the History of Israel and, still, what I've learned from this 20-Questions quiz was intriguing and invaluable. The one that blew my mind was #2, about the Arch of Titus, Rome, November 30, 1947. Great Quiz!!!

5.5.2022 9:41pm - Listen to the rhythm of the chase, the flute, and Chava Alberstein as she sings: "Yes, she is fearful, but as if not concerned, Will wait for the end of the chase"

5.5.2022 9:17pm - The Chase via @jewishjournal
5.5.2022 9:17pm - I don't know a better way of ending Israel Independence Day while, simultaneously, watching the axe-murders in Elad, than singing MIRDAF with Chava Alberstein and reading my translation of "The Chase" (approved by Yaron London), below.

5.5.2022 9:01pm - When we ask: "Where's the outrage" we give our enemies the impression that we rely on world's opinion, and we don't know that the world can so easily be deceived by labeling Tel Aviv residents "Settlers". To expose the deception, let's just show the axe & say "We're here to stay"

5.5.2022 8:35pm - You should not feel helpless. See what 65 professors did at USC upon realizing the faculty own the university, not the Chancellor, and not the mob.

5.5.2022 8:28pm - (Replying to @SAFECUNY and @ChancellorCUNY) I agree with the fact but not with the argument. Suppose the call does not mean Jews, just Zionists, would that make it less racist? Suppose it was "Latinos out of CUNY" or "Muslims out of CUNY" would @CancellorCUNY remain quiet? Once we mention "Jews" we lost the game.

5.5.2022 7:06pm - Hats off to your Grandpa, who went to fight five armies and managed to stop them for at least 74 years. Too bad I was only 12 then -- they wouldn't take us.

5.5.2022 6:45pm - (1/2) I can't think of a more rewarding way of celebrating Israel's Independence Day than the one I had this morning, speaking to 250 children age 6-10, holding Israeli flags, and telling them about the bomb shelter where I celebrated Israel zero'th birthday, & what it meant to us.
5.5.2022 6:45pm - (2/2) Unfortunately, it also saddened me to realize that we, the 1948 generation, are becoming an endangered species, like holocaust survivors and, soon, the 1948-deniers will raise their heads and play their trumpets.

5.5.2022 6:28pm - Harvard Faculty Group Denounces Crimson Newspaper BDS Endorsement
5.5.2022 6:28pm - My apologies to Harvard's faculty for having shamed their silence. They came out strongly against the attempted BDS-Crimson's wedding: @sapinker @NatalieLKahn

5.4.2022 6:00pm - (Replying to @MemphisSmitty and @NatalieLKahn) What makes "religious sects" more protection-worthy than history-bonded nations seeking self-determination, violently denied by their neighbors?

5.4.2022 5:42pm - (Replying to @Nedra @NatalieLKahn and 3 others) There are more effective ways Harvard alumna can support students; from my vantage point they have been silent, considering their latent ability to influence campus climate.

5.4.2022 5:19pm - (Replying to @mdubowitz and @NatalieLKahn) I'll join you in nominating @NatalieLKahn to Algemeiner's list of Top 100 People Positively Influencing Jewish Life, 2022

5.4.2022 5:04pm - (Replying to @Hoodview @JonathanSGlass and 2 others) Merci on @m_orlee , the siren songs of social acceptance may corrupt the best of us.

5.4.2022 4:50pm - Not sure if you received my comment, so I'm re-replying, and wish to restate my pride in you and your oped, and my renewed hope in Harvard students ability to deposit the Crimson Board episode deep down the Harvard Bin of Shame. Happy Independence Day

5.4.2022 1:55pm - (Replying to @soboleffspaces) Right! Where Z is a set of characteristics measured for people in the population, but unknown for the individual him/herself.

5.4.2022 11:07am - Thank you Natalie! It's the best gift I could have imagined for Independence Day. You have removed some of the ugly stains Crimson's Board has cast on Harvard's name.

5.4.2022 10:41am - (Replying to @EinatWilf) Me too! My scholarly friends say: too theatrical, too populist, too extravagandish, too this and too that. This is one day I want to be theatrical, populist, this and that. Think about it - 74 years!!!

5.4.2022 4:06am - The significance of Theorem 5 would be greatly appreciated recalling the Z is not a vector of characteristics of the individual of interest, but of the population at large.

5.4.2022 2:20am - The siren sound lasts for 2 minutes. But this fast-forward video amplifies what it means when a nation honors its defenders and stops to contemplate the meaning of its being.

5.4.2022 1:32am - (1/2) As it's a day before Israeli Independence Day, I opened an email received from Quora: "What's the Best Thing Israel Has Done for the World" I agree w/ the answer, and would summarize it in one word: Israel has given us the idea of "auto-emancipation"
5.4.2022 1:32am - "Auto-Emancipation", a word coined by Pinsker in 1881 means: "You, as a collective are a master of your destination. Yes, take all the help that others offer, as long as it further your goal - to be in charge of your future. Compromises? Yes, but stay the course, and it's doable.

5.4.2022 12:22am - Please join us in a moment of silence.

5.4.2022 12:10am - What in the world is NOT a a flagrant attack on the holy sites of Muslims? Can we breath? Eat? Our after-meal blessing says: "He will walk us in sovereignty to our land". Is that "a flagrant attack on the holy sites of Muslims?" What's holy for people w/ no respect for others?

5.3.2022 11:56pm - I tried to add a comment. I couldn't.

5.3.2022 11:27pm - Powerful words. Especially: “antizionism as an ideology is rooted in rage” because it aims for “the negation of another people, a concept as alien to the modern discourse as white supremacy." Almost the definition of Zionophobia. I hope it becomes an uncontested speech standard.

5.3.2022 10:21pm - (Replying to @BernieSanders @ninaturner and 3 others) Hard to swallow, but @ninaturner lost because she is associated with you, Bernie, and you have designated Linda Sarsour and James Zobgy as your advisors on Israel. You are a liability, Bernie, unless you come out strongly against BDS, very strongly, before November.

5.3.2022 6:56pm - (Replying to @SAFECUNY) CUNY Chancellor is waiting for Jewish faculty to draft a condemnation statement that he can issue. Have they? Administrators are not very creative, and they are scared stiff of Palestine Legal - they need help.

5.3.2022 6:44pm - And while we are at P(harm) and P(benefit), we should take a look at the newly revised version of "Causes of Effects ..." which was just accepted at IJCAI 2022, and provides new ways of shrinking the bounds on P(harm) and P(benefit).

5.3.2022 5:17pm - This is a beautiful paper, recommended for every decision maker (who isn't?), which starts with the same counterfactual foundations as our P(benefit) paper, and expands into many aspects of behavioral decision theory, including issues of moral obligation.

5.3.2022 6:47am - (Replying to @gottfriedmath @artistexyz and 2 others) I believe this belongs to another class of paradoxes, concerning reversal of preferences. Simpson's involves mixing of two datasets, each conditioned on different value of a third variable.

5.3.2022 1:12am - (Replying to @Soccermatics and @databoydg) I havrn't forgotten. I'm reading with interest how CRT morphed from benign anti-Eugenic movement into a movement perceived to be advocating race-based quotas. I admit to be biased by observations of the conduct of EDI offices on US campuses.

5.2.2022 11:40pm - Well documented history of how the enlightened attempted to "censor the intolerant". I'm surprise though that the instrument of "opposing the intolerant" (by university administrators) was not evaluated. I believe it is effective and protective of free speech.

5.2.2022 9:58pm - Exclusive: Supreme Court has voted to overturn abortion rights, draft opinion shows via @politico
5.2.2022 9:58pm - Breaking News, if not "Devastating News". This country is heading towards a civil war.

5.2.2022 6:34pm - For some strange reason, the music of Hatikva in this small hall, around 8-year-old Noam, sounds more meaningful than the one played in the huge Olympic stadium in Tokyo, 2021. Watch.

5.2.2022 1:14pm - A great summary of reactions to Harvard @thecrimson . I must add one Talmudic advice to @m_orlee : Ein Lecha Adam ShEin Lo Shaah (Avot 4:3) (There is no person that does not have his hour) Translation: Even Ex-Jews may someday understand what price they paid for social acceptance.

5.2.2022 4:36am - (Replying to @lennybendavid @ProfDBernstein and @m_orlee) Ein Lecha Adam ShEin Lo Shaah (Avot 4:3) (There is no man that does not have his hour) Translation: Even yester-Jews may someday understand what price they had to pay for social acceptance.

5.2.2022 3:00am - (Replying to @cunninghamjeff and @MerkinMuffley5) Patience comes from the Mishna. Raban Gamliel used to say: "The shy never learns, and the short-tempered never teaches." Avot, 2:5. I am working on my shyness.

5.2.2022 1:55am - (Replying to @MarkusSchacher) Check this: Bongers, Stephan, Patrick Forré, Jonas Peters, and Joris M. Mooij (2021). “Foundations of Structural Causal Models with Cycles and Latent Variables”. In: Annals of Statistics 49.5, pp. 2885–2915.

5.2.2022 1:29am - For those who have not read @EinatWilf book "The War of Return" this would come as a surprise. Still, surprise is better than innocence, when it comes to the Palestinian-Israel conflict.

5.2.2022 1:03am - I had a bunch of references, going back to Herbert Simon. But until I manage to dig them out, I am retweeting your question. Some of our readers would surely be able to help.

5.2.2022 12:58am - (Replying to @AChofshi @JGreenblattADL and 2 others) Very good questions, @AChofshi . Can any Jewish organization overcome its ego and admit to be learning from others? Can Hillel? Do we have any historical precedence?

5.2.2022 12:40am - Lavrov: So what if Zelensky is Jewish, even Hitler ‘had Jewish blood’ via @timesofisrael
5.2.2022 12:40am - This guy, Lavrov, is a college graduate. He even studied physics. Ivy Leage, like Harvard. So what if Hitler did not have Jewish ancestry? So what if Zelensky is not a Nazi? The more unbelievable the lie, the more it is believed by those who want to.

5.1.2022 10:06pm - You must be kidding. Anti-Israel pple engaging? The thing they fear most is the truth about Israel, and would do everything to keep that truth from exposing their ignorance and moral deformity. Education is needed, agree, but only to the willing. Zionophobes are beyond learning.

5.1.2022 9:19pm - (Replying to @KGilkarov) They try to harass you, as long as they think you are ashamed of something. Once you show them that you are proud of that very something, they are scared of you, because they know you can shame them with facts, logic and, mostly, with pride.

5.1.2022 8:40pm - When we marched on May 1st parades, we carried both red flags and Blue-White flags, and we sang "Arise, workers around the world" together with "Hatikva". "Free Palestine" meant "British go out". Who would imagine calls for freedom turning into calls for annihilation.

5.1.2022 8:18pm - (1/2) This is an important point to keep in mind: The way we estimate PC in does NOT require knowing the causal model. It should thus be rejoiced by even the most dogmatic statisticians on earth. Unfortunately, dogmatic statisticians never learned
5.1.2022 8:18pm - causal analysis, so, the idea of looking at observational studies drives them crazy. As Max Planck said (paraphrasing): Personalized decision making will have to wait for next-generation statisticians and lawyers.

5.1.2022 7:34pm - What kind of a soul must a person have to be ashamed of his/her soul? Israel's Independence Day is coming soon (May 4) always reminding me how one of my Jewish colleagues declined to join the celebration, lest his academic peers would take it as a political statement. Can't wait

5.1.2022 5:41pm - (1/2) More on Crimson's Editorial. Legal action against boycotters of Israel has its moral basis. If you run a restaurant and prefer some suppliers over others - fine. But if you hang a sign saying: "Not serving Blacks or Zionists" - freedom of expression no longer applies.
5.1.2022 5:41pm - (2/ ) Restaurant licenses are issued for the purpose of providing equitable service to the city residents, and your "Not Serving" sign defeats that purpose. Crimson Editorial is a shameful "Not Serving" sign on a campus newspaper; Harvard will reject hanging it on any other program.

5.1.2022 10:54am - (Replying to @saramagliacane and @tetraduzione) That sounds like a Chassidic song. Now I know where my grandfather got his songs from.

5.1.2022 10:29am - Agree! The best performance ever! Happy May 1st, everyone, Bandiera Rosa, Bandiera Rosa, To the millions and millions of people who sang this song in hope of a better world, And to the millions who sing it today Evviva il Ukraina e la libertà Evviva il Ukraina e la libertà

5.1.2022 10:06am - (Replying to @algentilini) Yes, we learned it in Kindergarten and sang it on May 1st. Than it became popular due to the movie which is one of the best movies I've seen in my life. I havn't stopped hamming it since.

5.1.2022 9:40am - @m_orlee , I am genuinely curious: Did the Board consider the wording: "Jews, like Palestinians, deserve a state", instead of your current version: "deserve nothing but life, peace, and security," which, as you know, is taken from BDS poetry book "A World Without Israel." ?

5.1.2022 9:11am - To May-1st: Avanti popolo, bandiera rossa Alla riscossa, alla riscossa Avanti popolo, bandiera rossa Alla riscossa, trionferà Bandiera rossa la trionferà Bandiera rossa la trionferà Bandiera rossa la trionferà Evviva il comunismo e la libertà Evviva il comunismo e la libertà

5.1.2022 9:02am - (Replying to @AsraNomani) She has a point. I used the term CRT to characterize its advocates' agenda. I read yesterday that it started in the 1970's as benign anti-Eugenic movement and morphed gradually into what we see today: legislation of race-based criteria to rectify statistical inequalities.

5.1.2022 12:56am- It's May 1st - International Workers Day. In my Kindergarten we were given red flags, and we march around singing: Arise, wretched of the earth Arise, convicts of hunger We sang high; being spoon-fed 24 hrs a day We then switched to Avanti popolo, bandiera rossa Join me!

5.1.2022 12:15am - Breaking News: After a 20-hour manhunt, Israeli forces arrested the two Palestinians who murdered Golev. May he rest in peace. He had seven brothers and sisters.

4.30.2022 11:24pm - To be fair, Harvard's Crimson Board was quite ecumenical. They generously said: "Jewish people — like every people, including Palestinians — deserve nothing but life, peace, and security." What they were unable to say was: "life, peace and a state", sending BDS to the roof.

4.30.2022 9:56pm - "Support for Palestinian liberation is not antisemitic". Says Harvard's Crimson Board. Except when you abruptly end here, as if it hasn't occur to you that Jewish "liberation" - Israel- could use some support too, in view of how Palestinians describe their pending "liberation".

4.30.2022 5:45pm - (Replying to @yudapearl and @MadamePratolung) If you allow me to judge RCT by effectiveness, and if we count EDI as an arm of RCT, then, based on my UCLA experience of past 10 years, EDI has been more obstructive than helpful. I'll soon post an account of some of my encounters with EDI, a uniform pattern in many campuses.

4.30.2022 4:53pm - (Replying to @MadamePratolung) Truly appreciating your input. If CRT is just anti-Eugenics then I would like to join the movement and actively promote it. But how did CRT get the reputation of advocating racial classification in school admission? Is it real or just bad mouthing? Can I judge it by effectiveness

4.30.2022 4:42pm - (Replying to @ESMDcan123 and @maria__cuellar) Finally, we focus on a concrete claim where we differ. I claim that some observational data, if added to the RCT finding can prove that treatment causes "harm" to some individuals. You say: "Impossible". I refer you to our paper, top of p.4. Convinced?

4.30.2022 4:25pm - (Replying to @matloff and @maria__cuellar) Yes, I saw the TOC. Fienberg was the first statistician who understood the importance of individual-level attribution, and said: "Impossible! Impossible" when I presented it at CMU in 2008-9 (see Causality page 396).

4.30.2022 4:07pm - (Replying to @ESMDcan123 and @maria__cuellar) "That causality can be inferred at the level of individual" is more than a "statement", its a mathematical proof, based on step by step demonstration. Which of the steps is shaky in your opinion. Our paper does not mention "design"; can you translate to mathematics?

4.30.2022 2:25pm - (Replying to @ESMDcan123 and @maria__cuellar) I believe I was able to demonstrate on this educational channel that Twitter is good enough to summarize and exchange ideas and principles or, at the very least, point to a paper and say: "This statement relies on assumption A, which does not hold in situation S." Can we try?

4.30.2022 2:11pm - (Replying to @ESMDcan123 and @maria__cuellar) Can you cite one such 'error' that YOU yourself believe to be an 'error', w/o referring to other threads, gossips or hearsays? I am open. Let's hear.

4.30.2022 1:22pm - (Replying to @ESMDcan123 and @maria__cuellar) Which statement of deserves "doubt" in your opinion? Please note footnotes 6 and 7 which, in my opinion, diffuse the "doubts" expressed by those you cite.

4.30.2022 12:57pm - (Replying to @ESMDcan123 and @maria__cuellar) And you agree with their "doubts"?

4.30.2022 12:52pm - (Replying to @AsraNomani) When people tell me "please be informed", I open my mind as wide as I can and ask: "tell me more". But when it comes to CRT, all I get is "it's not what you think". So what is it? What is that "social construct embedded in legal systems" that necessitates race-based schooling.

4.30.2022 12:37pm - (1/5) I took a closer look at @Maria_Cuellar summary of the "Causes of Effect" debate, and I have the following comments on her Chapter. 1. Regardless of whether one is using Bayesian or frequentist interpretation, the probability of causation should be defined by PN,
4.30.2022 12:37pm - (2/5) Eq. (12.1) or (12.5), NOT by PC, Eq. (12.2), as proposed by Dawid, Faigman and Fienberg (DFF). Proof: The conditioning event Y_1 =1 does not ensure that the outcome Y=1 or the action A=1 actually took place. 2. The two expressions coincide only when exogeneity holds,
4.30.2022 12:37pm - (3/5) i.e., when the action was taken at random. In litigation cases, actions are presumed executed by choice. 3. PC_i, as defined in Eq. 12.14 is obtained from 12.15 by consistency, which both Pearl and DFF assume; no point creating the impression that PC_i is different from PN
4.30.2022 12:37pm - (4/5) (as in Table 12.5) 4. The subscript i in PC_i is likewise redundant, because PN is a property of an individual, not of subpopulation. 5. Pearl would NEVER accept that, "since the data used to estimate PC comes from a population, the closest one can get to finding PC
4.30.2022 12:37pm - (5/5) for an individual is by conditioning further by using the covariates X." God forbids, see Bottom line, there is no controversy; DFF proposed a wrong expression for probability of causation (eq. 12.2) and, once corrected by PN, we sail the same boat.

4.29.2022 10:32pm - (Replying to @databoydg) Good to hear familiar terms, and substantive comments. But most of the RCT literature I read is fixated on rectifying statistical inequalities (eg Virginia schools), not their root causes; have I missed any of the latter?

4.29.2022 10:12pm - (Replying to @ToddTheLinguist) There are many crucial differences, the main one being that Stark was a racist while (most) CRT champions are anti-racists. But the difference you bring up does not show; CRT advocates the use of new legal means to fight statistical, not legal inequalities.

4.29.2022 7:35pm - I have my own theory on the Harvard scandal: Faculty silence. The experience of USC teaches us that Zionophobic hostilities come to an end when 65 professors step forward and tell the administration: Enough is enough, this is our campus too. See

4.29.2022 6:05pm - (Replying to @soboleffspaces and @SimonsInstitute) It is based on our PNAS paper, which my @SimonsInstitute talk mentioned briefly. But Elias has gotten more results since. I honestly can't figure out how ML can survive without fusion.

4.29.2022 5:34pm - As a former editor of Crimson, please teach us what's going on behind its editorial curtains. How is it possible for one or two big-mouths to silence voices of reason, hijack the good name of Harvard and parrot populist BDS slogans like some dumb white-supremacists. How?

4.29.2022 5:11pm - Glad Elias will be presenting "data-fusion" at MIT on Monday. The problem of piecing together multiple datasets collected under different conditions is one that both machine learning and econometrics have found to be insurmountable, and for good reasons:

4.29.2022 4:51pm - (Replying to @AsraNomani @JasonMiyaresVA and 5 others) I don't mind being called anti-Semite, as long as they don't call me anti-Zionist -- the ugliest label in today's market. But why do you call those people "Democrats"? People who call Asians "white adjacent" are just racist, not Democrats.

4.29.2022 10:25am - (Replying to @ToddTheLinguist) Agree. There are better examples of how the excuse of rectifying under-representations was misused to justify discrimination.

4.29.2022 12:23am - Inspiring thread with which to end Holocaust Remembrance Day. May the spirit of these people live on in all of us, and may we remember that it is now our turn to pass on the lessons learned of the time.

4.28.2022 10:43pm - (Replying to @InvisibleGoril1) Thanks for posting Plank's but I am not sure it is the article I had in mind. I distinctly remember Plank complaining about Jewish over-representation in academia, caused by their tendency to hire their kinsmen. I think I have a copy in my office

4.28.2022 9:03pm - (Replying to @osazuwa and @ubiquity75) I am not accusing anyone of arguing in bad faith, but we need to consider the irreversible effects institutionalizing society with race-based lenses, no matter how pure the intentions.

4.28.2022 7:55pm - (Replying to @EladNehorai) I am waiting for "the point of CRT" so that we can examine together where the analogy holds and where it does not. My point was/is that institutionalizing society to classify people by racial identity has unintended negative long term effects that need not be ignored.

4.28.2022 4:44pm - (Replying to @MerkinMuffley5) No analogy is perfect. And concerns about Black people under-representation are mostly genuine and noble. Still, concerns about over indulgence in statistics of racial divisions and the unwanted destructive effects of such indulgence should not be ignored.

4.28.2022 4:11pm - (Replying to @plattner_lisa) It isn't parallel in all ways, but it has some tiny parallels, as I understand what CRT is. And if I don't, by all means teach me/us what the "first thing" is about CRT that makes it unparalleled in ALL WAYS, including attention to statistical "representativeness" by race.

4.28.2022 1:59am - (Replying to @RRKRahul96) I don't know what identiraianism is, but Stark pointed out that the presence of Jews in academia far exceeds their percentage in the population, so corrective action must be taken for the sake of "equity" and "balance".

4.28.2022 1:49am - (1/2) I could not resist a paper titled "Causes of Effects and Effect of Causes" which appeared on my screen, and turned out to be a chapter in a book, in memory of my friend Steve Feinberg: " The chapter, by @maria__cuellar , summarizes a long debate on
4.28.2022 1:49am - attribution in legal settings, which should have been settled by recent papers on personalized decision making: and I wouldn't rule out the possibility that lawyers would adopt our results before drug trialists, which is a pity.

4.28.2022 12:38pm - (Replying to @ubiquity75) The precise definition of "critical race theory" is up to those who coined the term. I used quotes to evoke similarities between Stark's excuse and arguments used today to correct for statistical "under representations" by adjusting admission or employment criteria based on race.

4.28.2022 12:19am - We must remember that the holocaust was preceded by a successful program of "critical race theory". In April, 1938, Nobel Prize winner J. Stark explained in the British weekly "Nature" why Jewish professors at German universities must be replaced by "under-represented" Germans.

4.27.2022 11:21pm - (1/3) For me, "never again" means taking a concrete action to make the lesson of the Holocaust live and relive for generations to come. The action I have in mind is to plead/pressure/convince all 80 Holocaust Museums in the US to dedicate pavilion/corner/exhibit to the process by
4.27.2022 11:21pm - (1/3) which Israel has arisen from the ashes to provide new life for the survivors. Please join me in sending such request to the Director of the Holocaust Museum next to you. You can add this quote: ""Christianity wouldn't have become a world religion had its narrative stopped at
4.27.2022 11:21pm - (3/3) the crucifixion rather than the resurrection. Holocaust museums, from which most Americans get their glimpse of Jewish history, allocate zero floor-space to Israel - the Jewish resurrection." They will get the message.

4.27.2022 9:19pm - I was thinking: How can I commemorate my grandparents who were trapped in Poland in 1939, and taken to Auschwitz in 1941? Then I got this video and decided: To sanctify their memory, I am retweeting to all our readers.

4.27.2022 8:46pm - I was thinking: How can I commemorate my grandparents who were trapped in Poland in 1939, and taken to Auschwitz in 1941? Then I got this video and decided: To sanctify their memory, I am retweeting to all our readers.

4.27.2022 7:09pm - It takes a day like Yom Hashoah to show us how little the world has changed in the past 13 years.

4.27.2022 6:55pm - I normally hate ceremonies. But this one glued me to the screen for more than an hour, and made me understand, somewhat more than before, the meaning of the word "Hatikva" (The Hope). No, I still don't fully understand it...Perhaps my grandchildren will, some day.

4.27.2022 3:37pm - (1/2) A very powerful column on Holocaust education. Part of the declining interest in the Holocaust narrative (putting aside Palestinian objections and denials) has been our own failure to frame it in its right historical context and its inspirational message. I once twitted:
4.27.2022 3:37pm - (2/2) "Christianity wouldn't have become a world religion had its narrative stopped at the crucifixion rather than the resurrection. Holocaust museums, from which most Americans get their glimpse of Jewish history, allocate zero floor-space to Israel - the Jewish resurrection."

4.27.2022 9:04am - Thanks for introducing me to this useful resource. But if there is no text accompanying the audio clips, its hard to search for a topic of interest.

4.27.2022 2:59am - The thought that I am on the advisory board of an organization such as @newzionists always fills my heart with joy that, despite my age, I belong to the future. It's a future where Jews are not asking for approval to be themselves. Please join me in supporting them.

4.27.2022 2:36am - @KenRoth always reminds me how, when I coined the term Zionophobia, friends tried to convince me that people like @KenRoth may be ignorant, perhaps willfully ignorant but, for heaven's sake, they are not pathological cases. Ken proves me right, again and again.

4.26.2022 11:41pm - Gee, I've never seen this cover before. Is it a new edition? A translation to Swedish? Whatever it is, I hope the spirit of the New Science shines from it.

4.26.2022 6:38pm - Apologizing for not joining the quantum festivities. I have been so deeply immersed in managing causality to classical physics, that I neglected advances in quantum computing. Will get there soon.

4.26.2022 4:38pm - (Replying to @plevy @luislamb and 4 others) Agree, there is much more. But at least for that tiny portion of AGI that hinges on causal relationships, we do have semantics, and we need not climb Mt. Olympus to search for DL semantics.

4.26.2022 2:52pm - (Replying to @luislamb @GaryMarcus and 4 others) So why don't we adopt the theory of meaning from causal inference, of which DL is but a subtask?

4.26.2022 2:48pm - (Replying to @BryceGruber) Yes. The damage is done. X number of people arrested + injured. Y number of anti-Israel demonstrations around the world. The "envoy" can repent and relax.

4.26.2022 2:03pm - (Replying to @JGreenblattADL @elonmusk and @Twitter) Any idea how Zionophobic extremists feel? Say Linda Sarsour, Nora Erakat or Jiji Hadid? They should, but never have been banned from Twitter; are they worried or rejoicing?

4.26.2022 1:32pm - UN’s Middle East envoy appears to back Israeli account on recent Temple Mount unrest via @timesofisrael
4.26.2022 1:32pm - It's hard for me to believe that a UN "envoy" would confirm Israel's account of what happened on Temple Mount during Ramadan. Self reflection? New eye glasses? Shame? Born again? Hard to tell. Expect a change of heart soon.

4.26.2022 12:46pm - Retweeting, to pay tribute to Achad Ha'Am (1856-1927), the intellectual lighthouse of Jewish Odessa, whose essays influenced many of my generation

4.26.2022 12:17pm - (Replying to @HuelleAlexander) Zionophobes (like Rutgers' Nora Erakat) don't need to realize that they are antisemitic; it's enough if they realize that they are morally deformed and socially despicable.

4.26.2022 12:09pm - (Replying to @GaryMarcus @IntuitMachine and 19 others) Speaking practically, I would say "has partial access to whatever it takes to imitate the game of consciousness"

4.26.2022 12:00pm - (Replying to @AthenaAI2 @GaryMarcus and 21 others) "Knowing" is a matter of degree, agree. If your nightlight can also distinguish "she hurt my feeling" from "she flattered me" then your Marketing guy is not way off using the word "knows". Great philosopher, your marketing guy.

4.26.2022 11:46am - (Replying to @GaryMarcus @IntuitMachine and 19 others) Most humans know quite a bit about their blueprints, for example, whenever you and I say the word "I", or when we distinguish "she hurt my feeling" from "she flattered me". Reminds me: "He was a great philosopher, Adam, when he first uttered the word 'I' " (Achad HaAm, 1890)

4.26.2022 10:55am - (Replying to @IntuitMachine @GaryMarcus and 19 others) Can we just summarize it by saying: "Emotions" are hardware reactions to certain happenings in one's software. The mystery is in understanding those "happenings," not in wiring the hardware.

4.26.2022 10:24am - (Replying to @IntuitMachine @GaryMarcus and 19 others) A machine cannot have a complete image of itself (halting problem) but it can have a blueprint of its software (see #Bookofwhy) and that blueprint may be sufficient to "feel" compassion towards other machines. Is it really "feeling"? You can't fake it unless you have it.

4.26.2022 2:44am - (1/2) If I were a professor at @RutgersU , I would stage a hunger strike in front of the President office, and ask him to explain to NJ tax payers whether @4noura Erakat's criminalization of the identity of thousands of Rutgers students and faculty ("bedfellows of Nazism") stands
4.26.2022 2:44am - (1/2) in compliance with the norms of academic discourse and standards of diversity and inclusion at Rutgers University. @rutgersalumni @RutgersHillel @RutgersChabad @TheAENetwork @JGreenblattADL @deborahlipstadt @bandlersbanter @DionJPierre @JNS_org

4.25.2022 11:53pm - (Replying to @MSweetwood and @HillelNeuer) Amazing chart.

4.25.2022 11:30pm - @4noura is not demonizing Zionism, she merely "takes issue w Zionism" and labels it "political bedfellow of apartheid". This behoove us to soften our stance toward Zioinophobia and re-label it "political bedfellow of genocide." No defamation, just taking issue.

4.25.2022 10:57pm - I should add that the need to make the distinction between type-1 and type-2 antisemitism clear and crisp is what compelled me to coin the term Zionophobia, further emphasizing the moral depravity and obsessive nature of type-2 antisemitism.

4.25.2022 9:13pm - I am always surprised to see people surprised upon hearing Palestinians repeat what they have been saying since 1920: "No room for coexistence." Is it because we alluded ourselves: "It's a small minority" or "They don't really mean it"? They meant it! And never had a minority!

4.25.2022 8:49pm - (1/2) In discussing anti-Zionism and antisemitism, the latter is used to mean two different things: (1) Tropes & accusations that antisemites used through the generations, and (2) That which is socially unacceptable or morally reprehensible b/c it denies Jews the dignity of normal
4.25.2022 8:49pm - (2/2) nationhood and equality in the family of nations. Anti-Zionism is antisemitism in the sense of (2), even if it isn't in the sense of (1). For example, JVP and Neturei Karta are antisemites of the second kind, even if they speak Hebrew or lay Tefilin twice a day.

4.25.2022 4:03pm - For lovers of History. Today is April 25, the day of the San Remo conference, when in 1920, the League of Nations has echoed the Balfour Declaration (1917) and resolved: "in favor of the establishment in Palestine of a national home for the Jewish people"

4.25.2022 3:24pm - ADL should urge Georgetown administration to distance the University from the content of El-Kurd rhetoric, and stress the inevitability of co-existence.

4.25.2022 2:59pm - (Replying to @luislamb @GaryMarcus and 4 others) Unfortunately the data do not "represent" reality, they merely shadow it, the same way that a bottle represents the wine inside it.

4.25.2022 12:48pm - It's clear, I agree. Yet clarity carries a far reaching implication: testability. Any proposal of a new architecture for ML which promises to usher it towards GAI can be given a simple test, on a 3-4 variable toy problem, before committing 1 Billion $ to a new institute.

4.25.2022 11:43am - Status quo on Temple Mount? Yes! Conditional on end of violence and end to demands. Glad @rosnersdomain agrees: Else, the idea of barring Jews from Temple Mount only begets more violence.

4.25.2022 11:22am - Hillarious! Ukrainian MP Ivanna @IKlympush did not realize that MSNBC is situated in Al-Jazeera's mis-information bubble, and that any mention of Israel would bar her from future interviews. To keep his audience in the right bubble, Mehdi Hassan should interview only Russians.

4.25.2022 3:54am - What a superb description of how the "community of the good" operates.

4.25.2022 3:20am - (Replying to @ShMMor and @minhaipanema) I am relieved too, I haven't seen such level of moderation since Azam Pasha (1947). But where can we take the evil/good test?

4.25.2022 1:32am - (Replying to @ericjdaza @EpiEllie and @LucyStats) Off-hand, I see no connection, because Granger "causality" is non-causal. But I remain open to surprises if you can explain, conceptually, the principles by which you get "person-specific" effects, as we have done here

4.25.2022 1:05am - Tal Fortgang on Substack
4.25.2022 1:05am - Part of the problem is your honoring these Zionophobes with the title "anti-Semites." The latter begs for protection or empathy, and invites endless debates on definitions and counter-definitions. The former positions yourself as a prosecutor of a morally despicable ideology.

4.24.2022 10:36pm - (1/2) Though I haven't taken any Foreign Diplomacy class in my life, I would have said it differently: "Israel is preserving and will continue to preserve the status quo on the Temple Mount. But, if you dont stop this Al-Aqsa tantrum, we will reevaluate our position and consider
4.24.2022 10:36pm - (2/2) joining the family of enlightened nations where prayer is free for all.

4.24.2022 9:27pm - Replying to @ItamarIsJewish and @benzgreenfield Yes, we knew about it as early as 2009 Ed Koch was infatuated with Danny's last words, though he never explained to us why. We didn't pressure him, we knew why...

4.24.2022 8:37pm - When you put Nationalism, Democracy and Globalism in the context of chimps and Sapiens, the future of mankind seems less frightening. Yuval Harari at his best: Don't miss.

4.24.2022 1:56pm - (Replying to @reliesdecine) Hi @reliesdecine , I remember saying it (or writing it) someplace, can you remind me where? Thanks

4.24.2022 9:48am - The #Pallywood production of inciteful material will continue unabated, like production in the porno and prostitution industries, as long as customers vie for its products (eg., @nytimes , @amnesty , @unhrcpr .) Any thoughts towards criminalizing customers?

4.24.2022 9:37am - I've first learned about the Armenian genocide fairly late, in 1946, from a Hebrew translation of Franz Werfel's book (1933) "The forty days of Musa Dagh,". Why did it take so long?

4.24.2022 1:44am - (Replying to @rrtorrubiano and @GaryMarcus) I cannot comment on this debate in the abstract, but if you show me a DL system climbing from rung-1 to rung-2 of the Ladder with no external information I will begin to believe in unicorns and will stop wondering why others don't insist on seeing such climbing.

4.24.2022 12:56am - (Replying to @JoLuMorales @FundacionBBVA and @XLSemanal) I hope they say good things about me.

4.23.2022 4:22pm - (Replying to @westurner and @seanmcarroll) Which proves that the human mind is not organized to capture the "deep down" but the "deep up", i.e., the deep understanding of the macroscopic world.

4.23.2022 4:05pm - After Mariupol, is Putin aiming to top the Assyrians? See:

4.23.2022 1:52am - Q. What do Israelis and Palestinians agree on? Ans. Both take seriously what the latter say they would to do, given the chance.

4.23.2022 1:43am - "I am innocent" stands no chance against "he is guilty." The latter has a narrative, a demon, and noble slogans; the former, just the facts.

4.23.2022 1:37am - "What are they throwing rocks at?" In the age-old struggle between facts and fiction "I am innocent" stands no chance against "he is guilty." The latter has a narrative, a demon, and noble slogans; the former, just the facts.

4.22.2022 11:06pm - (1/2) Fascinating talk! I believe @seanmcarroll 's distinction between Macroscopic and microscopic states holds the key to a puzzle I wasn't able to resolve: Can we deduce, from the equations of classical physics alone, that the rooster crow does not cause the sun to rise?
4.22.2022 11:06pm - Another related puzzle: Can we tell, from watching a movie of Brownian motion, whether we run the movie forward of backward? See Causality, Chap 2, p.58-59, and on reversing the Statistical Time.

4.22.2022 4:30pm - "Cycle of violence" is CNN's vocabulary. Let's not equate maximizers of violence with minimizers, lest we begin to think like CNN and lose the distinction.

4.22.2022 3:45pm - The typical left-wing professor, chanting the "Israelis acting the way they do" mantra, is not a Hitler Youth grad. Quite the contrary. The only feature the two professors share in common is the astonishment that Jews defend themselves as equals --unheard of in their Youth Camps.

4.21.2022 11:06pm - Too easy? How about turning it into a college entrance exam (say in critical race theory). Q. Name a country where seeing someone praying offends people sensibilities. Ans.= ???

4.21.2022 10:32pm - Ask yourself why anyone should get upset watching someone else dance, pray, or just breathe. Last time I passed by a Church, they invited me in to pray. I know only one country where praying defiles someone's sensibilities.

4.21.2022 9:06pm - (Replying to @subyroy @Valerie_Harper1 and @kenaviba) "Not to be silenced" of course, but to be ignored, like the Sathmar Rabbi who thinks Zionism sinfully hastens God's redemption, and like Bernie Sanders who thinks Jewish redemption lies in the evaporation of Israel. The road to redemption is paved with false prophets.

4.21.2022 7:48pm - (Replying to @kenaviba @subyroy and @Valerie_Harper1) Beautifully put! Relatedly, he also said: “Zionism is a homecoming to the Jewish fold even before it becomes a homecoming to the Jewish land.” See

4.21.2022 4:42pm - (Replying to @BachmannRudi and @ArjanDevanamPKS) Great! Much relieved! I was worried one of those "cancel culture" priests complained about him spoiling the youth.

4.21.2022 12:58pm - If you can start a petition, I'll do my best to spread the word.

4.21.2022 11:02am - Twitter says: "Learn more about our approach to enforcement, including potential consequences for violating these rules or attempting to circumvent enforcement, as well as how to appeal." But there is no link to "appeal", all they talk about are their "rules".

4.21.2022 9:46am - Please help #FreePaulHuenermund

4.21.2022 4:24am - (Replying to @MindfulEm @_ChrisAlbertyn and 12 others) It's hard for me to remain unbiased, but when I try really hard, I can't resist recommending it very highly.

4.21.2022 2:26am - Do I hear a voice of reason from the Middle East? In the middle of Ramadan? About Al-Aqsa ? Hasty me was willing to bet that the combination of all three is a mathematical impossibility.

4.21.2022 1:28am - (Replying to @tetraduzione @dagstuhl and 4 others) I'm wondering: Why "imprecise probabilities"?. Have we exhausted what we can do with ordinary probabilities? Or that some people are more comfortable conveying knowledge using imprecise probabilities?

4.20.2022 10:48pm - No, UCLA has not adopted the IHRA definition. For this to happen, the opinion of "scholars" would probably be summoned, only to give administrators an excuse to say: "Even Jews do not agree on what antisemitism is -- let's appoint a task-force". Translated: license for inaction.

4.20.2022 10:37pm - The Nazarian Center is actually friendly to Israel, which doesn't prevent its "scholars" from showing off how knit-picking "scholarly" they can be, when straight answers is what's needed.

4.20.2022 10:05pm - (1/2) A mini-victory to causes and effects - our paper "Causes of Effects: Learning individual responses from population data" was accepted to IJCAI 2022. While I have published numerous papers at IJCAI and elsewhere, the kind of resistance that reviewers
4.20.2022 10:05pm - (2/2) have expressed against this paper almost made me worry that the idea of extracting individual behavior from population data would have to wait another generation to be accepted. Fortunately, our rebuttal was successful, and the idea, I hope will soon become standard CI tool.

4.20.2022 8:43pm - (Replying to @NathanWurtzel @ADLMidwest and 2 others) You don't defend your buddies by bowing to the bullies, you only embolden the latter to intensify their attacks. Ask any school teacher.

4.20.2022 8:33pm - A beckoning for UCLA Israel Studies via @JNS_org
4.20.2022 8:33pm - Who decides whether anti-Zionism is or is not antisemitism? "Scholars" whose livelihood hinges on keeping such discussions going ad nauseum? Or students who are clinging to the only word that characterizes the Zionophobic attacks they must endure on campus? I side with students.

4.20.2022 7:13pm - (Replying to @KordingLab @shansiddiqi and 9 others) I thought you were interested in the history of causal thinking from Aristotle to our days. If so, #Bookofwhy provides, so I believe, all key chapters in that history. For an abbreviated version, see (lacking 21st Century developments)

4.20.2022 3:14pm - Princeton Student Government Will Not Make a Statement for or Against BDS-Aligned Referendum via @jewishjournal
4.20.2022 3:14pm - (1/2) And the Carnival goes on and on. Everyone knows that Princeton administration will not divest from Caterpillar. Everyone knows that referendums are choreographed to malign, not to be acted on. Still, everyone acts as though the outcome of this referendum matters. The only
4.20.2022 3:14pm - thing that matters now is whether Princeton's President launches his own referendum: "Is BDS a racist movement? - Yes or No" It's the only way to end the BDS carnival without taking sides.

4.20.2022 1:37pm - Agree. The text of @amanpour 's question is reasonable and professional. But her facial expressions unveil intense animosity which, I am sure, she herself wouldn't be proud to watch. Painful and disappointing.

4.20.2022 4:59am - (Replying to @MarkRuffalo) Not so fast, @MarkRuffalo ! How about a juicy "Blame Israel" slogan, to sweeten the holiday, and make you feel creative and super-virtuous?

4.20.2022 4:47am - (Replying to @blakeflayton) Yester-Jews believed they could cement a new people on lox and bagel. Some still do, I see their kids on campus, unable to understand, why the Zionophobes won't take their bagels, even wrapped in slogans.

4.20.2022 2:38am - I realize that the King feels he needs to project virtues to his constituents, showing that, for him too, certain principles are immutable. Fine. Now here comes a police trying its best to protect these very principles and, what does he do, he throws a tantrum like spoiled child.

4.20.2022 1:41am - A penetrating in-depth analysis, not merely of the validity of the #amnesty report on "apartheid", but also of the mentality of its authors. A must read for anyone who thought: But there must be some truth in what #amnesty is saying!

4.19.2022 11:56pm - These facts and photos are known to everybody. The question is why a Jordanian King who enjoys a reputation of a rational being acts as if he hasn't seen these photos, or hasn't followed the sequence of events. My pathologists colleagues are forgivingly silent.

4.19.2022 11:26pm - (Replying to @LahavHarkov) What kind of response has Lapid expected? A two-sided response? A rational response? An adult response?

4.19.2022 8:23pm - (Replying to @yudapearl @makoyass and @TomChivers) To give an example of a complex evolutionary process whose end product is easily emulatable, take Pavlovian response. The dog's response to a bell can be mimicked by computing P(food|bell), In contrast, a child's empathy towards a stranger cannot be mimicked by such simple trick.

4.19.2022 12:33pm - (Replying to @makoyass and @TomChivers) My puzzle is not the process by which it was developed, but the final product, i.e., how it is represented in the mind, and whether that representation can be emulated by machines.

4.19.2022 6:54am - (Replying to @robertskmiles @DJMinded and @TomChivers) Toddlers learn quickly that differences in appearance is no hindrance to empathy. Machines should do it too, especially after realizing how similar their cognitive structure is to that of their human mentor.

4.19.2022 6:25am - (Replying to @WimCeelen) We in AGI do not accept "innateness" as a hindrance to mechanization. We ask: how is it represented in the mind, and what does it take to emulate it.

4.19.2022 6:20am - (Replying to @daniel_corcos and @stephensenn) I do not use "risk factor" terminology. I use back-door, and no confusion arises.

4.19.2022 3:23am - (Replying to @TomChivers) S. Russell advocates an imitation game, i.e., infer values by observing adult behavior. I am puzzled by how evolution managed to preprogram the basics of human values, because I suspect it is encoded in a structure similar to that used by children for acquiring causal information

4.19.2022 2:02am - (Replying to @AdityaRajKaul) My email is,

4.19.2022 12:24am - I enjoy reading S. Russell's thoughts on the difficulty of teaching machines to respect human values when we can't specify them ourselves: Yet we do succeed in doing so (barring exceptions) when it comes to children's education. Is the analogy valid?

4.18.2022 9:07pm - ASM unanimously passes resolution addressing campus antisemitism
4.18.2022 9:07pm - Resolutions to address campus antisemitism are welcome but, as the USC experience has taught us such resolutions turn covers for inaction unless accompanied with explicit and specific statements that Israeli and Zionist students are welcome on campus.

4.18.2022 8:20pm - How the image of a besieged and victimized Russia came to be so ingrained in the country's psyche via @ConversationUS
4.18.2022 8:20pm - A must read, to understand why Putin still appeals to most Russians, and how the myth of "Fortress Russia" (the perception of Russia as an island of goodness and beneficence in a hostile world) will shape Putin's next moves.

4.18.2022 7:58pm - (Replying to @eliasbareinboim) The feeling is mutual.

4.18.2022 7:02pm - It is out now. A video recording of the 20th Daniel Pearl Memorial Lecture, by Benard-Henry Levy has been released, and can be watched here: General Introduction: 0-6:15 My introduction: 6:15 - 12:00 BHL Lecture: 12:00- 38:00 Interview+QA: 38:00 - end

4.18.2022 2:49pm - Inside the Princeton BDS Referendum Controversy via @jewishjournal
4.18.2022 2:49pm - Insightful summary of the Princeton referendum. Yet, as I've tweeted earlier the damage is done; referendums are designed to slander, not to be acted on. "Is BDS a racist movement?" should be on the ballot next.

4.18.2022 2:17pm - (Replying to @deboerk07 and @lucasbergkamp) It is answered partially here:

4.18.2022 10:52am - (Replying to @MaartenvSmeden) Fascinating poll and discussion. I think the "Crash Course in Good and Bad Controls" has the answers to many of the questions.

4.18.2022 10:36am - A very happy holiday to all our readers celebrating #Passover, #Ramadan and #Easter around the world. Chag Sameach!

4.18.2022 1:50am - (1/2) Congratulations to @eliasbareinboim for getting tenure at Columbia University. Simultaneously, congratulations go to Columbia University for a securing its leadership in next-generation AI. This advancement strengthens causal inference research with an important academic
4.18.2022 1:50am - (2/2) anchor at Columbia University, and should inspire other universities to assess shifting trends in data science. Congratulations!

4.18.2022 1:32am - (Replying to @aliya_amirova and @enemiesnet) I feel bad not realizing that the photo was not taken this year, but I feel elated knowing that many families like those shown in the photo were able nevertheless to worship this year at the Temple Mount, and that Israeli police has dared the impossible to make it possible.

4.18.2022 12:54am - (Replying to @enemiesnet) It does not negate the content of my RT.

4.17.2022 11:50pm - University of British Columbia Rejects BDS Movement, Urges ‘Respectful’ Discourse on Mideast
4.17.2022 11:50pm - (1/2) Hats off to all students and faculty of the University of British Columbia for rejecting BDS demands last week. Still, the fact that a student union could issue such demands for the purpose of slandering Israel as a "system of apartheid" calls into question
4.17.2022 11:50pm - (2/2) whether UBC is doing all it can to curb Zionophobic racism on campus.

4.17.2022 11:11pm - I feel like breaking fast with these families, and thanking Israeli police for maximizing the number of worshipers allowed on the Temple Mount, while minimizing the number of arrests needed to enable the former. A monumental achievement.

4.17.2022 2:35pm - @nytimes style headline: Three Muslims on their way to Ramadan prayer at Al Aqsa Mosque, the holiest Muslim shrine in all of Jerusalem, intercepted by three Jews going elsewhere. @nytimes

4.17.2022 12:13pm - This holiday statement by Tom Nides @USAmbIsrael must have been generated by an old software package, before the advent of AI.

4.15.2022 4:31am - (1/2) Dear readers, I'll be taken a couple of days off Twitter, to celebrate Passover with my family. In addition to symbolizing freedom, renewal and hope, Passover reminds us that humans are historically driven animals. As the First Commandment says: "I am your Lord your God, who
4.15.2022 4:31am - delivered you from Egypt, from the house of bondage." Modern translation: "If you forget history, you'll forget me -- your sense of right and wrong." Happy Passover. Next year in Jerusalem!

4.15.2022 4:12am - Decades of lop-sided press coverage of the conflict gave Palestinians the idea that Western readers are stupid enough to believe that Israel is interested in escalating tensions on the holidays, even "to perpetrate a massacre"-- no less. Western readers can swallow elephants.

4.15.2022 3:38am - Their song is ringing in my ears: "And our footsteps testify: We are here!" They were there!

4.14.2022 5:49pm - (Replying to @AdanZBecerra1) Thanks for making my day, but it wasn't really my idea. DAGs were created by the mind so that we can see the world better, or, putting it differently, so that we can "understand" the world better; "seeing" is for the birds, and for deep learning folks.

4.14.2022 4:38pm - When the sewage overflows, you need a plumber. University administrators are waiting for ADL to draft a statement that simultaneously: (1) respects freedom of speech and (2) puts an end to Zionophobic slander on campus. Every sewage has its plumber.

4.14.2022 1:51pm - (Replying to @holdmytowel @gottfriedmath and @dylanarmbruste3) But don't forget why I am only a half Bayesian

4.14.2022 1:47pm - After last week's attacks, we need an uplifting story to start the Passover holiday. We bless every photon on its way to protect Israel from its predators. Rumor: @amnesty is preparing a 400 page report to ban "apartheid photons".

4.14.2022 1:19pm - (Replying to @soboleffspaces and @federicarusso) In the 1970's, expert systems tried to model the decision maker. It didn't work well. Bayesian Networks were successful because they modeled the disease, not the physician.

4.14.2022 2:43am - I believe most epidemiologist are now aware of the "Table 2 Fallacy". The problem is with authors of regression analysis texts, the bedrocks of statistics education. They rarely stop to ask "is this a causal effect?" and they rarely admit they secretly wished it was.

4.14.2022 2:29am - Is the enormous budget of the EDI enterprise just a glaring waste? Or a major obstacle to Equity, Diversity and Inclusion on US campuses? This well-documented article takes a position. I concur, based on experience with EDI offices on three campuses.

4.14.2022 12:43am - (Replying to @soboleffspaces @JanPotters and 2 others) I agree with @soboleffspaces that we can learn a lot about the dynamics of paradigm shifts by studying the one unfolding before our very eyes, in causality. The anti-shift forces are quite different today, compared with the time of Planck; the echo-chambers are bigger & tighter.

4.14.2022 12:36am - (1/2) Good news, mixed with somber thoughts. It is a Pyrrhic victory, because, as I've written elsewhere, BDS referendums are designed to malign, not to be acted upon.The maligning has been accomplished, Israel was put on the accused bench, and the next referendum will repeat the
4.14.2022 12:36am - allegations, year after year. Glad a counter-panel was organized and placed BDS on the accused bench (the only kind of panels I agree to join). I wish Princeton students would start next academic year with a fresh referendum: "Is BDS a racist movement?"

4.13.2022 1:29pm - (Replying to @huttoncp) I believe even the most traditional regression analysts today would admit that, secretly, behind all the talk of "statistical significance" and other legitimate concerns, what the analyst really wishes the reg. coefficients to represent are the (tabooed) causal effects.

4.13.2022 5:34am - (Replying to @AsraNomani and @JasonMiyaresVA) Since Israel is part of Asia, I see myself organically connected to the fight of these parents against anti-Asian discrimination. I also love the variable "experience", which the bigots are now using to hide "race", as if no one is watching.

4.13.2022 5:16am - We can read it in Palestinian newspapers of 1920's as well as in the books of Columbia professor Rashid Khalidi, crushing the Zionist movement has always taken first priority over establishing a Palestinian state.

4.12.2022 11:29pm - First time I am hearing this detailed history of CNN (convolutional neural net). Illuminating, though quickly overwhelmed by acronyms, my weakest spot.

4.12.2022 9:57pm - Universities betrayal of donors' intentions, a phenomenon to which I have been a witness (at UCLA and neighboring campuses) for over 2 decades, has now reached national attention. Needed: An independent arbiter, great caution, and decency in hiring and appointments.

4.12.2022 1:08am - Strange, it feels like I begin to understand Spanish.

4.12.2022 12:57am - And I was likewise excited to speak to a community of causal inference researchers that grew to such size, and bloomed in such colorful mosaic of young new leaders who are actively exploring the second phase of the causal revolution. It might be even more exciting than the first.

4.12.2022 12:40am - The murder in Tel Aviv has inspired several poetic expressions of Palestinian mentality, this one tops them all.

4.11.2022 1:14pm - Excited to hear of your new appointment, @noatishby , and looking forward to assist you in coordinating the activities of the many new Zionist organizations that have emerged recently, and are working tirelessly toward re-invigorating the inspirational power of Israel.

4.11.2022 9:43am - (Replying to @RunnerVgn) Please help me relate to my Quora answer. I don't believe I've used the term "outdated subfield" .

4.10.2022 9:33pm - (Replying to @FriedrichHayek) In AI, the reality that needs to be discussed is everyday experience, like driving cars, flipping switches and changing light bulbs. Once we mechanize this reality, going way beyond statistics, we hope to attend to grand scientific discoveries. Hopeless to start with the latter.

4.10.2022 2:13pm - (Replying to @pedrohcgs) One would assume that the "credibility revolution" would change this educational misdirection. Why hasn't it? Ans. Because economists are still refusing the First Law of Causal Inference, leaving econ. education in want of foundation.

4.10.2022 1:53pm - Anti-Israel Celeb Blasted by Palestinians for Condemning Tel Aviv Terror Attack
4.10.2022 1:53pm - Poor GiGi Hadid, she is being pressured again to decide if Tel-Aviv is, or isn't stolen Arab land, when all she wanted to do is create a myth that a non-negligible fraction of Palestinians can stomach the word "co-existence."

4.10.2022 12:59pm - Every once in a while, a reader "likes" an old post of mine, and tempts me to re-read and check if my thinking hasn't changed. This Quora post, on econometrics, statistics and machine learning reaffirms my current thinking, with very little to add.

4.11.2022 9:13am - I strongly urge all college presidents attending the Hillel - @AJCGlobal summit on antisemitism to pay good attention to this video, and to understand that, unless they nip this forum of fascism in the bud, college campuses will turn into bloody
4.11.2022 9:13am - (2/2) battlegrounds of harassment and intimidation. So, please, not another useless "task force" on antisemitism, like the one convened at U. of Toronto Start fighting Zionophobia with the tools, budget and personnel you already have to fight Islamophobia.

4.10.2022 12:59pm - Every once in a while, a reader "likes" an old post of mine, and tempts me to re-read and check if my thinking hasn't changed. This Quora post, on econometrics, statistics and machine learning reaffirms my current thinking, with very little to add.

4.10.2022 7:22am - Not to be missed. @blakeflayton speaking about #antisemitism on college campuses. Opps, I almost fell into this trap again. No! its about #Zionophobia on college campuses, i.e., the pseudo-respectable side of antisemitism, the side that truly counts these days.

4.9.2022 7:08pm - (Replying to @RobertoCarlosF @achavito and @wsosaescudero) Stock and Watson's book was reviewed here However, unlike authors of the other books reviewed, Stock and Watson consulted with me on possible improvements, and I assume their revised editions are free of inaccuracies.

4.9.2022 6:59pm - (Replying to @achavito and @wsosaescudero) Is it available in English?

4.9.2022 6:51pm - I am not sure which verdict would be harsher on @KenRoth : "KR is an anti-Semite" or KR is a Zionophobe". The former evokes stories of unfortunate upbringing, broken family, etc. The latter are calculating eliminationists, who know the genocidal consequences of their agenda.

4.9.2022 6:23pm - (1/2) (Replying to @soboleffspaces) Interesting observation. Normally we associate imbalance between trial arms with identification. But here the imbalance should be part of the estimation. Why? Because RCT imbalance (e.g., that 8 our of the 10 RCT samples are females) is attributable to the finiteness of the
4.9.2022 6:29pm - (Replying to @yudapearl and @soboleffspaces) the sample, whereas identification deals strictly with asymptotic analysis, no finite-sample considerations. Indeed, on the average, weighing the sample result by 50% will do better than 80%.

4.9.2022 6:16pm - (Replying to @achavito and @wsosaescudero) I am not familiar with the book “Qué es y que no es la estadística”. But I haven't met a statistician that understood the causal aspects of the paradox (see except perhaps the late Dennise Lindley (1923-2013).

4.9.2022 3:58pm - (Replying to @soboleffspaces) Why not RCTs? Because they do not need to know where Z sits in the DAG. All they need, for their purpose, is to assure that Z correlates with Y and is not affected by X. Bingo, their aim will be fulfilled (precision) w/o compromising bias in ACE.

4.9.2022 12:48pm - (Replying to @soboleffspaces) The needed knowledge comes from P(z), namely our assessment of the percentage of z's in the target population, hopefully confirmed by a large survey. It's no different than the knowledge needed for adjustment in observational studies. Where lies the problem?

4.9.2022 9:40am - (Replying to @soboleffspaces) I do not know what FDA aims to achieve by "adjustment for prognostic baseline covariates," because they do not say formally. I assume the purpose is to improve precision, not bias. I do not see why we are on square 1. I think we are on square 64 -- all is clear.

4.9.2022 9:04am - New acrobatics! @KenRoth quotes Zionophobic views of JVP ex-Jews to exonerate himself from charges of antisemitism. You are forgiven, Ken. History will judge you by your actions, their consequences, and the logic of your slander, not by labels you are trying so hard to dodge.

4.9.2022 8:28am - No, Secretary Blinken, Palestinian Terror is Not “Senseless” via @jewishjournal
4.9.2022 8:28am - (1/2) @SecBlinken , I fully agree with Suissa's observation that "The fundamental problem with characterizing terror as senseless is that it lets you off the hook". You know the "hook" as well as we do. The "hook" is Mahmoud Abbas, his TV anchors, his mosques, his school teachers,
4.9.2022 8:28am - (2/2) and their century-old eliminationist theology that you let off the hook when you hide behind the words "senseless" or "cowardice" or "not helping peace", etc, as if you never heard about the Middle East before. You shouldn't let them think you are as naïve as John Kerry was.

4.9.2022 7:09am - (Replying to @artistexyz @PetarV_93 and 3 others) I don't see any relation, because I don's see "reality" playing a role in this mathematics, nor do I see a "research question" that differs from standard classification, i.e., Rung-1.

4.9.2022 6:57am - (1/3) I like @TheAENetwork letter to Chancellor @MSuarezOrozco , in that it leverages voices of home-grown professors of @UMassAmherst . But I wouldn't use the virtue-projecting phrase "legitimate criticism of Israeli policies" which echo's BDS's propaganda that, for criticism to be
4.9.2022 6:57am - (2/3) "legitimate" it must be directed against Israel. Instead, I would use "legitimate criticism of Israeli and Palestinian policies," to remind readers of the real obstacle for peace. Additionally, restoring "Sabra" to dining halls would not prevent BDS cronies from
4.9.2022 6:57am - (3/3) repeating their campaign against another Israeli product, unless the Chancellor condemns BDS tactics explicitly. Recall, BDS campaigns are not aimed to boycott, but to bully. (I couldn't find a link to the Chancellor's statement.)

4.9.2022 6:16am - I'm retweeting because many readers are likely to ask the same question. The "bad and good controls" analyzed in our crash course refer to controls in observational studies. Controls in experimental studies need not worry about bias, except when Z is affected by X.

4.9.2022 4:29am - (Replying to @soboleffspaces and @analisereal) Good question! The answer is NO: In obs. studies, adjusting for Z won't give you ACE. But if you have an RCT, then averaging over the z-specific causal effects would give you ACE. Proof: E(Y|do(x)) = SUM_z E(Y|do(x),z)P(z|do(x))= SUM_z E(Y|do(x),z)P(z) if X doesn't affect Z.

4.8.2022 11:07pm - Congratulations! Readers are probably wondering why this paper will not appear in Econometrica where it belongs, given Angrist's quote on page 2. Ans. It is written for readers who are either familiar with causal modeling or are curious to quickly learn it from the Appendix.

4.8.2022 3:17pm - Father of Three Mourned as Toll in Tel Aviv Terror Attack Rises to 3
4.8.2022 3:17pm - Death toll of Tel Aviv attack rose to three. Barak Lufan died of his wounds. HYD, Hashem Yinkom Dammo.

4.8.2022 12:32pm - (Replying to @SeanYoungPhD @Scholars4Peace and 3 others) Adding grass-root professors to speak at the summit, plus, partnering with @Scholars4Peace and @TheAENetwork + to keep the "task forces" in focus.

4.8.2022 11:32am - Glad to see graphical causal models illuminating quasi-experimental thinking in ecological research despite theological prohibitions imposed by economists (e.g.,, the former champions of quasi-experiments.

4.8.2022 1:20am - (Replying to @ki_endoepi @oacarah and 2 others) Did you find clinicians able to interpret the path specific effect in situations where it differs from the total effect delivered by an RCT?

4.8.2022 1:16am - Let's send this query to the Mayor in the city of our residence. There are many decent Mayors in the US, they need a push.

4.8.2022 12:24am - (Replying to @ki_endoepi @oacarah and 2 others) what about the DAG in 8/. Is P(y|do(x)) identifiable?

4.7.2022 8:28pm - If chosen by @GEICO , I promise to speak an indigenous MENA language, sing indigenous MENA songs of strictly MENA holidays, narrate stories of indigenous MENA heroes of indigenous MENA history and landscape. No imitations! Promise!

4.7.2022 7:34pm - (Replying to @soboleffspaces) I am often surprised to see that people who are reluctant to distinguish Rung-2 from Rung-3 are precisely those who claim that what you can't get from RCT is non-scientific.

4.7.2022 7:25pm - Terrorist attacks are merely snapshots of the real killer - Palestinian ideology that Tel Aviv is stolen land. @SecBlinken should address the latter instead of decrying the former.

4.7.2022 7:13pm - In times like that, it is appropriate I believe to list the popular bands and pop stars and social media influencers, whose silence emboldens the next terror attack. I'm not into pop stars, so I would start my list with @amnesty , @HRW , Linda Sarsour, Rashida Tlaib and Ilhan Omar.

4.7.2022 5:15pm - Glad to hear that Geico cancelled Linda Sarsour appearance in their MENA Heritage Month. I'm still open though to invitations to represent Israel in their celebration. After all, how many countries do you know where toddlers speak a language spoken in the MENA 2000 yrs ago?

4.7.2022 2:46am - (Replying to @EinatWilf) Do you have a direct link?

4.7.2022 2:38am - To all lovers of counterfactual reasoning, this Passover song, Dayenu, has 14 nested counterfactuals in it, and it ends with one big thanks-giving praise: How grateful we should be that all those possible worlds have NOT materialized, and we can enjoy our actual world. Amen!

4.7.2022 2:02am - (1/2) Wait a minute, isn't Israel part of ME? Shouldn't a month celebrating ME heritage provide some glimpse at the culture of Israel? I know a couple of love songs written 2,000 yrs ago in the ME, in the language spoken then, hundreds of yrs before Linda concocted her hate songs.
4.7.2022 2:02am - I'm volunteering to represent Israel in Geico celebration, if they promise to let me face sister Linda, eye to eye, in song and dance.

4.7.2022 1:27am - Amazing phenomenon! I look at these soldiers and I feel brotherly kinship. Why? We are not the same age, not of the same religion or ethnicity. What is it? I know! I used to wear the same uniform in my army days, 70 yrs ago. I wish I could join their table and sing: Hallelujah!

4.7.2022 12:23am - (Replying to @ki_endoepi @oacarah and 2 others) Naive question: The model in your 11/ is not do-calculus-identifiabe. So how do you reconcile your identification results with the completeness of do-calculus?

4.6.2022 11:48pm - College and university presidents will gather in New York City this month for a summit on anti-Semitism on campuses run by Hillel International and the American Jewish Committee in collaboration with the American Council on Education.
4.6.2022 11:48pm - (1/3) I have two reasons to doubt the outcome of this summit. (1) Neither Hillel nor AJC have a "Faculty Advisory Board," skilled in negotiating with university administrators. (2) The meeting is billed as "summit on anti-Semitism" yet "Israel" and "Zionism" are not mentioned.
4.6.2022 11:48pm - (2/3) I am concerned that this summit will end up like many before it: Everyone swears to fighting antisemitism, everyone appoints a "task force" to pursue this noble cause. The task forces meet with Jewish philosophers and scholars. Five years from now, task forces conclude that:
4.6.2022 11:48pm - "Even Jews can't agree on what "antisemitism" is, so, let's continue to trumpet our commitment to fighting "anti-Semitism"." Organizations such as @Scholars4Peace and @TheAENetwork should help focus this summit on the actionable problem: Zionophobia. @HillelIntl @AJCGlobal

4.6.2022 10:57pm - (Replying to @ki_endoepi @oacarah and 2 others) Nice generalizations!!

4.6.2022 10:35pm - (Replying to @soboleffspaces) They will defend themselves saying that knowledge of which variable is affected by treatment is so mild and so reliable (even testable in RCTs) so it constitutes no threat, and should be welcome into the EBM community.

4.6.2022 10:11pm - (Replying to @soboleffspaces) EBM fears comes from the fact that adjustment methods in observational studies require knowledge of the DAG structure. In contrast, the adjustment recommended by FDA requires only that the predictors be non-descendants of the treatment, nothing else. I wish FDA spoke English.

4.6.2022 10:46pm - (Replying to @JudeanPF and @jjz1600) I never accuse Zionophobes of "lying". These poor souls can't tell facts from fibs. Their optics only permits them to distinguish stories that make Israel looks bad from those that make Israel look somewhat human. They call the former "facts", and dance to the light of the moon.

4.6.2022 9:54pm - James Zogby @jjz1600 can't reconcile to the fact that "separation" based on likelihood of violence and past behavior is not "apartheid" but first principle of law enforcement. Airlines act the same, when a passenger starts cursing the crew, let alone assaulting other passengers.

4.6.2022 9:39pm - I should add here that "bias amplification" does not threaten RCT's, because randomization removes the bias that gets amplified. It threatens IV though, where Hirano & Imbens (2001) (and perhaps other graph-avoiding economists) have fumbled. See

4.6.2022 8:54pm - Palestinian Apartheid Week as new approach to facing anti-Israel activity on campus via @JNS_org
4.6.2022 8:54pm - Finally, an anti-racism counter-attack, but this time based on truth, facts, and moral fortitude. I am proud of SSI for launching it nationwide. Please come to UCLA.

4.6.2022 7:37pm - I join @DeborahProudJew in urging you to email Duke University President Price to take action. I especially endorse bullet (3): Work with Zionist students on campus, not with Jewish organizations on whose watch Duke has turned a symbol of academic Zionophobia and tantric racism.

4.6.2022 6:57pm - The adjustment recommend by the FDA is valid, if done with an RCT, and if by "baseline" they exclude descendants of Treatment. This validity does not rest on any other modeling assumptions such as no-confounding, or no colliders, which adjustment for confounders does.

4.6.2022 1:34pm - (Replying to @soboleffspaces and @analisereal) Where do you see the "use the knowledge about outcome predictors for covariate adjustment after RCTs"

4.6.2022 2:09am - Princeton students to vote on anti-Israel referendum
4.6.2022 2:09am - (1/2) The problem with such "referendums" is that they are not designed to be implemented, but to malign. Jewish students at Princeton should respond with a "counter referendum" urging the university to condemn BDS for its racist character. Of course it won't pass, nor will this
4.6.2022 2:09am - (2/2) one. BDS "referendums" are not designed to pass, or to be acted on; they are proposed to gain publicity, a stage, a megaphone, and to point a finger at some imaginary accused, so as to bully pro-coexistence students into silence. Only a counter-referendum will stop this zoo.

4.6.2022 1:19am - Last month, I ventured to speak to the "Society for Health Economics and Outcome Research" -ISPOR, to learn more about EBM, RCT, etc. A summary of the interview is now posted here:

4.5.2022 9:09pm - After the war, Ukraine 'will definitely become a big Israel,' Zelensky says | World Israel News
4.5.2022 9:09pm - Someone should warn Zelensky that he underestimates Western mentality. 75 years from now, @amnesty will blame Ukraine for "apartheid", and #HRW will investigate him for "war crimes."

4.5.2022 8:37pm - My reasons for comparing the two perspectives was not to show economists how much they miss by being insular but, rather, to show economics students the vast research territories their field can acquire by taking down its confining walls.

4.5.2022 8:04pm - (1/2) Well put! Though, to pacify meta-philosophers who gets irritated by the term "actual world", I would add another layer: "model of the actual world," sitting between the data and the "actual world". Selection bias is an ideal pedagogical tool to stress the difference between
4.5.2022 8:04pm - (2/2) data and the "actual world". In the latter, two coins are independent. But if we exclude from our data all cases where the two coins come up "Head", then suddenly data show the two coins correlated. We live in a selection-biased universe, trying to interrogate the "actual world."

4.5.2022 7:36pm - I'll be speaking at the Simon Institute tomorrow morning and will try to give my talk a more theoretical-epistemological flavor.

4.5.2022 10:32am - A striking difference of perspective. Scholkopf's paper devotes 1/2 page to Instrumental variables (IV). Yet according to some economists, e.g.,, IV methods "transformed the modern practice of applied microeconomics". A huge "selection bias."

4.5.2022 3:42am - I'm compelled to Retweet this superb video by Rabbi Wolpe on "Why Antisemitism Makes ‘Sense’: Reflections on the Psychology of the Antisemite" I'm writing a piece on the psychology of the Zionophobe, so this video is a real treasure.

4.4.2022 11:21pm - One of the most comprehensive papers I've seen on causal analysis. Concluding: "Some of the hard open problems of machine learning and AI are intrinsically related to causality, and progress may require advances in our understanding of how to model and infer causality from data"

4.4.2022 11:16pm - (Replying to @svendaj and @kareem_carr) But those concepts can be either about the distribution that governs the data or about the ropes that push and pull that distribution. Statistics deal with the former. The "world" include the ropes.

4.4.2022 10:29pm - The "polite and cruel zoo" of academia was downed on me the morning after UCLA's rhinoceros festival, here I've recovered since, and learned to stand up to rhinos.

4.4.2022 7:45pm - (1/3) I think I'm getting it, but here is how I would explain it to my students: You want to test the assumption that (1) there are no more hidden confounders. If you can measure additional auxiliary variables Z, and you are willing to make assumptions about (2) how the Z's are
4.4.2022 7:45pm - (2/3) related (causally) to other variables, put those Z variables in your DAG, according to assumptions (2), then look at the DAG and see if it has any testable implications. If so, you have tested (1), on the basis of (2). Thus far, I haven't mention IV. But I can add that some
4.4.2022 7:45pm - (3/3) economists prefer (for cultural reasons) to cast this rather general test scheme in IV language, possibly because their students are not yet versed in graphical models, and/or because IV's are deemed common in empirical applications. Have I explained your paper correctly?

4.4.2022 6:51pm - Thanks for an illuminating interview. Now I understand where you got your calling - Corbynism. I got mine in academia - a more polite and cruel zoo - so when I spoke to ACF, I had a different message to students and alumni:

4.4.2022 5:03pm - (Replying to @kareem_carr) Curious: Any recent incident? A paper? A vibe?

4.4.2022 12:39pm - (Replying to @CausalHuber) But if you have (2) why bother with IV? Can't we get CATE without an IV? Another question: is your test a d-separation type test, or goes beyond it? (eg, Verma constraint?)

4.4.2022 11:21am - Remember our "Personalized Decision Making" We passed it by a trialist who also knows causal inference (rare species) and his revisions now make the paper palatable to all (I hope) who are curious about personalized medicine and improved RCTs. Enjoy.

4.4.2022 10:41am - Very interesting. Unfortunately, the notation makes it hard for me to see if it is related to the instrumental inequality ( Causality, p. 274) which is a testable condition for the classical IV structure. Conditioned on X we have a classical IV structure.

4.4.2022 2:29am - Palestinian Muslim Scholar Claims Jews Started Ukraine War to Establish a Second Jewish State
4.4.2022 2:29am - Nassar's imagination is topped only by @amnesty . But Westerners much prefer the latter.

4.4.2022 2:12am - Last time I stormed Damascus Gate was in 2014. I storm it each time I visit Israel, and I can't wait for my next storming, perhaps in June. Wish me a meaningful storming.

4.4.2022 1:48am - The "credibility gap" is moving to the UN. Who would you believe? Prof. Steinberg, or the @UN_HRC ?

4.3.2022 11:20pm - (1/2) I do not understand the logic of this debate. Israel claims Francesca Albanese is biased. Francesca claims she is not. As a student of "bias", I can resolve it in one question: "Dear Francesca, tell us something about Israel, her birth, existence, accomplishments, etc." I bet
4.3.2022 11:20pm - (2/2) I bet even 3-yr old would be able to tell if she is, or is not biased. After all, we now have a solid theory of bias -- no one can cheat!

4.3.2022 10:21pm - (Replying to @dylanarmbruste3) I can think of at least 3 books missing from the collection.

4.3.2022 10:01pm - (Replying to @SaraHirschhorn1) Thanks for your comradeship, Sara. But when I hear you mentioning "peace" in the same breath with lady Fatima, I am crushed by realizing how far away a light year can be.

4.3.2022 9:50pm - (Replying to @siriniok) Relax. Zionophobes do not mean what they say.

4.3.2022 9:47pm - (Replying to @PercySotoBecer1 @instrumenthull and 3 others) I didn't realize there is a debate about Medelian Randomization.

4.3.2022 8:00pm - (Replying to @siriniok) Zionophobes' knowledge has no effect on their shouting.

4.3.2022 6:14pm - (Replying to @skyniels and @BruinsForIsrael) For students, I would recommend SSI (Students Supporting Israel) and, for faculty, Academic Engagement Network (AEN) or @Scholars4Peace . Your use of the term "fascists" has sparked an alarm of the kind of tactics we are facing, and the kind of complacency we are seeing.

4.3.2022 5:23pm - Even @MishelanuIAC is with us. Let's get the Nazarian Center @israelstudies to spearhead our campaign by writing to Block, Drake and the Regents that no Center can function in an environment that de-legitimizes its very existence. @SSIUCLA @BruinsForIsrael

4.3.2022 4:53pm - Extreme pleasure to be retweeted by SSI UCLA, my own campus. All we need now is to get @BruinsForIsrael behind us and organize to convince Chancellor Block that SJP has no place on our campus, unless they declare preparedness to treat other student groups with equal respect.

4.3.2022 2:42pm - (Replying to @stephensenn @venkmurthy and 2 others) Every Instrumental Variable is predictive of outcome. Yet if we treat it like other covariates and adjust for it (as did Hirano and Imbens 2001) we will improve power, OK, but will also amplify bias, as explained here:

4.3.2022 9:54am - (Replying to @ale_martinello @HenningStrandin and @ProfLaurenRoss) Before "listening" I used "wiggle". But then people start complaining: what about non-manipulable variables that you cannot "wiggle". Listening requires no manipulability.

4.3.2022 9:44am - (Replying to @soboleffspaces Almost. Listening is focused, reacting may not be. I react to bad news, but I listen to the TV only. Formally X--->Z---->Y Y may react to changes in X, but it listens to Z only.

4.3.2022 8:27am - Most scientists think that anthropomorphic analogies are non-scientific, a mistake which explains why the question "what's a cause" got stuck in circular traps, and why science has not given us a language for causation. "Listening" has given it to us, together with logic & tools.

4.3.2022 7:00am - (Replying to @analisereal and @soboleffspaces) The discussion revolved around Z being an ancestor of both treatment and outcome, as in our model 10, or even model 12, because only correlation was considered, not ancestry.

4.2.2022 11:07pm - You are right, if any group would call for cancelling Muslim professors, for any reason, it would be ousted in a day. But this is not the time to complain why our university administrators tolerate the intolerable. It's time to present them actions they can take - disqualify SJP.

4.2.2022 5:25pm - (Replying to @soboleffspaces) I am as embarrassed, but I've learned to expect the weirdest things from people who are speaking RCT without DAGs to guide them. Imagine how confused the world would be like through the lens of "conditional ignorability." This is what some folks are forced to work under.

4.2.2022 4:56pm - Agree. Hey, who pressed this "like"? How timely. The harmful effects of "harmless" are beginning to erode our RCT's and we would soon take drugs approved by DAG-less folks.

4.2.2022 4:31pm - (1/2) Dear Fatima Mohammed, I am one of those Zionist professors whom you wish to shun. Let me assure you that your organization, SJP, will be ousted from my campus way before me or any of my students, precisely because of racist demands such as those you have made today.
4.2.2022 4:31pm - (2/2) Academia cannot tolerate student groups that call for the ethnic cleansing of other student groups. SJP, as well as other Zionophobic or Islamophobic groups, have no place on any university campus.

4.2.2022 3:53pm - (1/2) Agree: "It does not make sense to select *additional* variables" -- even if they are correlated with y. This is the "bias amplifying" mistake made by IV folks, and corrected here: and here: Surprised to see it nibbling RCTs.
4.2.2022 3:53pm - (2/2) This is serious, because RCT folks are more vulnerable to such mistakes, not having any training in graphical methods, and the ML folks that guide them suffer from similar training deficiency. Hopefully, our crash course is read before bias amplifying is enshrined.

4.2.2022 3:08pm - (Replying to @JacobusCilliers and @dmckenzie001) I can't find links to the relevant papers, but two questions would guide me when I do: (1) Do they specify the causal estimand that their method is aimed to estimate? (2) Have they taken our "crash course in good and bad controls"? @VC31415

4.2.2022 8:39am - My recollections of the 2005 Doha Conference were published in the Jerusalem Post (June 16, 2005) Note how different (and similar) the world was 17 years ago. The disappearance of Israel was on everyone's mind, though not in anyone's words.

4.2.2022 7:21am - (Replying to @dylanarmbruste3 @gottfriedmath and 5 others) Tiny correction: Causal inference users of observational studies rely on TWO inputs: (1) Data, (2) Model assumptions.

4.2.2022 7:17am - (Replying to @RJsnda and @DavidDeutschOxf) I don't see how a concept (EBM) can be "false" if it is not "falsifiable". My reading is that EBM is merely a list of hopeful desiderata, eg. "lets use all the evidence we have". How can anyone falsify such noble aspiration if we are not told how to combine pieces of evidence?

4.1.2022 11:46pm - (1/2) I don't know how one can be a Zionist and not be proud. When someone asks me if I am a Zionist, I find it hard to contain my busting joy knowing that I am part of that intriguing journey called Jewish history and that astounding miracle called Israel. I have an anecdote
4.1.2022 11:46pm - that goes with this pride. When I visited Doha (2005 conference on East-West relations), I was interviewed by an Al-Jazeera reporter who asked: "Are you a Zionist?" And I answered, tongue in cheek: "Of course, isn't everybody?" Needless to say, the interview was not published.

4.1.2022 11:24pm - (Replying to @soboleffspaces) I do not see any "misconception" fueling these issues, in the sense that I haven't seen a question about experiments and observation that we cannot answer today, based on what we know from causal analysis. Have you?

4.1.2022 2:24pm - (Replying to @plusqperfect @ducha_aiki and @wellingmax) No. You were off, you haven't heard correctly. Zionophobes often lapse into mishearing or mis-thinking. Understandable, no problem,

4.1.2022 11:47am - I assume this post is addressed to @amnesty and @hrw ? Or is it to @NewIsraelFund and @PeaceNowUS ? No, it must be to all UCLA departments that howled "solidarity" in May, and turned mute in March.

4.1.2022 9:03am - (Replying to @viejojudio and @situraku666) Asking Zionophobes for the facts is like asking an alcoholic for the chemistry of alcohol. It is unfair, and could be traumatic.

4.1.2022 8:04am - (Replying to @soboleffspaces and @gottfriedmath) The inequality is not in the definition, but the equality P(x|do(y)) = P(x) is a corollary of X-->Y and the definition.

4.1.2022 8:23am - This Tweet has an anecdotal continuation. Such conversation happened to me twice in public lectures and, after a long pin-drop silence, I said: "I'll tell you what, state it loud w/o meaning it, just move your lips, we won't quote you .." -- Pindrop silence, no lips moving, Nada.

4.1.2022 7:27am - (Replying to @eoteromuras @gilbellosta and @Picanumeros) My Spanish did not allow me to follow the entire conversation, but I would confirm my current opinion that, whenever ML folks say "inductive bias" they mean: "We must rely on domain knowledge, we're not sure how to do it, so we'll call it 'inductive bias', it sounds ML-doable."

3.31.2022 10:42pm - (Replying to @HillelNeuer and @FranceskAlb) Now we see where @HRW got the "apartheid" theory.

3.31.2022 10:04pm - (1/2) The argument of "shoot up pedestrians" has stopped working on modern "Zionophobes". In their mindset, every Israeli pedestrian, even a 6-yr old girl, is a potential mother of a soldier, hence an "oppressor", if not a "war criminal". An argument that often works for me is:
3.31.2022 10:04pm - (2/2) "No! Palestinians do not have the right to "resist" defenders of a country they wish to destroy." At this point Zionophobes usually say: "All Pal want is freedom..." To which I reply: "So, Israel does have the right to exist, can YOU state so publicly?" -- Pindrop silence.

3.31.2022 8:57pm - (1/2) The impetus to switch to CSM was more prosaic. Tom Verma and I were looking for an intuitive interpretation of DAGs, so intuitive that people would stop asking us "what does it mean". Also, we wanted to embrace cycles, which is impossible with conditional probabilities. The
3.31.2022 8:57pm - result was this paper (see Def. 2, first published 1991). Pedagogically, the impetus can be seen in Causality page 35; the minimal "shieling" set of Y is {0}, which doesn't match intuition, since X does affect Y. (i.e., Y listens to X in any state of u).

3.31.2022 3:16am - Another Stabbing attack in Israel -- screwdriver in a passenger's neck., while Blinken legitimizes the arch-ideologues of hate:

3.31.2022 2:55am - The definition of a causal DAG is rather simple: The set parents of each variable X is a minimal set of variables that, if intervened upon would "shield" X from all other interventions. By "shield", we mean keeping P(x) unaffected by such interventions.

3.30.2022 11:05am - Looking back, how often have we consulted College Ranking? How often have we participated in one? Time to re-evaluate.

3.30.2022 10:43pm - Clearest exposition I heard on why America can't afford the risks of the proposed "Iran deal."

3.30.2022 10:40pm - Mazal Tov @deborahlipstadt , In your strength comes ours.

3.30.2022 9:29pm - University of California system considers mandatory ethnic-studies requirement via @JNS_org

3.30.2022 3:59pm - Honored to participate (April 6) in this distinguished lecture series at the Simon Institute: Glad to see causal inference inspiring several of these lectures.

3.30.2022 12:11pm - (Replying to @soboleffspaces @dylanarmbruste3 and 4 others) Yes. When I speak of RCT, I have in mind an "ideal" RCT, i.e., a study capable of estimating E(Y_1 - Y_0). Though I know, based on our conversations, that practicing trialists think differently. They do not have an estimand in mind; RCT is whatever an approved protocol delivers.

3.30.2022 11:09am - (Replying to @ducha_aiki and @wellingmax) Illuminating post. Key quote: "I can’t imagine that our country can be an aggressor." Which matches a lecture I heard from an insider, and which we find hard to believe, the Russian people do support Putin. Most people under totalitarian regimes do. Leaders speak to peopl hearts.

3.30.2022 4:01am - (Replying to @KenRoth) Mr. war-crimes, people expect you to say "I'm sorry". See why:

3.30.2022 3:51am - The terrorist who chose the town of Bnei Brak did not contemplate exposing to the world the mosaic of pluralistic Israel, nor did he intend to reveal the human consequences of Western anti-Israel orgies. Honk, Honk, howled @amnesty . Honk, Honk, yodeled @KenRoth .

3.30.2022 3:19am - Condemn? Wait a minute! What has your country (Belgium) done to curtail European support of the "pay for slay" enterprise? Last I heard, Brussel is still in Belgium.

3.30.2022 3:09am - They sought refuge from one inferno in another. Two countries whose neighbors wish eliminated, and my academic colleagues say: you have to look at both sides.

3.30.2022 2:51am - (Replying to @wellingmax) On the positive side, they are talking time table of 12-18 months. By that time, I hope, we would know what portion of the planet will not go up in flames.

3.30.2022 12:35am - In Defense of Straight Talk via @jewishjournal
3.30.2022 12:35am - Another bold opinion about where our society is heading. Makes you think: when was the last time I've exercised self-censorship? Shouldn't we, in privileged academia, be the most outspoken straight talkers around?

3.29.2022 10:56pm - (Replying to @soboleffspaces @gottfriedmath and 3 others) Trying to help clinical experts is tricky. They tell you that they think CRTs, but my suspicion is that they secretly think SCM. Ask them: "Assume you intervene on X, would adding interventions on Z make a difference on the distribution of Y?" See if they feel comfortable.

3.29.2022 10:48pm - A revised version of this paper is linked here A more comprehensive treatment of the entire causal hierarchy can be found here: (Chapter 27 in the ACM book

3.29.2022 10:30pm - (Replying to @gottfriedmath @soboleffspaces and 3 others) Interventions, be they physical or hypothetical, are used to decide the topology of a DAG in Causal Bayesian Networks. Counterfactuals, encoded as functions (not distributions), are used to decide the topology of an SCM. The difference is shown vividly on page 35-6. Key to Rung-3

3.29.2022 10:13pm - (Replying to @soboleffspaces @pablogerbas and 2 others) Agree! The causal reading of the parents-->child arrows is a corollary of the invariance assumption P(v| pa(V), do(x)) = P(v|pa(V)). Page 24 of Causality gives a complete characterization of Causal Bayesian Networks and states that the local property is not sufficient. (homework)

3.29.2022 9:55pm - (Replying to @FJnyc and @mehdirhasan) Yes, @Mehdirhasan blocked me. I hate to speculate why.

3.29.2022 6:57pm - "Data should be a means to knowledge, not an end in themselves." A friend sent me this article in Nature: which reminded me of our discussions on the radical data-centric philosophy of machine learning:

3.29.2022 6:10pm - (Replying to @DanielHGill @amnestyusa and @KenRoth) Hillarious! I wish she would realize what she said and rush to apologize. Then I can see @KenRoth scolding her for bowing to Zionist pressure. God bless Zionist pressure, it keeps the planet from going insane.

3.29.2022 5:09pm - This vold opinion piece by Moshe @vardi deserves serious thought by every computer scientist.

3.29.2022 5:03pm - Nelson Mandela's Grandson Blames Russian Invasion on 'Ukraine Neo-Nazis' and 'Apartheid Israel Dogs of War'
3.29.2022 5:03pm - When I labeled anti-Zionists "Zionophobes", friends told me: No way, they hate but they are not out of their wits. I was right! Zionophobes can't tell right from wrong, truth from falsehood, fair from foul. They're literally consumed by hate. Observe Mandella's grandson.

3.29.2022 4:01pm - (1/2) AEN sent a letter this morning to UC President Drake regarding UC Centers that are still institutional members of MESA, thus implicating their campuses in a BDS campaign to incriminate faculty and students who collaborate with Israeli universities.
3.29.2022 4:01pm - (2/ ) I should add that I am one of those professors; I have a "visiting professor" appointment at the Technion (my alma mater) and there is nothing that would give me greater pleasure than finding out that one of my algorithms is helping the Technion contribute to Israel's defense.

3.29.2022 3:37pm - Now I understand why @mehdirhasan blocked me. He knew I'd remind him that Palestinians were merely asked to accept the legitimacy of their neighbor, which Putin does not, Arafat did not, and Mehdi cannot (mother milk?).

3.29.2022 3:18pm - (Replying to @BenMFreeman @blakeflayton and @mehdirhasan) Ben, what did @mehdirhasan tweet this time? I've just found out he blocked me, so I can't read his pearls of wisdom.

3.29.2022 2:48pm - I'm not sure which theory deserves greater admiration, that Rothchild paid Darwin to invent evolution, or that Israel is an apartheid state. @amnestyusa and @KenRoth are still undecided.

3.29.2022 2:32pm - Bnei Brak is my home town. This is where I first learned to say "Salam" (peace) to the Arab kids who came over with their donkeys to play in the orange groves. We played soccer too, and often took a swim in their village pool (we had none). Bnei Brak is my home town.

3.29.2022 6:13am - (Replying to @GeraldNGOM @aarondmiller2 and 3 others) Confession is good if you are willing to learn from your mistakes. @aarondmiller2 is still unwilling to let go of the key presumption that has led Americans predictions astray: that "Palestinians' national ambitions" could be fulfilled, short of a total destruction of Israel.

3.29.2022 5:31am - Thanks for posting. We have seen some of these KGB documents and quotes in various articles, but to have them all compiled under one cover is a treasure.

3.29.2022 5:12am - Rumor has it that even Herzl was a Palestinian.

3.28.2022 10:50pm - When you think how many lives they saved by that split-second decision to charge forward, rather than take shelter, you begin to appreciate their commitment, their comradeship, their humanity -- our loss. Yehi Zichram Baruch

3.28.2022 8:54pm - Bernard-Henri Lévy's has returned from Ukraine with: " A prayer for Odessa", see: Recall, Lévy will deliver the 20th Daniel Pearl Memorial Lecture this coming Thursday. For information, see:

3.28.2022 7:22pm - (1/2) You don't need to be a Middle East scholar to urge your university to cut ties with MESA, see why: And, if they must affiliate with a ME association, there are alternatives, eg @asmeascholars at a scholarly association devoted
3.28.2022 7:22pm - (2/2) to Middle East research and a free exchange of ideas.

3.28.2022 6:50pm - "Changing the narrative" sounds revolutionary today. But to one who grew up with this same narrative since kindergarten, it sounds more like a precautionary alarm: "Will it last?" "Will it spread?" "Will it reach our neighbors?"

3.28.2022 2:46pm - (Replying to @dylanarmbruste3 @RWJE_BA and 5 others) I meant to reaffirm your observation that there is something strange with EBM logic. Instead answering you question of what the estimand is, and how to tell whether an RCT delivers an estimand, what we hear is repetition of the mantra that there is nothing above the RCT.

3.28.2022 1:20pm - (Replying to @dylanarmbruste3 @RWJE_BA and 5 others) You are assuming that EBM folks have an "estimand" in mind, or on paper; thus far I havn't seen evidence of that assumption.

3.28.2022 9:00am - (Replying to @soboleffspaces @dylanarmbruste3 and 5 others) Curious, what is the 2-model approach to estimating ATE?

3.28.2022 8:57am - (Replying to @soboleffspaces @dylanarmbruste3 and 5 others) I too have noticed this phenomenon. EBM priests ask: to what degree studies resemble RCTs, instead of asking: to what degree an RCT helps answer a research question.

3.28.2022 8:04am - The tragedy is that Ben and Jerry themselves sound like decent folks, just trying to project some virtue, like most Jewish Americans, to feel somewhat good, Tikun-Olamniks, enchanted by BDS's slogans, blinded to its ultimate aim.

3.28.2022 6:29am - Guess what news hit me two hours ago, in the midst of Putin's brutal attack on Ukraine. Our publisher says #Bookofwhy will soon be published in Moscow. No way! But there was a contract to sign, and a thousand thoughts to reign in. Moscow? My book? I'm begging for a cheerful one.

3.28.2022 1:59am - (Replying to @kareem_carr) Thanks for posting. There are so many people to whom I would love to write portions of this letter, that I know it will come handy soon. But, note, this is a letter of "disengagement" not "cancellation", written on the basis of professed ideology, not on rumors/hearsay.

3.27.2022 1:04pm - Scott, it seems that your followers really like the crash course, and I can't figure out why. To my mind, all these tricks are just corollaries of the First Law. Try sending them the First Law and see if they wouldn't appreciate it even more.

3.27.2022 10:29am - (Replying to @causalinf @jmwooldridge and 2 others) Yes, I read it in Feb of 2019 and commented: According to others (eg @marcfbellemare , @PHuenermund @causalinf), economists are beginning to rebel against the tyranny of outdatedness. Evidently, only econ students are rebelling, as we can see on this channel.

3.27.2022 1:12am - I like the paradigm shift from the "3 No's" of Khartoum to "3 Yes's of #AbrahamAccords that we are about to witness tomorrow, at the Negev Summit. Only dinosaurs old enough to remember the Khartoum's 3 No's can appreciate the devastating effect it has had on the soul of peace.

3.26.2022 10:56pm - My kids tell me that the word "barbarians" may be offensive to some people. I tried to contain myself.

3.26.2022 10:06pm - Anyone who supports BDS has no credibility speaking about discrimination, oppression, freedom and human rights.

3.26.2022 9:52pm - (Replying to @soboleffspaces @pablogerbas and 2 others) What makes the pa(X)→X dependency 'causal' is the property: P(x| do(pa(X))) = P(x|pa(X)) plus additional properties, specified on page 24 of Causality.

3.26.2022 9:39pm - (Replying to @causalinf) But make sure you hit refresh -- it was revised March 2022.

3.26.2022 6:07pm - Join me in celebrating this historic event. Note however that the "peace" has been extremely cold, until very recently. E.g, Egyptian journalists were forbidden (by their peers) from visiting Israel. Things changed lately, perhaps Abraham accord, perhaps Iran, perhaps commonsense

3.26.2022 5:51pm - Here are tangible examples on fighting boycotts with inclusion. Urge you institution to follow these directions, or initiate your own - to multiply the effect. Thanks for posting.

3.26.2022 5:35pm - (Replying to @anndvision @noah_greifer and 5 others) I would never say is is not reasonable to ask a question about a human being. But I would like to exhaust first the possibilities latent in non-chaotic systems and, then, if proven hopeless, explore chaotic directions. So far, I've been mighty encouraged by those possibilities.

3.26.2022 3:43pm - (1/2) Great article on the dangers of academic boycotts. Still, I am in the opinion that academic boycotts should be fought with academic embracing. If you have ties with an Israeli university, strengthen them. If you don't, initiate them. Invite an Israeli scholar to your lab,
3.26.2022 3:43pm - connect your students with an Israeli team working on related topics. In short, fight destruction with construction. And do it publicly, to show destroyers defeated by inverting their intentions. Who said academia can't fight its predators?

3.26.2022 1:04pm - (Replying to @soboleffspaces @anndvision and 5 others) Yes, Y(x,u) has a definite value given x and u, in any SCM model M. Moreover, a SCM model M assigns a truth value to every counterfactual statement.

3.26.2022 12:57pm - (Replying to @soboleffspaces and @attilacsordas) In SCM, the justification of consistency is easy. It goes: What is Y(x,u)? It is defined by the First Law. Therefore, consistency is a proven Theorem. In PO, Y(x,u) is defined by intuition about treatments, individuals and outcomes. Therefore, consistency must rest on intuition.

3.25.2022 3:12pm - Quote of the Week: Potential outcomes without the First Law are 2 decades behind the times. Have a nice weekend.

3.25.2022 5:37pm - (Replying to @soboleffspaces) This sounds too self serving. I prefer to name it just "First Law," get people curious, and let them find out for themselves how fundamental it is, and how everything, literally everything in causal thinking follows from it. See

3.25.2022 11:32am - (Replying to @bveg3 @noah_greifer and 4 others) It is good to agree on these points; we even have the math to prove us right. I hope we can also agree that bounding ITE holds promise for personalized medicine, and that potential outcomes without the First Law are 2 decades behind the times.

3.24.2022 11:32pm - (Replying to @AJCGlobal and @amnesty) Why is @AJCGlobal still using the outdated phrase "does nothing to advance the cause of peace," as if anyone in @amnesty or @unhrcpr cares. Replace with: "emboldens Palestinians hopes that eliminationism works and the West is falling for it."

3.24.2022 10:25pm - (Replying to @soboleffspaces @causalinf and 2 others) HMMM... But E(Y_1) and E(Y_0) do not give us EIT = E[(Y_1-Y_0)>0]

3.24.2022 9:34pm - Good news. Brandeis University terminates its membership in MESA, over BDS. But I wish they would explain the move with a bit more compelling argument: BDS is an eliminationist movement, not simply one that opposes academic freedom.

3.24.2022 9:12pm - Finally, the word "apartheid" is given semi-formal definition, with mathematical equations, rather than empty slogans.

3.24.2022 6:03pm - Down with outdated definitions. Apartheid -- a word emptied of its content by Israel maligners (Newest Oxford Dictionary).

3.24.2022 2:22am - While UCLA is examining its relation with MESA, I have decided to take unilateral, one-man action. Now that MESA has turned Israeli students and academics into a disadvantaged minority, I am starting a "reverse discrimination" program to assist them in any way I can - join me!

3.24.2022 2:03am - (Replying to @RekaSundaram) I'm the luckiest person on earth, coming to know that what he does for fun inspires others to do even more. Thanks for letting me know.

3.24.2022 1:14am - (Replying to @CotlerWunsh) An example of how a bunch of pseudo-academics form a "clinic", hijack the good name of Harvard University and issue "A legal analysis.." Too bad I am not in the UCLA Law School; I would issue a "Legal Analysis" of those I dislike, criminalize them, and let the press rejoice.

3.23.2022 7:00pm - Shouldn't we expect all Centers for Jewish/Israel Studies to urge the Harvards, Princetons, and Brandeises to withdraw their MESA memberships? How can a University undermine its own Jewish/Israel Centers, which were created to promote academic collaboration with Israel?

3.23.2022 6:35pm - (Replying to @TerLucile and @soboleffspaces) If I were purist, I would say: Interpretation? Why? The formulas themselves provide an interpretation! What else you need? Luckily, I am not purist, and I think you are seeking "causal interpretations", in terms such as "direct" of "indirect effects" right?

3.23.2022 5:50pm - I will be speaking April 6 to the Simon Institute: This time, my abstract opens with: "What does it mean to “understand” a phenomenon, a domain, or a situation?" If you are still curious, tune in.

3.23.2022 3:52pm - (Replying to @Martin_Kramer and @J_d_parker) Is the Near East Study the only UCLA dpt affiliated with MESA? Is it a departmental affiliation or University affiliation? How will continued affiliation with MESA hurt UCLA's reputation?

3.23.2022 3:29pm - The identity of pa(X), the CAUSAL parents of X is crucial, and cannot be replaced by non-causal predictors of X. See "Confounding equivalence," which discusses: What's the minimum information needed for the adjustment formula. Good paper. @VC31415

3.23.2022 3:09pm - (Replying to @VC31415 @soboleffspaces and 2 others) The LHS is defined in Causal Bayesian Networks, which are Rung-2, and does not need functional relations as in fully specified SCMs.
3.23.2022 3:09pm - (Replying to @soboleffspaces @VC31415 and 2 others) This is not my writing. I wouldn't write w/o distinguishing Rung-2 from Rung-3.

3.23.2022 1:57pm - (Replying to @soboleffspaces @causalinf and @nickchk) We are asking: what information must we have before the adjustment formula is valid. Ans. We must have (1) Rung-1 data to estimate the cond. probabilities involved, and (2) Some causal information coming from the graph (Rung-2). For example: Who are the (causal) parents of X.

3.23.2022 6:45am - Eleven (11) new orphans have woken up this morning, with eleven new questions. Israel's demonizers are printing new reports, no time for questions. Hyenas laugh, hyenas print, can't hear no question.

3.23.2022 3:10am - (Replying to @soboleffspaces and @JohanDH2O) I have naively assumed that all arrows are causal (as defined above). But you are telling us: No need, trialists hate causation, they love conditional independencies. Lets use the graph to convey independencies, not causal arrows. I say: OK, just tell me which independencies.

3.23.2022 2:57pm - We all know that CI is becoming fashionable, but someone just sent me this CACM article to prove the point: It should provide ML students powerful talking points, in case their professors are still thinking data-fitting; many still do.

3.23.2022 1:45am - (Replying to @soboleffspaces and @JohanDH2O) You are describing "causal" dependencies by talking about "the state of the node" as opposed to "the state of our belief in the state of the node". The latter goes both along the arrow and against the arrow - symptoms change our belief about the disease.

3.23.2022 1:38am - (Replying to @soboleffspaces and @JohanDH2O) The causal assumption conveyed by Z1--->X--> is P(z1|do(x)) = P(z1) I am not saying we need this assumption, but if someone claim we do NOT need it, all we need are CI's, I would like to know what CI's we do need.

3.23.2022 12:41am - (Replying to @soboleffspaces and @JohanDH2O) I can't answer for @JohanDH20, but a typical "causal assumption" would be X--->Y, namely P(x|do(y)) = P(x). Why we call it "causal"? Because it answers an interventional question that cannot be expressed in the language of probabilities.

3.22.2022 11:04pm - (Replying to @soboleffspaces @causalinf and @nickchk) Take any graphical representation, DAG, or undirected graph, or chain graph. What does it say? It is a list of CI's that can be read from it according to some rule of reading. I am trying to pinpoint those CI's that you are trying to capture with the graph.

3.22.2022 10:24pm - (Replying to @LMplusG) It was Darryl Issa who was briefed by Malley. We must go but what he learned.

3.22.2022 10:20pm - (Replying to @soboleffspaces @causalinf and @nickchk) The semantics of a graph lies in the set of conditional independencies (CI) it coveys. That is why I am trying to understand what CI's you wish to represent before asking how we represent them.

3.22.2022 10:15pm - This is probably the most depressing message I've retweeted on this channel. Everything that Robert Malley has said in the past turned out dead wrong, but this new Iran deal is much worse; its heading towards a hyper-disaster, for America, for Israel and for the West.

3.22.2022 8:47pm - Reminds me of poor AOC. AOC: "Israel puts children in cages!" Host: "Cages?" AOC: "Well, I am not a Middle East expert..." It makes me wish I was a biologist.

3.22.2022 8:06pm - Rejoice! Rejoice! Your prayers to get a revised and corrected KINDLE copy of Primer, have been answered. Wiley swears that you'll now get the revised version. Make sure the number 20,000 appears on the "authors information" page, 3rd line from the bottom.

3.22.2022 6:32pm - (1/2) (Replying to @soboleffspaces @causalinf and @nickchk) I have two comments on your interpretation. First, you limit yourself to studies where the research question is P(y|do(x)) only (assuming Y does not cause X), and you miss on all the gazzions of other causal effects that we may wish to estimate from the model. Second, you
3.22.2022 6:38pm - (2/3) (Replying to @yudapearl @soboleffspaces and 2 others) may be right that, in such limited study, the only causal assumption we need is X-->Y and the rest are conditional independence assumptions. I can check on whether you are right only if you specify precisely what additional assumptions are needed. Assume you have five
3.22.2022 6:46pm - (Replying to @yudapearl @soboleffspaces and 2 others) variable in your data: X,Y,Z1,Z2,Z3. What cond. Ind. assumptions do you need? (1) those implied by P(z1, z2,z3)? or (2) by P(x,z1,z2,z3)? or (3) by P(x,z1,z2,z3,y) or (4) by others? Once we know which, we would be able to check if they are sufficient for the task.

3.22.2022 7:24pm - (Replying to @soboleffspaces @causalinf and @nickchk) How we encode the independencies is important, but first we need to agree on what independencies we need. Have you decided on (1), (2) (3), or (4) ?

3.22.2022 12:15pm - (Replying to @soboleffspaces @causalinf and @nickchk) I am not sure I can agree with that, but willing to learn new interpretation. I'm under the impression that DAG relationships must be causal. Otherwise, if only conditional independencies, how would we be able to tell X--->Y from Y--->X ? Both convey same independencies.

3.22.2022 4:15am - (Replying to @dad_zibao @leads_connect and 2 others) You ask "Where is the #SCM?" I'm afraid ML folks wouldn't know what you are talking about. It's not in their education, nor training, and when you remind them that you can't make decisions w/o causation, they answer: No problem, its just "inductive bias".

3.22.2022 2:30am - So, how come I can listen to Assad all day, and my stomach turns upside down when listen to @amnestyusa ? Is it just the feeling that Assad truly believes in what he says? And that what he says is truly believed by millions and millions of his subjects -- Israel's neighbors.

3.22.2022 1:27am - (1/2) Continuing this line of thought, the graphical model can be viewed as a parsimonious summary of Zillions of RL simulations, involving all possible actions or combinations of actions, and all possible outcomes or combinations of outcomes. The miracle of parsimony comes from
3.22.2022 1:27am - (2/2) the fact that, if the data generator is a DAG, then those zillions of combinations are not totally independent of each other, but must conform to three constraints articulated in Definition 1.3.1, page 23-4 of Causality (2009). A mighty miracle, w/o which we won't survive.

3.22.2022 12:43am - (Replying to @DrRJCarter and @mounirshita) I would phrase it a bit differently. It's not the do-operator that takes you from Rung-1 to Rung-2, but the information encoded in the graph, parsimoniously, which allow the do-operators to produce (together with data) the consequences of every action or combination of actions.

3.21.2022 8:19pm - If you are connected with a UC campus, please consider signing this important Letter to the UC Academic Council asap, opposing new, politically loaded admission requirements: You can sign it by filling the Google form here:

3.21.2022 3:23pm - Remember our Crash Course in Good and Bad Control? Well, it has been revised again (to pacify another reviewer) and it's now close to perfect, with references to everything that was ever done on "control", "adjustment" "covariate selection," OVB, Enjoy!

3.21.2022 2:56pm - A new paper by Bernard-Henry Levy in the wake of President Zelensky's speech at the Israeli Knesset: BHL will deliver the 20th Daniel Pearl Memorial Lecture next week, see

3.21.2022 3:58am - Going back to ACM's book, Chapter 32, on "Actual Causation" can be accessed here: Chris Hitchcock speaks for a circle of philosophers that take structural models seriously, as mental representations of commonsense knowledge.

3.21.2022 3:12am - (Replying to @offaltube and @sullydish) Yes, thank you for correcting me. It was April 13 1948 when the medical convoy to Hadassah was attacked. And I don't take such errors lightly, b/c truth, facts, dates and figures is our only remaining weapons these days, when history is distorted by professional fact finders.

3.21.2022 2:27am - (Replying to @JFronczekMD @soboleffspaces and @RWJE_BA) To do better, a trialist should ask herself: What scientific principles made me fall in love with CRTs? Perhaps these same principles would allow me to improve on what my fellow trialists think is the best we can do? This question will take you to CI and to a new way of thinking.

3.21.2022 12:37am - Yemeni-Houthi leader: Ukraine at war because it is led by a Jew
3.21.2022 12:37am - I like the philosophers in the crowd!

3.21.2022 12:20am - Usually, when @mehdirhasan is "disappointed" we know that something positive took place in the ME. This time, alas, it is old news: Palestinians support Putin. Only Zionophobe @MaxBlumenthal seems surprised; he has hoped Mehdi would join him in condemning NATO.

3.20.2022 8:16pm - (1/n) Yes. It's the best discussion you will ever hear: I listened to it again to understand the host @sullydish ; what makes a seasoned journalist believe or pretend to believe in the fable that Israel was created with the intent to displace Palestinians,
3.20.2022 8:16pm - (2/3) Worse yet, what makes him fall for recent Palestinian logic: "We fought tooth and nail against co-existence, and managed to prevent it; this proves that it is not a viable solution - quit!" I came to a conclusion that he actually believes in it, he is not pretending,
3.20.2022 8:16pm - (3/4) and the reason he fell for it is that we stopped telling our narrative, even to our children and our Rabbinical students. We assumed that our struggle for coexistence is so just and self-evident that it needs no repeating. So we relaxed and let E. Said and I. Pappe take over,
3.20.2022 8:16pm - (4/4) and the actual story of 1917-1948 Zionism disappeared. Even good souls like @sullydish and @Sethrogen missed the story that I grew up on:, cast in the slogan "Two states for two peoples, equally legitimate and equally indigenous." Time to retell it!

3.20.2022 5:02am - (1/3) Agree, it's the best discussion of the Palestinian-Israeli conflict that has ever been aired anywhere. Yet I can't bring myself to believe that @sullydish never heard of the medical convoy of 1947 or the Arab invasion of 1948. More stunningly, he seemed unaware of
3.20.2022 5:02am - (2/3) the today's super obstacle to peace -- Palestinians DNA-deep rejectionism of any form of Jewish sovereignty. This blind spot, I believe, casts doubt on his sincerity. A seasoned journalist of his caliber can open an eye or two, rather than recycle cliches. Yossi,
3.20.2022 5:02am - (3/3) admirably, refrained from jumping: "Come on, Andrew, Don't tell me you are unaware of X or Y or Z..." I say "admirably," because I wouldn't have been able to control myself.

3.19.2022 11:09pm - (Replying to @chuluutiinam) She is not! Decrying conditions of some and devalue lives of others, not less human, befits a racist, not a Congresswoman.

3.19.2022 10:57pm - When an Amnesty Zionophobe is telling a people who they are and what's good for them, I would not try to confront him with facts, logic and values. Racists do not get it.

3.19.2022 5:53pm - Chapter 33 from the ACM book, reads: "Causal Diagram and Social Science Research" by Kosuke Imai (Harvard). It can now be accessed here: Economists who see economics as a branch of social science should appraise Imai's vision.

3.19.2022 4:51pm - Only a rare combination of profound ignorance and obsessive racism can breed mentality like AOC's, for whom Israel is not a moral imperative of a nation fighting for its existence, but a disposable, mentally disturbed child abuser.

3.19.2022 1:48pm - Curious readers asked me to reveal the name of that economist who, according to Causality page 216, footnote 10, was the only one to solve for ATE in a cyclic demand-price model. I can't resist: his name is Ed Leamer, well known to many, and a remarkably open-minded economist.

3.19.2022 1:24pm - (Replying to @HL327 @artistexyz and 9 others) If there is one thing I wish readers to take from #Bookofwhy, it is that RCT's and all their variants must be based on ATE, not the other way around. The latter is a stable property of reality, the former, man-made tricks that (occasionally) glimpse at reality.

3.19.2022 1:14pm - How new hopes emerge from the dark stage of history

3.19.2022 8:59am - (Replying to @howardrgold1 @AsraNomani and 2 others) One reason UCLA is 33% Asians: I tell all my Israeli friends to check the "Asian" box on their application forms. It works - Moses turned east upon hearing about college quotas.

3.19.2022 8:41am - Please join me in welcoming @JaiceHarmon the 60,000-th reader to our educational channel. True, this number is significantly lower than some Rung-1 channels but, recall, it was only 30,000 years ago that evolution started climbing Rung-2 and Rung-3. We are on the right Ladder.

3.19.2022 8:33am - ATE (Average Treatment Effect) is defined in cyclic as well as a-cyclic processes. See Causality page 215-7 (don't miss the anecdote in footnote 10). Moreover, ATE is defined the same way in both systems, using the do-operator. The Wikipedia article is 3-decade old. Volunteer?

3.19.2022 3:14am - A colleague in Kharkiv writes: In these terrible times for our country, we ask for your support and ask you to sign an open letter by filling out the form We have over 8,000 signed, inc. respected scientists in the global community. We need your vote.

3.18.2022 10:50pm - (Replying to @soboleffspaces @HL327 and 9 others) I like the title "agricultural concept of ATE", though I am not sure trialists would agree that ATE is what their trials are attempting to estimate/maximize.

3.18.2022 5:00pm - (Replying to @soboleffspaces @ADAlthousePhD and 9 others) So where is the center of the trialist camp? Is it the Society for Clinical Trials (SCT)? Are they as secret as our discussants? I see they have a conference in San-Diego in May. Would they be open to a few "getting to know you" questions from outsiders?

3.18.2022 4:22pm - (Replying to @soboleffspaces @ADAlthousePhD and 9 others) A proud profession never jumps towards progress. Compliance with prevailing practices is its holy measure of success. That's why we hear so often: "This is not what we do" and so rarely: "This is what should should do." The latter requires a consensus of AIM.

3.18.2022 4:02pm - (Replying to @stephensenn @EngineerDiet and 9 others) Let's not get hung up on whether the "ideal" is realizable. Our question is what RCT is aiming to estimate. "Aiming" defines the "ideal". If there is no aim, there is no direction and no sense of better or worse design. I can't wait to hear what the AIM is, one formula please.

3.18.2022 3:55pm - (Replying to @soboleffspaces @ADAlthousePhD and 9 others) I believe I have done it for RCT as well, but it was a terrible mistake, never tell a proud profession what you think they are trying to do, let them articulate it themselves; we will soon be hearing from trialists, in one formula, what it is that RCT is aiming to estimate.

3.18.2022 3:28pm - (Replying to @EngineerDiet @stephensenn and 9 others) I am not familiar with acronyms such as "RTM-inducing selection thresholds", and I hate to speculate, b/c whenever I do, 12 trialists jump on me with indignation: You do not know about RCT. You can help us by just explicating what "ideal RCT" means to you. One formula please.

3.18.2022 2:47pm - (Replying to @ADAlthousePhD @soboleffspaces and 9 others) Illuminating document, except I wouldn't know where judgment ends and science begins.

3.18.2022 2:41pm - (Replying to @EngineerDiet @stephensenn and 9 others) Please define for us what an "ideal RCT" aims to estimate. "Ideal", not "mystical". And please, please, please, if at all possible, cast it in a formula, not hand-waving. @stephensenn assured us there is a notation for this aim.

3.18.2022 2:30pm - (Replying to @Jon_Williamson_ @federicarusso and 3 others) Your definition of "Evidential Pluralism" is so broad, that it embraces the definition of Science itself. Is there anything a scientist does that is not collecting and integrating pieces of evidence, hence "Evidential Pluralism". It explains why many agree that we need it badly.

3.18.2022 2:21pm - (Replying to @Jon_Williamson_ @federicarusso and 3 others) "Helplessness" is when people lecture on how nice it would be to be able to do X, decry/assume the absence of X in our current store of tools, and do not demonstrate how X can actually be done. X= Evidential Pluralism. None showed how combining evidence produces something new.

3.18.2022 4:23am - @Jon_Williamson 's craving for "evidential pluralism" reminded me of an Open Letter I wrote in 2010 to Nancy Cartwright, regarding a related craving, "causal plurality," I havn't received a reply yet, except an indirect one, through

3.17.2022 6:09pm - (Replying to @stephensenn @HL327 and 9 others) I am eager to see how your notation expresses what an ideal RCT aims to estimate.

3.17.2022 5:37pm - (Replying to @HillelNeuer) Plus, I would remind Ken Roth of what Mordechai told Queen Esther: “Do not imagine that you will escape with your life by being in the King's palace” “For, if you keep silent in this crisis, deliverance will come to us from another quarter, while you will perish."

3.17.2022 5:10pm - (Replying to @HL327 @stephensenn and 9 others) The nice thing about doing things formally is there are no "unrecognised assumption buried" they are all explicit and visible. Collapsible RD? This sounds like gossip from the 1990's. Can you explicate? Equations do not hide, they expose.

3.17.2022 3:15pm - (Replying to @federicarusso @soboleffspaces and 3 others) As someone in love with SCM, but always openminded about other adventures, can I benefit from your conception of "mechanism"? Is there one case of "pluralistic evidence" that I would miss? Hating to miss adventures.

3.17.2022 2:09pm - (Replying to @stephensenn @HL327 and 9 others) I wonder if trialists would agree on what "ideal RCT" is. Absent notation, everything is at the mercy of intuition, interpretation and mis-interpretation.

3.17.2022 12:26pm - (Replying to @HL327 @stephensenn and 9 others) Our analysis makes no assumption about the mechanism behind the data, save for the assumption that the RCT is ideal (no selection bias). You get different results by assuming some mechanism. Can you explain the difference? Is your RCT ideal? I'm under deadline pressures today.

3.17.2022 12:56am - (Replying to @lenlayton @psychunseen and @UmarN91) Saying "RA Fisher has done damage" would sound blasphemous to most statisticians. But in truth, the current resistance we see in CRT's circles to formalizing ATE in counterfactual terms has its root with Fisher's animosity to J. Neyman. #Bookofwhy describes what Fisher lost.

3.16.2022 10:13pm - I made the Epilogue available to our readers: It's a good read, enjoy.

3.16.2022 6:45pm - More information here: about our speaker, Bernard Henry-Levy, and the annual Daniel Pearl Memorial Lecture.

3.16.2022 6:38pm - (Replying to @HL327 @stephensenn and 9 others) The explanation you provide for the findings in the example are plausible. Our point is that the conclusion holds regardless of the explanation. Moreover, I would not attribute the surprising data to the RCT, which is assumed ideal. The culprit is whimsical free-choice behavior.

3.16.2022 4:04pm - (Replying to @stephensenn @soboleffspaces and 8 others) "Ask your doctor" is my poetic way of telling enlightened colleagues: It's really promising.

3.16.2022 3:38pm - (Replying to @soboleffspaces) The manipulation vs. randomization confusion is a result of not formulating the research question: What are we trying to estimate with RCT? Intuitively, they know it's "treatment efficacy," but if you deny it a symbol and a formula, you invite mis-communication.

3.16.2022 1:50pm - Pedagogically, it is always more effective to start with reality, or our model of reality, and then, only then, go to describe how we go about interrogating reality. If we do not have questions about reality, why bother conducting trials.

3.16.2022 1:32pm - Delighted and honored to invite all our 59.8K followers to attend (virtually) the 20th Daniel Pearl Memorial Lecture at UCLA and to listen to one of my most admired author/philosopher/speaker Bernard-Henry Levy, March 31, 2022, 12pm. Register below.

3.16.2022 7:24am - (Replying to @Foreman1David) I'm relieved. We hardly find stupidity or racism among Artificial Intelligence researchers. What an unfortunate coincidence of acronyms.

3.16.2022 6:21am - (Replying to @Foreman1David) Can you elaborate? Who in AI can make such statements?

3.16.2022 5:36am - Amnesty International, Israel and Race-Baiting –

3.16.2022 2:33am - (Replying to @stephensenn @soboleffspaces and 8 others) To eliminate any "baffling," please interpret everything I said or say about RCT to be about "ideal RCT", no imperfections, aiming to deliver an estimate of "Average Treatment Effect" (ATE) on some population of interest. I truly hope this clears away any baffling in the future.

3.15.2022 10:32pm - (1/2) A mini-victory at the University of Southern California! USC has revised its policy to disallow political statements that imply they are made on behalf of USC. As a tribute to the 65 professors who started the "USC faculty uprising" (see here:
3.15.2022 10:32pm -, I am retweeting my Purim Op-Ed, which talks about the risk and duty of speaking up out-of-order: Tonight, "The Fast of Queen Esther", it is fitting to remember her courage in taking that risk.

3.15.2022 9:28pm - As promised, I am posting another chapter from the ACM book Chapter 30 - Alison Gopnik on "Causal Models and Cognitive Development"

3.15.2022 8:57pm - (Replying to @soboleffspaces @f2harrell and 6 others) I am not sure. As I read @stephensenn 's papers I see concerns with selection bias, emanating from a variety of reasons, but it's still a bias associated with the disparity between the target population and the sample population. (They don't use this language).

3.15.2022 8:45pm - (Replying to @yudapearl @stephensenn and 9 others I truly do not know why you need us, CI folks? You have most trialists following your writings, we have none. Your theories are implemented and tested, ours are not. Do you want ask to consider problems you are having, or possibilities you haven't considered? How can we help?

3.15.2022 8:06pm - (1/2) (Replying to @f2harrell @JFronczekMD and 5 others) This is precisely what our definition of "selection bias" stands for: non-representative treatment effects in the study sample. Except, it is expressed formally, leaving no room for ambiguity. But now that we agree on the issue, do we agree on the solution? I've sent you
3.15.2022 8:17pm - (Replying to @yudapearl @f2harrell and 6 others) our solution in full details here: Is any of our theorems a special case of yours? Are there new principles in yours that we have failed to leverage? Let's be a bit technical, can we? 3.15.2022 8:06pm - (1/2) (Replying to @f2harrell @JFronczekMD and 5 others)

3.15.2022 7:32pm - (Replying to @jmwooldridge and @steventberry) I think you are still fighting the "silly SEM" period in the history of econ, to which PO contributed a wakeup call. But nowadays, when we start with "non-silly SEM" where autonomy is a defining feature, & from which POs emerge organically, if needed, do we need to mention them?

3.15.2022 4:03pm - (Replying to @yudapearl @steventberry and @jmwooldridge) added tools, algorithms or insight? I have seen those flowing one way: from SEM to PO, not the other way around. Calling y "potential outcome" serves only social function: assuring the PO priests that you still pledge allegiance to the PO flag, and haven't defected to SEM camp.

3.15.2022 3:57pm - (1/2) (Replying to @steventberry and @jmwooldridge) Every function, say y=f(x), has a counterfactual flavor to it: Had X been 3, Y would be f(3), etc. So, every time an economist put down a function, some PO enthusiast can claim: for me, y is just a potential outcome. But what purpose does it serve to say so? Does it come with
3.15.2022 4:03pm - (Replying to @yudapearl @steventberry and @jmwooldridge) added tools, algorithms or insight? I have seen those flowing one way: from SEM to PO, not the other way around. Calling y "potential outcome" serves only social function: assuring the PO priests that you still pledge allegiance to the PO flag, and haven't defected to SEM camp.

3.15.2022 11:56am - (Replying to @stephensenn @soboleffspaces and 8 others) Agree. If you only knew how eager CI folks are to learn from trialists, both problems and solutions. Unfortunately the language barrier stands in the way. We need Esperanto for translation.

3.15.2022 11:52am - (Replying to @stephensenn @JFronczekMD and 8 others) If you have "necessary" conditions, then they must embrace the conditions reported in as a special case. Do they?

3.15.2022 11:20am - (Replying to @soboleffspaces and @causalinf) The word "mechanism" is much sexier than the word "function". So Russo and Williamson are assured unanimous readers support for their romantic call for the necessity of mechanisms. Luckily, the features they hope would emerge from mechanism have emerged already from "functions".

3.15.2022 6:05am - Indeed, it's highly recommended, and answers many of the questions raised in our discussion of "inductive bias". The published version can be accessed here:

3.15.2022 5:53am - The least we can do for the defenders of Kiev is to visualize with them the Great Gate of Kiev, about 30 minutes into this performance of Musorgsky's Pictures at an Exhibition: Retweet to Putin: "Spare this one!"

3.15.2022 5:20am - (Replying to @f2harrell @JFronczekMD and 5 others) "Selection Bias" occurs when those who are recruited to the RCT are not the same as those who would eventually use the treatment. Formal definitions, examples and remedies are given here: [Nice paper; should be read by every trialist]

3.15.2022 3:33am - (Replying to @prem_k @EduMinOfIndia and 4 others) #MachineLearning of yesterday was based on correlation alone, not the #MachineLearning that I know.

3.15.2022 2:11am - As promised, I'm posting chapters of the ACM book to make them accessible to all. The first is S. Greenland on the causal foundations of statistics I like his: "basic statistics can and should be taught using formal causal models"

3.15.2022 1:49am - (Replying to @soboleffspaces @jmwooldridge and 3 others) I think @jmwooldridge had a valid point. He who hasn't seen econometrics confusion over SEM (in 1970-80) would fail to understand the zealotry with which PO folks resisted SEM's comeback (in the 2000-10) to offer a scientific basis for PO and ways to circumvent its weaknesses.

3.15.2022 12:55am - Happy Purim! @GalGadot . It used to my favorite holiday too. Recently, however, it has become more of a calling: We can't keep silent in such times and, as M. said to Esther: "Who knows, perhaps you have attained to royal position for just such a crisis."

3.14.2022 10:59pm - Refreshing to read an article by a philosopher who understands Causal Inference: It has a nice section on the choice of "causal variables."

3.14.2022 6:52pm - (Replying to @JFronczekMD @artistexyz and 5 others) I believe you got a few things totally reversed. Med folks prefer ATE over PNS because Observational studies disagree with RCTs when wrongly interpreted. Whenever RCT results clash with obs. studies, I would check the former for imperfections, e.g., selection bias. Tests exist.

3.14.2022 5:55pm - (Replying to @razingarizona) There is nothing wrong with a "Two State Solution Act" if it focuses on removing the main obstacle to Two States: Palestinian eliminationism, rather than the wishful obstacles.
3.14.2022 5:55pm - Reverse Orientalism galore, an Amnesty Zionophobe telling the Jewish people who they are and what's good for them. Amnesty's mentality: Israel ought not to be (nationally) Jewish, Mexico ought not to be Mexican and Palestine ... No, he wouldn't dare.

3.14.2022 5:36pm - All 25 Jewish Congressional Democrats Decry Amnesty Director's 'Disturbing' Remarks That Israel 'Shouldn't Exist as Jewish State'
3.14.2022 5:36pm - Reverse Orientalism galore, an Amnesty Zionophobe telling the Jewish people who they are and what's good for them. Amnesty's mentality: Israel ought not to be (nationally) Jewish, Mexico ought not to be Mexican and Palestine ... No, he wouldn't dare.

3.14.2022 4:41pm - (Replying to @HL327) We explain it in the paper: "It appears as though many women who chose the drug were already in an advanced stage of the disease" and would have died anyhow. Recall, the observational study is laden with confounding bias.

3.14.2022 4:24pm - (Replying to @JFronczekMD @artistexyz and 5 others) This long history, and its justified fear of confounding bias, is partly responsible for the stiff resistance of trialists and EBMists to welcome observational studies when they are no longer a threat, but needed for estimating a totally new parameter: Individual Causal Effect.

3.14.2022 12:02pm - (Replying to @paulgp @instrumenthull and 7 others) If you define "potential outcome" in counterfactual language, ie "had X been 0", you are OK. But if you define it in "treatment-assignment" language (as Rubin did) there is a gap to fill, since many SEM models do not have "treatment" defined.

3.14.2022 11:52am - (Replying to @steventberry @causalinf and 7 others) Except that "ceteris paribus" is undefined w/o SEM or DAGs. "other things being the same"? You certainly do not mean to keep the outcome the same. To define "other things" you need SEM

3.14.2022 11:42am - (Replying to @jmwooldridge and @steventberry) The 1980-90 regressional invasion of economentrics (See Causality, 2000) was traumatic, agree, and led to the PO rebellion and overswing. But the recovery has been traumatic too, and dogmatic, purging SEM & First Law from econ literature. Now people are asking for "the bridge".

3.14.2022 10:16am - (Replying to @jmwooldridge and @steventberry) Given the one-to-one mapping between SEM and PO, the usage of one or the other is a choice of which you prefer to keep implicit. In small problems you can afford to keep SEM in your head, in large ones you can't. Imbens says: most eco problems are small. I doubt it's by choice.

3.14.2022 9:56am - (Replying to @jmwooldridge and @steventberry) Each equation in SEM can be given a PO interpretation, and every PO expression can be evaluated from SEM. Therefore there is no "poor application of SEM" , nor "silly". But where is it more meaningful to start, SEM or "ignorability" assumptions in PO notation? Choice = ?

3.14.2022 3:29am - (Replying to @soboleffspaces and @Nichogachet) I've been more generous to Rubin's consistency here since, after all, he did not have anything resembling a model to derive it from, and had to pose it as an assumption to connect observables with PO.

3.14.2022 1:23am - (Replying to @jmwooldridge @causalinf and 5 others) I thought the question was about the bridge between "classic" econometrics and PO. I do not see the former in your proposal. Moreover, don't you agree that the work you hope to do is already done by the First Law? If not, what part of it is missing?

3.14.2022 1:12am - (Replying to @instrumenthull @soboleffspaces and 6 others) Good question! Y^x is defined in terms of SEM, which every economist learn in Econ 101. "Potential outcomes" however are defined (by Rubin) in terms of "treatment" and "assignment" which do not exist in classical economics. So, Y^x is NOT exactly PO, the former is well defined.

3.14.2022 1:03am - Never seen it done? Isn't the First Law of Causal Inference providing us the needed bridge, between classical economics and potential outcome? See Isn't that the bridge that @Nichogachet was looking for? @soboleffspaces

3.14.2022 12:14am - (Replying to @BethCarey12) I don't know about humans, but I for one wouldn't know how to infer "Can I trust him?" from "He tried to help you" w/o a causal model to tell me how "tried" and "help" are connected to "trust".

3.13.2022 11:30pm - Turkey to return Israel's most important ancient inscription
3.13.2022 11:30pm - I'm speechless. I've heard about this tablet since, as a 10 yr old, I studied the story of King Khezkiyau's tunnel. Thank you Turkey; A Tablet belongs in a country where 10 yr olds are taught its story.

3.13.2022 11:00pm - (Replying to @SAFECUNY) I would innocently raise my hand and ask (politely): "Where did Putin get the idea of proclaiming his neighbor "illegitimate"?

3.13.2022 10:24pm - Unless the language contains causal sentences, e.g., "Had Oswald not killed Kennedy ..." or, "if it weren't for Covid-19,..." or, "He meant to help you..."

3.13.2022 6:38am - (Replying to @tarvaina @KyleCranmer and @wellingmax) I have a suspicion that you are coming from ML culture. Why? You forgot to list a third option: Keep the same data, keep the same learning algorithm but incorporate my knowledge into it.

3.13.2022 3:08am - (Replying to @KyleCranmer and @wellingmax) The more I listen to how ML folks describe what "inductive bias" is, the more I am tempted to ask: What would be lost if we replace the words "inductive bias" with the words "whatever we do in ML" ??

3.13.2022 1:15am - After over three decades of covering Russia, Nick Robertson (CNN) leaves in despair and summarizes both history and current events:

3.13.2022 12:08am - (Replying to @hsvgbkhgbv) Renaming for the sake of ownership is forgivable, every field prefers homegrown over imported vegetables -partly to subsidize local farmers, partly for the sense of freshness. But when renaming emboldens laziness I begin to worry.

3.12.2022 10:47pm - (Replying to @wellingmax) How can you say that you do NOT EQUATE "inductive bias" with "domain knowledge" when you do not allow any part of the latter to affect system's decisions unless it is produced by inductive bias? Can some domain knowledge circumvent the walls of inductive bias?

3.12.2022 8:48pm - (Replying to @KyleCranmer and @wellingmax) What you are saying (and I agree) is that, from input-output view point, your decision is a product of two inputs: Data and domain-knowledge. How you combine the two is an internal detail, part of which may be inductive-bias or some regulation trick. These details are secondary.

3.12.2022 4:16pm - (Replying to @wellingmax) I almost celebrated our convergence into model = data + knowledge, but your last sentence takes us back to knowledge = inductive bias. Isn't "domain knowledge" something we assume in our model? Or do you insist on discarding all domain knowledge that is not learnable from data?

3.12.2022 4:04pm - (Replying to @wellingmax) If I take you paradigm at face value: model = data + inductive bias, I find a paradigm that is driven by two and only two inputs: data and inductive bias, leaving no room for domain knowledge. Can we converge by broadening our paradigm to read: model = data + knowledge

3.12.2022 11:17am - (1/3) It seems that ML folks use the term "inductive bias" to replace what other folks call "domain knowledge". I see two problems with this terminology. (1) It is inconsistent with canonical examples of inductive bias (see Wikipedia) which are universal rules of preference among
3.12.2022 11:17am - (2/3) fitness functions. "Domain knowledge," as the name implies, is domain specific, not a preference relation applicable to all problem instances. (2) What is gained by replacing a familiar term, "knowledge", with one that is slippery and easily misinterpreted? I can only think
3.12.2022 11:17am - (3/3) of "laziness" or "laxness", i.e., a message of "don't bother with issues of knowledge representation and usage, everything will be taken care of by some all-encompassing bias principle, like Occam's razor, or symmetry invariance, that is ML's past specialty. Don't bother."

3.12.2022 1:04am - (Replying to @wellingmax and @tdietterich) Your paradigm model = data + inductive bias jolted me to looked up Wikipedia for "inductive bias" and come back, horror-stricken, to ask: What connection whatsoever is there between "inductive bias" and causation, not to mention viewing the latter as a product of the former?

3.11.2022 7:55pm - (1/2) (Replying to @wellingmax and @tdietterich) I don't think we disagree on what kind of hybrid systems AI will eventually converge on, but I may be more concerned than you about slowing the process by viewing causality as a biproduct of "inductive bias", which makes people lazy, hoping that causality would hatch out
3.11.2022 7:56pm - (Replying to @yudapearl @wellingmax and @tdietterich) automatically if only they learn what "inductive bias" is - it won't. Causality has its own idiosyncrasies which must be understood and attended to, as this paper is describing in the language of deep learning.

3.11.2022 5:27pm - (1/2) (Replying to @david_lagnado @nerdy_grad and 3 others) I didn't think anyone would deem me capable of such trickery. Plus disseminating the ideas of Personalized Decisions is of much higher priority than promoting a book. The former has so many paradigms to shift that I can't resist linking to it again, here:
3.11.2022 8:35am - (Replying to @tdietterich and @wellingmax) I agree with your rating, but I cannot bet against the Ladder of Causation which is enshrined in math. I don't think @wellingmax meant to bet against it either (correct me if I'm wrong), but the appeal to ""Differential symbols" may sound like resisting symbiosis.

3.11.2022 7:56am - (Replying to @ThosVarley and @Abelaer) The so called "statistical measures of causality" explored here are all plain wrong, because they are expressed in terms of conditional probabilities instead of counterfactuals. The latter is what the authors of those measures have in mind but, left as such, lead to inconsistency

3.10.2022 11:54pm - You right, the correct link is:

3.10.2022 11:43pm - (1/4) My reading of this Tweet is: You'll never make DL people admit they erred; they are determine to own every wall they hit. "Differential symbols of sorts" are created to ensure that no matter what turn AI takes in the future it will carry a "made in DL" label. That's why
3.10.2022 11:43pm - (2/4) I'm not fond of the neural vs. symbol dichotomy, which is so easy to blur. I'd rather distinguish frameworks by asking what information they need, ie, is it purely data-centric, or a data + model symbiosis? This distinction makes it harder to cheat, because we know what
3.10.2022 11:43pm - (3/4) can and cannot be done with data alone. For example, data alone cannot warn you that there is something wrong with a drug that is good for men, good for women and bad for people. Put another way, I'm waiting to see how data-centric DL, be it in neural or symbolic dressing,
3.10.2022 11:43pm - (4/4) will answer simple causal questions, like those described in #Bookofwhy.

3.10.2022 9:34am - (Replying to @JFronczekMD @soboleffspaces and 4 others) The only time you can point estimate P(benefit) from an RCT alone is when monotonicity holds.

3.10.2022 8:30pm - Following discussions with trialists and EBM experts, Scott and I polished our paper on "Personalized Decision Making", and have added to it a section on: Monotonicity, Number Needed to Treat (NNT), and P(harm). Enjoy.

3.10.2022 1:05pm - (1/2) Sorry to disappoint Williamson's enthusiasts. Just read his and must conclude he lives 3 decades behind our time. Quote:"Pearl's mathematisation of causality in terms of conditional probabilistic independence." I renounced this mathematisation in 1992,
3.10.2022 1:05pm - (2/2) and have replaced it with Structural Causal Models (SCM) which, evidently, Williamson is not aware of. Had he known of SCM's he would have formulated his 2007 crave for "mechanism" using the crispier notion of "functions" in SCM.

3.10.2022 12:37pm - Some readers complained about the high Pay Wall on this book. I'll try to fight ACM but, in the meantime, my papers can also be accessed here: and some of the others will soon be linked to you through the backdoor.

3.10.2022 11:56am - Yoseph Haddad's Brilliant Defense Of Israel In The Irish Parliament via @VINNews
3.10.2022 11:56am - I wish I could be half as articulate as my compatriot Yoseph Haddad, and half as knowledgeable as he is about what's really going in the minds of Israeli Arabs.

3.10.2022 12:40am - (Replying to @CharlesMonneron and @johncarlosbaez) Next we will be hearing the slogan "cycle of violence," as we did in the Gaza war, blind to aims and intentions.

3.10.2022 12:18am - (Replying to @johncarlosbaez and @InstMathStat) Agree, the statement is nauseating in its emptiness, and casts scientists as morons.

3.9.2022 10:03am - It's heart-warming to see one "illegitimate" country helping another, while their "legitimate" neighbors are busy seeking destruction.

3.9.2022 9:32am - 3.9.2022 1:44am - (Replying to @offaltube and @mehdirhasan) Hasan may be intelligent enough to cover up his bigotry and inconsistency in splendid acrobatics, but not courageous enough to excise that bigotry. I've found him to be an Israel-eliminationist to the core, openly and obsessively.

3.8.2022 8:26pm - No, No, Mehdi, there is another thing we can all agree on: Calling a neighboring country "illegitimate" is brutal, immoral if not racist. And that applies not only to BDS, but also to eliminationist commentators like @mehdirhasan .

3.8.2022 6:26pm - Readers ask, justifiably, what the lesson of USC has been. Here is the general background and here is my personal summary, as narrated to students and educators:

3.8.2022 5:52pm - (Replying to @NonPert @davidshalev6 and @UofT) Theoretically, nothing is wrong which such benign move, until you listen carefully (in Arabic) to what they obligate their children to do (to the "colonizers") once they take over.

3.8.2022 5:42pm - (Replying to @NonPert @davidshalev6 and @UofT) It's not how they BDS spokespersons describe themselves in English that counts, but what they tell each other in Arabic (intensely monitored in Israel). Putin describes himself as the liberator of Crimea and a "de-Nazifier" of Ukraine.

3.8.2022 4:49pm - 'Antisemitism is Wrong,' Says University of Wisconsin After Three Recent Reports
3.8.2022 4:49pm - Students at the U. of Wisconsin have not learned the lesson of USC: "Every commitment to fighting antisemitism which does not mention antizionism is but a cover up for inaction." @DionJPierre2

3.8.2022 4:32pm - New York Times Correction Gets Close, but Can't Quite Acknowledge Jerusalem as Israel's Capital
3.8.2022 4:32pm - How can readers trust what a newspaper writes about Ukraine, or even NYC, when openly and repeatedly, same newspaper lets wishful thinking suppress truth and bare facts? nly time you can point estimate P(benefit) from an RCT alone is when monotonicity holds.

3.10.2022 8:30pm - Following discussions with trialists and EBM experts, Scott and I polished our paper on "Personalized Decision Making", and have added to it a section on: Monotonicity, Number Needed to Treat (NNT), and P(harm). Enjoy.

3.10.2022 1:05pm - (1/2) Sorry to disappoint Williamson's enthusiasts. Just read his and must conclude he lives 3 decades behind our time. Quote:"Pearl's mathematisation of causality in terms of conditional probabilistic independence." I renounced this mathematisation in 1992,
3.10.2022 1:05pm - (2/2) and have replaced it with Structural Causal Models (SCM) which, evidently, Williamson is not aware of. Had he known of SCM's he would have formulated his 2007 crave for "mechanism" using the crispier notion of "functions" in SCM.

3.10.2022 12:37pm - Some readers complained about the high Pay Wall on this book. I'll try to fight ACM but, in the meantime, my papers can also be accessed here: and some of the others will soon be linked to you through the backdoor.

3.10.2022 11:56am - Yoseph Haddad's Brilliant Defense Of Israel In The Irish Parliament via @VINNews
3.10.2022 11:56am - I wish I could be half as articulate as my compatriot Yoseph Haddad, and half as knowledgeable as he is about what's really going in the minds of Israeli Arabs.

3.10.2022 12:40am - (Replying to @CharlesMonneron and @johncarlosbaez) Next we will be hearing the slogan "cycle of violence," as we did in the Gaza war, blind to aims and intentions.

3.10.2022 12:18am - (Replying to @johncarlosbaez and @InstMathStat) Agree, the statement is nauseating in its emptiness, and casts scientists as morons.

3.8.2022 2:42pm - A new book around the block. ACM has decided to eulogize me (Hey, I am still alive!) with a volume about my works, 1980-2020, here is the TOC: If you want to find all my heretical ideas (+ critics) under one cover, Google scholar tells me it's out now.

3.8.2022 6:2am - (Replying to @JewishonCampus_ @SAFECUNY and @UofT) There is no quote of U of T Provost Cheryl Regehr announcement. The way she phrase her reasons is critical.

3.8.2022 3:52am - (Replying to @soboleffspaces @dylanarmbruste3 and 6 others) Last I heard of Russo and Williamson (2011) they were still stuck on the "probability raising" conception of causation. Change of heart?

3.8.2022 3:44am- Those of you who liked Prof. Darwiche 2 lectures on Causal Inference will surely enjoy the sequel, on counterfactuals: —Part A: —Part B: It would be especially useful for stat and ML folks who have difficulties with the basics.

3.7.2022 10:19pm - (Replying to @papa_cientifico) Well put. This is what happened to me. I just can't understand those who resist this transformation -- my weakness.

3.7.2022 9:58pm - University of Toronto to Withhold Thousands in Graduate Student Union Fees Over Israel Boycott
3.7.2022 9:58pm - I'm mighty proud of my virtual alma mater @UofT (honoris causa 2007) for taking a meaninful step toward a Zionophobia-free campus. I hope the provost explicates the moral imperatives for this step.

3.7.2022 6:15am - Russia publishes an official list of states 'unfriendly' to it
3.7.2022 6:15am - If you are writing a joint paper with a Russian colleague, dont be surprised if it's slowed down, or if you get hacked.

3.7.2022 5:42am - (Replying to @kleintob and @heckmanequation) I commented on it here: and here: It seems that the authors are still trying to avoid the First and Second Laws of Causal Inference, hence that the obstacles listed here will continue to slowdown and misguide their followers.

3.6.2022 10:58pm - (Replying to @laurabialis) They won't move a pebble. Palestinians were given a permanent seat in the Church of the Saintly Victims; nothing they do or say will change their sainthood.

3.6.2022 10:26pm - The doctrine of "de-legitimize your neighbor" breeds strange bed fellows: Putin and Palestine

3.6.2022 10:18pm - To understand Ukraine is to watch the gigantic see of human misery in photo below.

3.6.2022 3:56pm - (Replying to @stephensenn @hemanth_kanak and 5 others) I was tempted to request link to Lubsen Tijjsen but then recalled that, if he/she is not using mathematics of CI, I'd be unable to decipher the method, nor evaluate its promises and limits. So, I'll be waiting for a translator, while insisting on the completeness of Fusion.

3.6.2022 3:43pm - (Replying to @elikesprogramm1) I have hard time separating my hopes from my realistic assessments, both are fairly gloomy, but the latter is truly depressing, ending up with Europe under Russian boots, Asia under China and Iran, and meek US struggling to stay afloat, lamenting how it failed to protect freedom.

3.6.2022 3:12pm - (Replying to @hemanth_kanak @stephensenn and 5 others) It is surely "not as robust as A vs B". But recalling that A vs B in itself is not very robust to the target population of end users, which is vastly different from both A and B, insisting on A vs B, while discarding information from the target population sounds ill advised to me

3.6.2022 1:35pm - (Replying to @JFronczekMD) The limitation of observation studies (eg confounding) can be turn into information-providing advantageous, when we combine them with RCT. Note also that no ignorability-type assumptions are needed for the method to work.

3.6.2022 1:28pm - (Replying to @RaulMachadoG) I am not teaching this year, but I posted here several recommended courses on CI, at UCLA and elsewhere.

3.6.2022 9:30am - (Replying to @alexis_andreyev @SAFECUNY and 3 others) I have been there, in October of 1947, before anyone was kicked, and I have heard Azzam Pasha promising us "Rivers of blood" and Monumental massacre", then 5 armies attempting to implement his promise, all before anyone was kicked out. Still, Biden is talking 2022, not 1948.

3.6.2022 8:56am - (Replying to @prem_k and @SPilleron) Yes. The goal is level-3

3.6.2022 8:54am - (1/2) This article reinforces my theory that the current anti-CI discussion on Twitter is not representative of the entire RCT community. As to whether there is a better way to personalise treatment than to explore heterogeneity in an RCT, my answer is Yes,
3.6.2022 8:54am - (2/2) exploring heterogeneity in some mixture of RCT and observational studies, as shown in What you get from RCT are always population-level decisions, never individual-level decisions.

3.6.2022 8:27am - (Replying to @hemanth_kanak @stephensenn and 5 others) Studying "What we can gain" entails studying "what we can lose". It merely excludes the current attitude of "Let's rally to the defense of our judgment-based culture".

3.6.2022 8:14am - (Replying to @f2harrell @ADAlthousePhD and 5 others) We need to cultivate such researchers, but I doubt we can do it by neglecting asymptotic bias and focusing on p-values.

3.6.2022 8:12am - (Replying to @JFronczekMD @soboleffspaces and 6 others) If you have a c hance to read the article again, you will see that the whole point is NOT treating the two the same. Moreover, the focus is not on how people practice trials today, but on how they should practice tomorrow.

3.6.2022 8:05am - (Replying to @JFronczekMD @soboleffspaces and 6 others) The current practice, of deligating the problem to the designer's an-aided judgment is not less vulnerable to "loss of certainty that the drug will actually work," perhaps even more volnerable.

3.6.2022 7:58am - I pray to see this award inspire Technion's students the way The Technion inspired me 65 yrs ago.

3.6.2022 7:49am - (Replying to @jonnyjkane) If you havn't seen such combination before you should check the prism through which you judge "bias"

3.6.2022 7:44am - (Replying to @stephensenn @soboleffspaces and 5 others) Forget about "random sampling" - you can use your own favorite way of uncovering the confounding involved. The main thing is to ask: what can we gain by deviating from current practices?

3.6.2022 7:33am - (Replying to @alexis_andreyev @SAFECUNY and 3 others) The honest answer to this question is: Israel has now ample evidence showing that this would not end Palestinian's inbred commitment to the destruction of Israel. As Biden said: "I've been around the ME, and am yet to meet ONE Palestinian willing to accept Israel."

3.6.2022 6:58am - (Replying to @f2harrell @soboleffspaces and 5 others) Of course not everything is math. Gamblers acquired extremely powerful intuitions before Bernoulli and De Moivre. But I have hoped to hear more voices say: "Gee, the CRT world, currently in charge of our health, can gain much from the new math we have; it's worth looking into."

3.6.2022 12:31am - (Replying to @michaeldickson and @nickcave) The comparison is in fact informed:

3.6.2022 12:15am - No one has told you that Putin reminds the international community of the Palestinians commitment to the destruction of their neighbors? We are talking 2022 commitment, not 1948.

3.5.2022 10:40pm - (Replying to @blackcat_1iii) eliminationism

3.5.2022 10:23pm - Ukraine or not Ukraine, attacks on Israel must go on

3.5.2022 10:15pm - Israel's Naftali Bennett informs Berlin after meeting with Vladimir Putin
3.5.2022 10:15pm - I do believe that, if there is a person capable of talking Putin into stopping the Ukraine madness it's Naftali Bennett. But I do not believe Putin is stoppable.

3.5.2022 10:05pm - (Replying to @alexis_andreyev @SAFECUNY and 3 others) When Zionophobes say "LOL", we are witnessing the shock of seeing their ugliness in the mirror for the first time, to the light of moral norms.

3.5.2022 9:39pm - (Replying to @soboleffspaces @hemanth_kanak and 5 others) There are still questions about getting the ATE correctly, when the target population is shifting. In CI, we do it through do-calculus. How they do it in EBM, w/o math, is a puzzle to me. But I am open to learning, perhaps they are hiding a secret math that is much superior.

3.5.2022 9:04pm - (Replying to @hemanth_kanak @soboleffspaces and 6 others) This is a very helpful exposition of RCT practices. What is not clear is where judgment ends and mathematics takes over. Absent equations, or reference to technical articles, one may get the impression that all listed requirements can be satisfied by the designer's good judgment.

3.5.2022 8:38pm - (Replying to @stephensenn @ADAlthousePhD and 5 others) Disagree. It is perfectly legitimate for me to inform readers that I have examined a few articles and failed to find equations that speak about efficacy, assumptions, estimands and shifting populations. While I'm not familiar with the field, I'm a fairly good equation detector.

3.5.2022 3:37pm - (Replying to @f2harrell @ADAlthousePhD and 5 others) "Putting off" is a blessing when a complex problem can be broken into two almost independent subproblems, to be solved sequentially. Can't we solve the asymptotic bias first, and then attend to estimation, ie, going from finite samples to distributions? Has RCT solved the former?

3.5.2022 4:55am - (Replying to @SPilleron) Level-3 (counterfactuals) of the Ladder of Causation. #Bookofwhy p. 28. Level-3 contains all the important things we want to know about, that cannot be obtained from RCT's, for example, P(benefit) or P(vaccine was the cause of Joe's death).

3.5.2022 10:30am - (Replying to @stephensenn @ADAlthousePhD and 5 others) The way forward is not to ask what I read or not read, but to communicate how you normally specify (mathematically) (1) a research question, (2) a RCT finding, (3) the assumptions that permit you to go from (2) to (1), and (4) the machinery of doing (3). All it takes is 4 lines.

3.5.2022 10:11am - (Replying to @hemanth_kanak @ADAlthousePhD and 6 others) (Speaking of the "non-just-silly" literature) In what language is the "estimand" specified? Example = ?

3.5.2022 6:37am - (Replying to @ADAlthousePhD @PavlosMsaouel and 5 others) For every procedure you describe, there is a mathematical expression that encodes the asymptotic outcome of that procedure. What I found missing in RCT literature is the research question, and/or the target population.

3.5.2022 6:25am - I know these kids from Odessa. I grew up with them in a spiritual sense. Most of the poets and authors I studies in my childhood came from Odessa: Byalik, Mendele, Achad Haam, Jabotinsky, ... I know these kids, and I pray for their safety.

3.5.2022 6:08am - (Replying to @PavlosMsaouel @GustafRydevik and 5 others) I am pretty sure what statisticians generally mean by random sampling is P(Y,X). P(Y,X|S=outcome of coin) = P(Y,X) Because, in the graph, the coin node is not connected to any other node.

3.5.2022 5:56am - (Replying to @PavlosMsaouel @GustafRydevik and 5 others) I (et al) am interested in all useful research questions that a study can answer, regardless if it is currently in the RCT protocol or deviating from it. BTW, P(Y|X=x, S = random sampling) = P(Y|X=x)

3.5.2022 5:46am - (Replying to @liquidizer) I am disappointed. And terrified to think about the Iran Deal that he is cooking for us in Vienna. Literally terrified.

3.5.2022 5:44am - (Replying to @PavlosMsaouel @GustafRydevik and 5 others) Random sampling gives simply P(X,Y). P(Y|do(X=x), S=random sampling) encodes what we get if we first sample people randomly to the study and then randomly assign them to X=x. "Concurrent control" gives: P(Y|do(X=x), S) where S is "Selected to the study" which may not be randomly.

3.5.2022 5:11am - Just heard Tony Blinken on CNN. My take: (1) The decision to sacrifice Ukraine was made way before the invasion. (2) US has squandered the opportunity of arming Ukraine with massive supply of anti-tank & anti-aircraft weapons as soon as the invasion was deemed imminent.

3.5.2022 4:55am - (Replying to @SPilleron) Level-3 (counterfactuals) of the Ladder of Causation. #Bookofwhy p. 28. Level-3 contains all the important things we want to know about, that cannot be obtained from RCT's, for example, P(benefit) or P(vaccine was the cause of Joe's death).

3.5.2022 4:38am - (Replying to @yudapearl @stephensenn and 4 others) Mr. Hyde: What do you mean "improve"? Concurrent control is the ultimate! Dr. Jekyl: It isn't. Recall, we promised the public P(Y_1-Y_0|target-population). Moreover, they (CI) have a logic to go beyond ACE and ask "What else can we obtain from studies"

3.5.2022 4:27am - (Replying to @stephensenn @ADAlthousePhD and 3 others) Mr. Hyde: Why listen to them? They (CI) don't even know what concurrent control is! Dr. Jekyll: They know, they just have a formal word for it P(Y_1-Y_0|S) which allows them to understand what they obtain from the study, and how to improve on it!

3.4.2022 11:53pm - Fair question. In chp 7 of @Bookofwhy, the goal is estimating ACE, for which RCT provide unbiased estimates. In our paper on personalized decision, the goal is totally different - to estimate P(benefit) for an individual. RCT alone cannot capture that goal b/c it's level 3.

3.4.2022 4:43pm - (Replying to @yudapearl @stephensenn and 2 others) Congratulate me! I passed the test! Just searched #Bookofwhy and found ZERO mention of "random sampling", albeit lots of E[Y|x]. We can now leave all talks of "random sampling" to those who have N objects, equally accessible, awaiting their attention, feeling safe and protected.

3.4.2022 3:57pm - (Replying to @stephensenn @soboleffspaces and @f2harrell) Good point. Accordingly there is no such thing as a random sampling, b/c we do not have the N object sitting there awaiting. It is better to talk in term of the ultimate objective P(y|x) of the procedure, not the procedure itself. e.g., "The mean age E(X)", standard in stat texts

3.4.2022 2:44pm - (Replying to @soboleffspaces @stephensenn and 3 others) Assuming they wish to learn the language of causality. From what I see on tweeter, the trialist community is likely to be the last to do it; which is unfortunate, I have hoped they would all jump in glee upon realizing what they can do with minor tweaking. Luckily, some do jump.

3.4.2022 2:01pm - (Replying to @soboleffspaces @stephensenn and 3 others) The problem is that professional interrogators do not have names for any result that is not estimated by their interrogation, even results that are highly desirable, like P(benefit) - no name.

3.4.2022 2:44pm - (Replying to @soboleffspaces @stephensenn and 3 others) Assuming they wish to learn the language of causality. From what I see on tweeter, the trialist community is likely to be the last to do it; which is unfortunate, I have hoped they would all jump in glee upon realizing what they can do with minor tweaking. Luckily, some do jump.

3.4.2022 2:01pm - (Replying to @soboleffspaces @stephensenn and 3 others) The problem is that professional interrogators do not have names for any result that is not estimated by their interrogation, even results that are highly desirable, like P(benefit) - no name.

3.4.2022 1:45pm - (Replying to @ADAlthousePhD @stephensenn and 3 others) Here you are simply wrong. It is easy to ""randomly select people and assign them…". What you probably mean by "there is no" is that the result will be biased. This is far from "there is no".

3.4.2022 1:37pm - (Replying to @GustafRydevik @stephensenn and 4 others) Yes, Randomization estimates P(Y|do(x)) and random sampling P(Y|x) . These two quantities have life of their own, regardless of whether you estimate them through randomization or other methods.

3.4.2022 1:27pm - (Replying to @soboleffspaces @stephensenn and 3 others) As J. Fisher-Box said (#Bookofwhy): RCT is THE SKILLFUL INTERROGATION OF NATURE: This is indeed the scientific view of RCT, rarely mentioned in statistics: There is such a thing as Nature, and Nature has some properties which need to interrogate. They exist independently.

3.4.2022 1:15pm - (Replying to @stephensenn @ADAlthousePhD and 3 others) The distinction is so fundamental in CI that it even acquired a notational distinction and, once enshrined in notation, it is guaranteed protection from ambiguity, to the point where it need not be mentioned in the text.

3.4.2022 1:05pm - (Replying to @stephensenn @ADAlthousePhD and 3 others) Which word has "precise definition" in two different fields?

3.4.2022 12:45pm - (Replying to @yudapearl @stephensenn and 4 others) P(Y|do(x)) is an intrinsic property of the population which is defined and exists independently of any estimation technique. For example, causal assumptions can replace randomization to obtain unbiased estimate of P(Y|do(x)).

3.4.2022 12:36pm - (1/2) (Replying to @stephensenn @ADAlthousePhD and 3 others) Agree that this represents the world "from within the statistics world" (at least some circles in stat). From AI perspective, it looks a bit differently: "Randomization" is a technique, whose purpose is to estimate a population property P(Y|do(x)) = "effect of action do(X=x)"
3.4.2022 12:36pm - (Replying to @yudapearl @stephensenn and 4 others) P(Y|do(x)) is an intrinsic property of the population which is defined and exists independently of any estimation technique. For example, causal assumptions can replace randomization to obtain unbiased estimate of P(Y|do(x)).

3.4.2022 6:30am - (Replying to @stephensenn @ADAlthousePhD and 3 others) That "randomization is not random sampling" we can see in the equations with zero ambiguity. The less trivial thing we can see in the equation is that results from a non-representative sample, can be repaired as if it came from a representative sample. But this takes math.

3.4.2022 6:14am - (Replying to @ADAlthousePhD @f2harrell 4and 3 others) Consider our paper on "Personalized Decision" as an existence proof of what can be done outside medicine, and as inspirational for what can be modified and transported to medicine at the hands of creative and motivated trialists.

3.4.2022 6:03am - (Replying to @f2harrell @ADAlthousePhD and 3 others) "Runs the risk", of course. At the same time it should not be dismissed apriori is "completely inapplicable" but examined carefully as inspirational for similar methods in medicine, appropriately modified, but leading to truly valuable goal -- personalized medicine.

3.4.2022 4:53am - (Replying to @ADAlthousePhD @stephensenn and 3 others) I am so glad you are no longer angry at me on this count. Now, at the risk of getting you angry again, note that things are different outside medicine. In e-commerce, you can randomly select customers to one ad or another. Same in social experiments. Angry again?

3.4.2022 4:43am - (Replying to @ADAlthousePhD and @soboleffspaces) Agree. This is indeed sufficient for routine work and requires no symbols, just words. It fails however when a new task arrives, e.g., estimate P(individual benefit), that no one has done before, and you need to decide if it is doable with the data sources available.

3.4.2022 4:30am - (Replying to @ADAlthousePhD and @soboleffspaces) This question was answered by one of our trialists: (one way) You go ahead and randomly assign them to A or B, plus ask them about their habits at home. Use the answers obtained in lieu of an observational study. Now, since you are here: A formal definition of comparability = ?

3.4.2022 4:00am - (Replying to @ADAlthousePhD @RWJE_BA and @soboleffspaces) You are underestimating how reliably one can make such generalizations w/o reading every article written, based on one fact: The task involved requires a formal language (counterfactuals) which, based on inquiries and syntactic analysis, is non existent in a given literature.

3.4.2022 3:43am - If you are a member of the University of California Faculty Senate, you should consider signing this letter if you object to using the name of the University or any of its units for religious, political or business activities. I've signed it.

3.4.2022 2:21am - An open letter, from over 100 Nobel laureates, calling for an end to Russia's aggression in Ukraine.

3.4.2022 1:38am - (Replying to @stephensenn @TAH_Sci and 6 others) Please point me to the item that I might have missed.

3.4.2022 1:33am - (Replying to @stephensenn) The hole you treat in your Nature article is only one among several. The hole of mixing benefits and harms to get same ACE is another, which continues to plague us when outcome is binary and deterministic. @soboleffspaces

3.4.2022 1:19am - (Replying to @stephensenn @TAH_Sci and 6 others) Great! Now that we know how to combine studies in a way that respects concurrent control and other holy grails, let's do it for the goal of obtaining information on individual behavior, PNS, a quantity you agree is worth estimating.

3.4.2022 1:13am - (Replying to @kleintob and @heckmanequation) I think this paper creates more confusion than it clarifies. See why:

3.4.2022 12:53am - (Replying to @TAH_Sci @statsepi and 6 others) Can you point to a "fundamental challenge" posed to the idea of combining experimental and observational studies? So far we heard non-fundamental challenges of the type: "This is not how trials are done today". And it was far from "ignored" - a formal explication was requested.

3.4.2022 12:13am - What do you know? The Polish translation of #Bookofwhy will soon become a best seller in Poland. I know my mother would have loved to hold a copy and it show off in her high school, in Kielce, Poland. Unfortunately, she hasn't taught me a word of Polish to enjoy this translation.

3.3.2022 11:22pm - (Replying to @statsepi @raj_mehta and 6 othersq) Explanation = ?

3.3.2022 11:11pm - (1/2) I can't begin to tell you how reassuring it is for me to hear an idea-centered voice rising above the sounds of emotional resistance that have dominated past conversations of our paper. What I hear you saying is: when we've got a good idea, someone will find
3.3.2022 11:11pm - (2/2) a way of putting it into good use, even if it means shifting the foundations of received protocols. Thank you.

3.3.2022 10:43pm - (Replying to @PWGTennant) Recovery from selection bias and establishing external validity are two slightly different tasks, nicely distinguished using causal diagrams, see

3.3.2022 8:31pm - (Replying to @raj_mehta @statsepi and 6 others) In the passage you quoted from Hernan and Robins "exchangeability" is defined as "Ignorability", or "non-confoundedness", or "backdoor admissible". All the latter three hold in RCT, where the question of "comparability" still exists, according to our exchange with trialists.

3.3.2022 5:41pm - (1/2) (Replying to @soboleffspaces) I'm not sure trialist reject CI a priori. I have the sense they are emotional thinking CI folks do not share their intuition about participants in the RCT being different than non-participants. Not only we share it, but we have notation to express it and tools to minimize
3.3.2022 5:54pm - (Replying to @yudapearl and @soboleffspaces) such differences. Perhaps the reason they do not see benefit in making this intuition formal is: (1) they do not have the notational machinery, and/or (2) they do not see the next question we're about to ask: Are the current RCT's protocols optimal? Such questions require math.

3.3.2022 2:49pm - (Replying to @PWGTennant @statsepi and 5 others) We have a Rosetta Stone, called SCM, which tells you, for every (formally defined) relationship in Stat, Epi, Econ, and Cs how it relates to others, and how it is derived from an CSM. What it can't tell you is what hand-waved relationships mean. That's why I am working on CRT's.

3.3.2022 1:26pm - "There is nothing new under the sun" if you are a guardian of professional pride. "There are lots and lots of new things under the sun" if you like to have fun with science.

3.3.2022 1:21pm - (Replying to @ADAlthousePhD @f2harrell and 3 others) You seem very impatient with me. Sorry. I perfectly understand the intuition in your example, but I am asking for a formal criterion that makes A and B more comparable than A and O. Why formal? Because once you put it down formally you would discover some "holes" in thinking.

3.3.2022 1:16pm - (Replying to @raj_mehta @statsepi and 6 others) How is it related to "comparability" in RCT studies?

3.3.2022 1:10pm - (Replying to @f2harrell @ADAlthousePhD and 3 others) So, in your judgement, RCT literature IS GUIDED by a formal framework? I'm willing to accept it, if you say: Yes. So far, I haven't received a Yes from anyone, and I havn't received an answer to the Q: what property is common to A and B that is different to A and O ?

3.3.2022 12:58pm - (1/2) "Professional pride" and "We don't need it" are now the two main obstacles to scientific progress. The former creates the latter as an excuse for stagnation (eg, see economics, where the former is giant). Things get to be really bad when the choice is
3.3.2022 12:58pm - (2/2) not between one notational system or another, but between having or not having a formal framework to work with.

3.3.2022 12:43pm - (Replying to @statsepi @PWGTennant and 5 others) Can't take the time? Unfair! It's been three days now that I am begging RCT experts to tell me what is common to A and B (which they judge to be "comparable") and different to A and O (which they judge to be "incomparable"). Isn't "comparability" part of RCT basic vocabulary?

3.3.2022 12:32pm - I doubt RCT experts would live another decade without a formal framework. Now that an increasing number of researchers are being exposed to CI, they will start noticing "holes" in RCT literature. One such hole is NNT and the distinction between P(benefit) and ACE. See Wikipedia.

3.3.2022 12:07pm - Another causal inference class to learn from. Welcome to the causal revolution.

3.3.2022 11:58am - (Replying to @ADAlthousePhD @stephensenn and 3 others) See footnotes where we describe "selection bias"

3.3.2022 11:56am - (Replying to @ADAlthousePhD @stephensenn and 3 others) Frankly, assuming RCT experts do not have a formal framework to capture their intricate considerations, it would be very wise of them to benefit from the tools developed in CI, or any other formal framework in existence. Why not? Professional pride?

3.3.2022 11:46am - (Replying to @ADAlthousePhD @stephensenn and 3 others) Please do not put words in my mouth. I would never say "my preferred notation". I asked whether there is ANY formal language that trialists use and, having learned there isn't any, I proposed the only one I know about, potential outcomes (augmented with DAGs). Care to switch?

3.3.2022 11:35am - (Replying to @totteh @WvanAmsterdam and 3 others) Sure. The PNS bounds are expressed in terms of P(Y_x) and P(Y_x') , not P(Y_x|S) and P(Y_x'|S), assuming that some smart trialist managed to infer the former from the latter (after reading CI papers). After all, who is interested in effects on the idiosyncratic selected (S).

3.3.2022 7:05am - (Replying to @stephensenn @f2harrell and 2 others) CI is incomplete so long as it does not embrace the intricacies of RCT, as they have been encountered by generations of trialists.

3.3.2022 6:37am - The tension between American and Israeli Jews
3.3.2022 6:37am - Grateful to Eric Mandel for quoting me on the current crisis of Jewish survival.

3.3.2022 6:26am - (Replying to @stephensenn @f2harrell and 2 others) I am trying to learn how RCT's are run with one object in mind -- to improve them. I don't understand how you can say that RCT's are outside the province of CI (the hammer) when they are all about cause and effect and, from what I can tell thus far, have not been formalized yet.

3.3.2022 6:05am - (Replying to @soboleffspaces @WvanAmsterdam and 2 others) I think the node named "RCT" is meant to represent "selection for RCT study"

3.3.2022 5:55am - (Replying to @WvanAmsterdam @stephensenn and 2 others) Nice, but we conduct RCT to find causal effects on some target population. Someone must take the RCT findings and generalize them to the target population. Now, why shouldn't the "generalized RCT" be applicable to those participating in the observational study?

3.3.2022 5:28am - (1/2) (Replying to @stephensenn @f2harrell and 2 others) Here is a link to the new version of our paper on personalized decisions: I am trying desperately to study RCT's, mainly because I know that CI will prove extremely helpful to trialists once their problems are articulated in the language of modern CI.
3.3.2022 5:40am - (Replying to @yudapearl @stephensenn and 3 others) Unfortunately, I have a problem of understanding even the most rudimentary concepts such as "clusters" "non-concurrent" "within study" etc. unless someone translates them to CI vocabulary. Has such translation ever been attempted? Can we start with A,B and O ? as I asked Frank.

3.3.2022 5:05am - A UC academic named Bazian equates Ukraine with Palestine. Dr. Bazian, Ukraine does not seek to eliminate Russia and, before you try to squeeze yourself out of this difference, tell your audience what you personally think, today, about Israel's right to co-exist with its neighbor

3.3.2022 3:51am - (Replying to @stephensenn @f2harrell and @soboleffspaces) We are in 2022. We have powerful notational system that allows us to avoid ambiguities and misunderstanding using symbols. Can we use it as we communicate? Again, it's not a crime to be unfamiliar with it, I'll respect you just as well.

3.3.2022 3:44am - (Replying to @stephensenn @f2harrell and @soboleffspaces) My humble question is: Can we make your question formal? If we don't, we will continue to go in circles over my inability to precisely interpret what "observed outcome rate" is, and, more generally, what is common to A and B that is different in O.

3.3.2022 3:33am - Teachers and students. UCLA is offering 3 courses in causal inference. The one in computer science is taught by Prof. Adnan Darwiche Here is a recording of his first two lectures: —Part A: —Part B: Enjoy.

3.3.2022 2:43am - Fair is Foul and Foul is Fair, and the world is chaotic and problematic and full of hostile devils when one is determined to ignore its beauty and its possibilities. I would rather fight the devils scientifically.

3.3.2022 2:08am - (Replying to @shannonjensen) And I sincerely hope that when you're 85, I'd be around to remind you of how skeptical you were about the longevity of your fighting spirit.

3.3.2022 1:27am - (1/2) (Replying to @stephensenn @f2harrell and @soboleffspaces) It is easy to imagine that A and B could be different from O, but this is not the question asked. 1. We count on O being different, else it would not add information on top of RCT. 2. We seek formal definitions of "comparable" "similar" & "different" to avoid confusion.
3.3.2022 1:41am - (2/2) (Replying to @yudapearl @stephensenn and 2 others) We innocently trying to verify, w/o judgment, whether the stat literature outside CI has developed a formal language to distinguish such nuances, or made them by appealing to healthy intuition. If the former, it can be expressed in one line: P(Q|A) = P(Q|B) =/=P(Q|O) where Q
3.3.2022 1:49am - (3/3) (Replying to @yudapearl @stephensenn and 2 others) where Q is some well-defined causal property. If the latter, it is NOT a crime to say: "No, the literature has not developed the machinery to make this distinction formal." It will not be held against you or your peers, promise.

3.2.2022 9:35pm - (Replying to @sterndavidi and @heckmanequation) Thanks @sterndavidi , I have commented on it yesterday: It would be interesting to see how many NBER members would have the courage to comment on it. I can't, not a member of the exclusive "NBER family".

3.2.2022 9:26pm - (Replying to @soboleffspaces) Who says "this" is what precision medicine seeks to address.? Does the EBM literature ever mention a "crisis"? or "precision medicine"?

3.2.2022 8:52pm - This makes it clear, thanks. And despite our long and lively discussion yesterday with dozens of RCT experts, so adamant against our proposals, I am still no sure if they have recognized the distinction in the past and/or consider it a problem worth thinking about today.

3.2.2022 7:20pm - (Replying to @ArabsForBernie and @JGreenblattADL) What would you call people denied a day of normalcy for 74 years? Victims? Not in my vocabulary. Let's just call them resilient role models to Ukraine and other freedom-loving nations.

3.2.2022 7:07pm - (Replying to @KownUnkowns and @JGreenblattADL) Putin's slogans still work on some people.

3.2.2022 12:43pm - (Replying to @marosenbergCO) Everything is geared toward individual level. Once we understand the mathematics, all kind of creative ideas will come up from enthusiastic trialists, so as to satisfy the conditions specified, and get the quantities needed.

3.2.2022 12:38pm - (Replying to @ADAlthousePhD) BTW, for the sake of future communication, the difference between those assigned and those randomly selected is called "selection bias", and I hope trialists are somewhat familiar with the mathematics of overcoming it, since the end users are patients in the population at large.

3.2.2022 12:33pm - (Replying to @ADAlthousePhD) Great! This kind of creativity is what I've hoped to see from trialists. Moreover, we do not really need group C. Once we select A and B the standard way, we also ask them to fill a questionair about their habits under choice and use the answers to represent observ. study.

3.2.2022 8:22am - (Replying to @f2harrell and @soboleffspaces) I believe I made it very very meaningful by describing how we select the 3 groups, A,B,C, I'm waiting for an explanation on why the groups will NOT be comparable. Repeatedly saying they are not comparable does not make them so. Can you try?

3.2.2022 8:10am - (Replying to @pablogerbas @ashtroid22 and @ADAlthousePhD) Of course they are not "equivalent", else we would not get new information from the observational study. Must catch some sleep now and, by the time I wake up I hope we continue in technical language; the slippery language of past discussions is becoming tiring & unmanageable.

3.2.2022 7:54am - (Replying to @stephensenn @raj_mehta and @SMuellerLab) I have defined "same populations" 3 times. Also, I expect personalized medicine to become widely practiced once practitioners understand the ramifications of the math. And many do, so "Ask your doctor", it may be sooner than thought.

3.2.2022 7:38am - (Replying to @ADAlthousePhD) Now we need the help of some math, to tell us formally what it means to be "actually drawn from 'the same population'". Would you care to try? Or you prefer to keep the discussion at the intuitive level and "prove me wrong" at this rather slippery level?

3.2.2022 7:26am - (Replying to @ADAlthousePhD) On the contrary. I make the distinction clearly between "random sampling" which involves no intervention, and "random allocation to treatments" which does involve intervention. All my math revolves around the distinction between P(y|x) and P(y|do(x)). Try another route.

3.2.2022 7:20am - (Replying to @ADAlthousePhD) Not big deal. Randomly select A B C from those who show up at your clinic, assign A and B to treatment and control, and ask members of C: what did you use this morning, when it was your choice.

3.2.2022 7:14am - (Replying to @totteh @stephensenn and 2 others) This is true for product testing but, in medicine, recruitment is both expensive and sensitive to preferential sampling.

3.2.2022 7:09am - (Replying to @ADAlthousePhD) I would be ready to acknowledge if shown what the difference is. What factor makes them different? Is it the fact that the C group are at home and A+B in the hospital? Naming factors is better that just saying "not the same".

3.2.2022 7:00am - (Replying to @ADAlthousePhD) Eager to hear why "the observational population and trial population are not the same population" What makes them different?

3.2.2022 6:56am - (Replying to @ADAlthousePhD) I confess to have never worked on a trial. Yet, as an outsider, you could do good not merely by lessening burden on patients but also by acquiring a math language of causal inference, so that you can communicate with outsiders like me, who might (very rarely) have a good idea.

3.2.2022 6:45am - (Replying to @ADAlthousePhD) I said it clearly: You select 3 group at random, A, B, and C. Where do you get the "non randomized" from? The nice thing about group C is that you can select 10,000 of them, for little extra cost.

3.2.2022 6:31am - (Replying to @ADAlthousePhD) They are "free" because you dont need to call them to the hospital and make sure they take the assigned dose. You call them and ask them: "what dose did you take this week?" or "what drug did you prefer this week?"

3.2.2022 6:25am - (Replying to @ADAlthousePhD) Sure. You randomize them to C by calling random numbers from the phone book (kidding).

3.2.2022 6:22am - (Replying to @ADAlthousePhD) I am not looking at it that way. You get 200 that are randomized, fine, but the C group is free, so you can survey 1,000 of them. Beside, what's sacrificing 100 sample for getting such precious information as individual response.

3.2.2022 6:06am - (Replying to @ADAlthousePhD) If you are referring to "selection bias" (or non-representative samples) it's a different issue, that can be handled separately, but bottom line, eventually we get some estimate of treatment efficacy, else what is the purpose of the RCT?

3.2.2022 5:52am - (1/3) For readers having difficulties with the notion of conducting experimental and observational studies "on the same population". Here is a simple description: We select at random three groups of patients, A, B and C from the same populations. Say 1000 samples in each.
3.2.2022 5:52am - (2/3) We now assign A to treatment, B to placebo, and submit C to non-interventional survey, asking each patient: what dosage of drug you take daily? how is your blood pressure, etc etc. Our analysis takes the results of the three studies, and tells us
3.2.2022 5:52am - (3/3) quantitatively how a randomly chosen patient (drawn from the same population) would react to BOTH treatment vs. denial of treatment. A typical assessment: "This individual has 40% chance of benefitting from the treatment and 10% chance of being harmed by it". No complications

3.2.2022 5:05am - (Replying to @ADAlthousePhD) Honestly, I do not understand. I count on you to articulate those point in a layman language. Please try.

3.2.2022 5:03am - (Replying to @ADAlthousePhD) Perhaps you can explain to me the very real issues conveyed in: "That random sampling takes place is demonstrably false." What is it? Can't we draw three groups at random, A, B, C, without being dismissed as: "demonstrably false."? Eager to grasp the "very real issues".

3.2.2022 4:08am - (Replying to @dnunan79) Partnering is always welcome. Note though that EBM fears of CI originate from two sources: (1) CI relies on assumptions which EBM object to apriori. (2) CI talk counterfactuals, which is scary. Personalized decision has none of (1), but EBM folks can't get it, (2) = Swahili

3.2.2022 3:39am - Two scientists are looking at an equation. One says: I havn't seen this phenomenon before, there must be some possibilities here, let's explore. The second says: I havn't see this phenomenon before, there must be something wrong here, let's dismiss.

3.2.2022 2:43am - Fair is Foul and Foul is Fair, and the world is chaotic and problematic and full of hostile devils when one is determined to ignore its beauty and its possibilities. I would rather fight the devils scientifically.

3.2.2022 1:10am - (1/n) (Replying to @stephensenn @raj_mehta and @SMuellerLab) I agree with your conclusion: "Numbers Needed to Treat [as usually estimated] cannot be used to assess the proportion of responders in clinical trials." But I do not understand the way you go about showing it. For me, "responders" is counterfactual notion, whereas "trial" is
3.2.2022 1:15am - (2/n) (Replying to @yudapearl @stephensenn and 2 others) an interventional notion (level-2). Done! No further analysis is needed to show their incompatibility, except if you wish to ask: what information does the former provide about the latter. This requires a formal language for counterfactuals which I havn't seen in your paper.
3.2.2022 1:25am - (3/n) (Replying to @yudapearl @stephensenn and 2 others) This paper answers the question formally and I have hoped you would be excited to help turn its findings into personalized medicine practice. Instead, you refer me to papers that lack formal treatment, which you hope would convince me to give up.
3.2.2022 1:32am - (3/n) (Replying to @yudapearl @stephensenn and 2 others) To give up I'd need to see formally why my conclusions are wrong, or why they face practical difficulties of implementation. I havn't seen it. "Same population" is not a "fallacy" but simply explained as follows: We select at random three groups of patients, A, B and C.
3.2.2022 1:42am - (4/n) (Replying to @yudapearl @stephensenn and 2 others) We now assign A to treatment, B to placebo, and submit C to non-interventional observations: what dosage of drug you take daily? how is your blood pressure, etc etc. Mathematics takes the results of the three groups, and tell us something on how a randomly chosen individual
3.2.2022 2:00am - (5/n) (Replying to @yudapearl @stephensenn and 2 others) would react to treatment and (the same individual) to denial of treatment. I think this finding calls for some informed curiosity, perhaps even a sweeping enthusiasm. I am always open to examine criticism, once it is cast in the language of counterfactuals, not probabilities.

3.1.2022 4:32pm - (1/2) Glad we have Zelensky's own words on record, to reaffirm the unmistaken parallels between Israel and Ukraine. The moment I heard Putin declare: "Ukraine is not a real nation" I heard him echo Omar Barghouti (BDS-cofounder) declaring: "they [Israelis] are not a people",
3.1.2022 4:32pm - to the cheerful welcome of my tenured colleagues at the UCLA's history department, in January of 2014. Putin's/BDS's audacity to define their neighbors as they please cannot be missed by any serious observer of the Ukraine invasion and the Middle East tragedy.

3.1.2022 3:15pm - He completely misses 3 decades of causal analysis, according to which "Bayesian analysis" is no longer an acceptable replacement of causal models. Evidently he, and so many well meaning statisticians, have not read why they should be only half-Bayesians:

3.1.2022 12:15pm - (Replying to @JGreenblattADL) I beg to disagree. Putin's attack of a neighboring country is appropriately reminiscent of the 1948 attack of Israel by its 4 neighbors + Iraq. Putin's declaration about Ukraine having no right to exist is appropriately reminiscent of Khomeini+BDS declarations about Israel.

3.1.2022 11:46am - (Replying to @PavlosMsaouel @StableMarkets and 5 others) Believe me, I've tried, but I lack the language, my weakness. I thought you could help me with ONE research question; not tons, just one.

3.1.2022 11:12am - (Replying to @sbmisi) I am not looking for recognition of novelty. I am looking for open-mindedness which I only find among those who appreciate the critical role of new notation, and the importance of "rethinking" old problems in new light. Lacking notation is not "doing everything wrong".

3.1.2022 5:19am - (Replying to @AnoopRKulkarni @casualinfer and 3 others) Wait! Don't you want to hear about PO and do-calculus? Fascinating topic!

3.1.2022 3:06am - (Replying to @raj_mehta @stephensenn and 4 others) Note, Altman is treating NNT(benefit) and NNT(harm) separately, assuming monotonicity in each case. But he does not consider a population as being a mixture of two types of individuals, some "benefitting" and some "harmed". See Wikipedia:

3.1.2022 2:53am - (Replying to @raj_mehta @stephensenn and 4 others) I, too, suspect that Altman would have agreed with me. He seems to be open minded, open even to the realization that he DID NOT note the counterfactual, unobserved nature of each "benefit" and "harm". Namely, his results are valid ONLY if treatment can never harm anyone.

3.1.2022 2:09am - "Why Putin already lost the war?" My ultimate take from this article by Harari is: "Nations are ultimately built on stories"

2.28.2022 9:29pm - It's hard to tell how many economists fall for this narrative which, on one hand, introduces new notation (semantically equivalent to do-operator) and, on the other hand, does it in the name of traditional (pre-Cowles) econometric methodology that demanded no new notation.
2.28.2022 9:29pm - Regardless of its appeal to econometric students, the narrative should be read in conjunction with an alternative narrative of Haavelmo's contributions, and an awareness of the cost of giving up modern techniques such as these:

2.28.2022 6:00pm - Elementary, gentle and brief introduction to DAGs for instructors and newcomers, by Sander Greenland.

2.28.2022 4:21pm - @nberpubs , Can one comment on the recent NBER posting of Heckman and Pinto? or are comments restricted to "members of the NBER family only," thus isolating NBER readers from the rest of the world? Any NBER member among our 60K readers willing to help me pass through this wall?

2.28.2022 4:02pm - (Replying to @StableMarkets @RWJE_BA and 5 others) I can't see how you can "violate ignorability," or suffer from misspecification, if the ID assumptions hold. Consequently I can't see how nonparametric Bayes can help. Example please...

2.28.2022 1:06pm - (Replying to @mogens_fosgerau @nberpubs and @PaulaAureo) Thanks for posting this new paper by Heckman and Pinto. It seems that the authors are still trying to avoid the First and Second Laws of Causal Inference, hence that the obstacles listed here will continue to slowdown and misguide their followers.

2.28.2022 7:21am - Just got a recording of my podcast with @EpiEllie and @LucyStats on AI, Causal Inference and mainly on personalized medicine. Enjoy.
2.28.2022 7:21am - You can tell that I was heavily influenced by

2.28.2022 3:09am - I think I'm getting old. Things that I wrote two years ago sound so fresh and truthful to me today that I begin to doubt my authorship, though not their truth.

2.28.2022 12:47am - (Replying to @Daroff and @Conf_of_Pres) I am honored by your offer, please email me on I also volunteer to testify at the next meeting of @Conf_of_Pres on why the crisis is so urgent, and why we, at the trenches, see the leadership of @Conf_of_Pres to be irreplaceable, regardless of scopes and roles.

2.28.2022 12:31am - This version is slightly updated: and addresses a wider audience.

2.27.2022 11:30pm - Glad you asked, because my papers on the impossibility of model-free data science: are almost 2 years old and, in Twitter environment, anything older that 24 hours is either skipped or forgotten.

2.27.2022 11:09pm - (Replying to @PavlosMsaouel @ESMDcan123 and 3 others) I love this attempt to cohabitate methods developed for RCTs design with those developed in the graphical-models literature. Question: What is the most valuable method, tool or idea that the latter should learn from the former?

2.27.2022 10:38pm - In my humble opinion, this is an opportunity for @Conf_of_Pres to extend the "outside" into "inside", recognize that Jewish education on college campuses is already 60% hijacked, and that no other leadership is capable of preventing a total collapse in 3-5 yrs @YKleinHalevi

2.27.2022 2:04pm - (Replying to @ninoristeski and @vardi) Self-learning is great, if you start with a model of the world, not with a model-free data.

2.27.2022 3:33am - (Replying to @HuelleAlexander) "Anachnu Po" is the Hebrew translation of "mir seynen du". I have had the honor of singing it with Theodor Bickel, he did it in Yiddish and I in Hebrew. Here he is: "Mir Seynen du." I wonder if President Zelensky ever imagined that he would inspire us to remember those partisans.

2.27.2022 2:41am - (Replying to @schraubd and @DavidHirsh) I was not referring to the Nazarian Center, and you are minimizing the danger of appointing a person viewing Israel as a liability to head up a Center to which Israel need be an inspiration. There is such a thing as "fit"; we dont appoint a creationist to head a dpt of evolution.

2.27.2022 1:32am - (Replying to @GeraldNGOM @DavidHirsh and @schraubd) Hijackers of Jewish education truly do not understand how central Israel is to most donors, and to most Jews who haven't bowed to social acceptance. Perhaps the analogy of hiring a Kahanist to head a Center of Palestine Studies would tickle some understanding of that centrality.)

2.27.2022 12:24am - When Ukrainian President Zelensky Said, “Listen, I Am Here,” He Made the Ultimate Jewish Statement via @jewishjournal
2.27.2022 12:24am - Inspiring piece by David Suissa, reminding me of another "Hinneni" -- the marching song of the Jewish partisans in WWII: "We are here" -- "U'Mitsadenu Od Yarim Anachnu Po" Anachnu Po!

2.26.2022 10:52pm - (Replying to @GeraldNGOM @AnneHerzberg14 and 4 others) True, but I would rather think of remedial actions. The reason I mentioned @Conf_of_Pres is that every Jewish Federation has a "community fund" that advises innocent donors where their wishes can best be realized. If we educate this fund, chances are even EDI gates will open.

2.26.2022 10:27pm - I would expect a responsible organization such as @Conf_of_Pres to compile a credible "Annual Guide to the Informed Philanthropist". Are they awake? Anyone with a clout to alert them to what's going on on US Campuses? @ADL ?? @AJC ?? @AEN ? I was almost going to type @HillelIntl

2.26.2022 8:55pm - Are these the same Ben and Jerry who pulled out their ice cream from another country who defended itself against indiscriminate rockets?

2.26.2022 8:46pm - Only a "professors uprising", like the one we have witnessed at USC can stop EDI's betrayal of public trust. To commemorate dim memories, EDI = Equity, Diversity and Inclusion.

2.26.2022 8:33pm - My advice to philanthropists who contemplate donations to "Jewish Studies" or "Israel Studies" programs: Please take a look first at what happened to these programs at UCLA and USC. I'll be happy to take you on a guided tour. @GeraldNGOM @StandWithUs

2.26.2022 8:08pm - Since no one volunteered to correct Wikipedia's "Number Needed to Treat", @smueller just did it. See I hope authors who bemoaned the improper use of NNT will find its proper use refreshing, unassailable and promising. @miquelporta @f2harrell @suiss70

2.26.2022 3:43am - (Replying to @desai_pratik) We need to ask, thanks for the alert.

2.25.2022 9:48pm - I do not know why, but they sent me another video of the lecture I gave today. This one is a "cloud recording" Enjoy.

2.25.2022 8:10pm - To all teachers, students and fans of Primer, unusually good news from Wiley: "Corrected copies of the book are now available." I just got a physical copy in the mail. Warning: Watch for the statement "Reprinted with Revisions 2021" on page iv. Enjoy!

2.25.2022 5:47pm - Oh, great! Now I can watch myself through the eyes of the audience, and enjoy a discussion on "Deep Understanding". Thanks.

2.25.2022 12:10am - (Replying to @stephensenn) We make no assumption of such similarity. On the contrary, the more dissimilar studies are the more information we get from the observational study; confounding works for us because it carries information on how different units are from each other.

2.24.2022 11:50pm - Helplessness is when all you can do for the 44 million people of Ukraine is retweet this emotional statement by their leader. He will be remembered as the one who didn't abandon ship.

2.24.2022 6:59pm - (Replying to @artistexyz and @stephensenn) It's a theorem in SCM, since Y, Y_1,Y_2 are each independently defined by operations on the model.

2.24.2022 6:05pm - (Replying to @mendel_random @stephensenn and 6 others) I heard it said about science: Ah if only science could answer the important questions of life... It can't, therefore, back to astrology...

2.24.2022 5:01pm - (Replying to @stephensenn) The only assumption we are using is "consistency", which Rubin writes as Y = X(Y_1) + (1-X)(Y_0) without which nothing can be inferred in CI. Can you express "study effect" formally? Does it violate consistency? Isn't it what is sometimes called "placebo effect"?

2.24.2022 4:35pm - (Replying to @PavlosMsaouel @AndersHuitfeldt and 3 others) The do-operator is more than just a "tool". The do-expression distinguishes Level 2 from Level 3 of the reasoning hierarchy, namely between questions that can be answered by RCT and those that can't. Same with a "causal model" (not DAG) - nothing can be inferred without it.

2.24.2022 4:13pm - (Replying to @PavlosMsaouel @AndersHuitfeldt and 3 others) My My, I felt right away that this terminology comes from a foreign source - authors who never used the do-operator, perhaps not even a causal model. I think our paper gives more compelling reasons why P(benefit) should be of concern to both policy makers and individuals.

2.24.2022 2:48pm - (Replying to @stephensenn @RWJE_BA and 5 others) We have hoped all such questions will be settled in the revised, conceptually compelling version If we failed, please point to the equation you have doubts about.

2.24.2022 2:40pm - Here is a youtube announcement for regestration

2.24.2022 12:59pm - No. @amnesty has not begun to "address" the situation in Ukraine. All they have done is tapped themselves in the shoulder for being "humanitarian", but have not attached a single negative label to Putin, none comparable to the slanderous labels they splashed on Israel.

2.24.2022 12:41pm - (Replying to @soboleffspaces @AndrewPGrieve and 2 others) The EBM establishment pushes a false construct for anyone interested in more than just CACE. But we can know with certainty the proportion of people benefiting, once we have combinations of RCT and nonexperimental data.

2.24.2022 12:29pm - Hats off to the entire US delegation @RepBarbaraLee @RepRoKhanna @RepAdamSchiff @RepTedDeutch @USRepKeating @RepSwalwell @RepAndyKimNJ currently harassed by pro-BDS organizations and other conflict-prolongers, emboldened by @Amnesty 's vilification of Israel.

2.24.2022 11:08am - (1/2) Sure. NNT = 1/P(benefit) is the correct interpretation of all the verbal definitions of NNT that I could find. It must be distinguished from NNT_w = 1/ACE which is a (wrong) numeric interpretation used in all the calculations that I could find. NNT_w=NNT under monotonicity,
2.24.2022 11:08am - namely when treatment can cause no harm to any individual. Otherwise, based on RCT alone, we have the bound: 1/CACE < NNT< infinite Anyone volunteering to correct Wikipedia and other Encyclopedias? @AndrewPGrieve @suiss70 @miquelporta

2.24.2022 6:10am - Bernard Henry Levy on "The Rape of Ukraine"

2.24.2022 2:12am - (Replying to @RWerpachowski) We cannot stop Putin's tanks, but perhaps we can stop Guterres from assuming that he has any credibility left.

2.24.2022 2:07am - Of course Putin thinks none of humanity. But Guterres finds it necessary to project humanity in the name of an organization that has lost all credibility to speak on the side of humanity. He acts as if unaware of the UN credibility crisis.

2.24.2022 1:24am - Guterres to Putin: 'In the name of humanity stop this war against Ukraine'
2.24.2022 1:24am - I am not sure Guterres has the needed moral rectitude, when speaking for an organization that never told Iran: "In the name of humanity stop this war against Israel"

2.24.2022 12:54am - (Replying to @AndersHuitfeldt @AndrewPGrieve and 2 others) My understanding is that none of the papers written on NNT captures the counterfactual nature of the word "prevent". How can I tell? Easy. You can't articulate this notion in the language of RCT's and, as we show, you can't estimate it from RCT data alone. Time to correct!

2.24.2022 12:02am - (Replying to @AndersHuitfeldt @AndrewPGrieve and 2 others) The point of our paper is that even RCT-based NNT does not capture the verbal definition "The average number of patients who need to be treated to prevent one additional bad outcome" (Wiki). "Prevent" means to flip an otherwise bad outcome into a good one.

2.23.2022 11:19pm - Readers ask: How is P(benefit) related to NNT (Number Needed for Treatment). Ans. The former is correctly (ie, counterfactually) defined, the latter is not. I ask: What would it take to correct a literature that is 3-decade wrong? As noted by @AndrewPGrieve @suiss70 @miquelporta

2.23.2022 9:33pm - I was assured by the organizers at UCI that all who try to register will be accommodated. They also mentioned some type of recording.

2.23.2022 5:18pm - Upon a 2nd reading, I am very happy to my comparison on Quora, which even economists, so I believe, have found to be unassailable.

2.23.2022 5:10pm - (Replying to @AleksYankov) If I heard him correctly, Putin claimed that Ukraine itself is an artificial creation and has no right to exist.

2.23.2022 4:30pm - Not immediately. They will first give the Russian delegate a chance to explain why certain countries do not have the right to exist.

2.23.2022 2:48pm - Thanks for posting, @DKedmey . I didn't know this interview with Marvin Minsky exists; it's full of great ideas, some new to me. I had one serious conversation with Marvin in 2011, on free will. He was astonished that anyone takes free will seriously in the age of computers...

2.23.2022 1:07pm - Scott and I decided to make "personalized" decision making understandable to all, so we revised our paper to make it both conceptually compelling and numerically demonstrable. I hope we succeeded. "Speak to your doctor!"

2.23.2022 12:57pm - A recent announcement by the American Mathematical Society calling for boycott of a scientific conference in Russia, raising thoughts whether this would be an appropriate/effective step by scientific societies.

2.23.2022 4:38am - (Replying to @nelkhadiri) Your options are (1) submit it to a do-calculus engine that will tell you if it can be estimated at all, and is so how, or (2) if the answer is NO, you resort to parametric assumptions, or ask the engine for auxiliary experiments you can run, or new variables you need to measure.

2.23.2022 12:05am - Nancy Pelosi repeats her keen observation that "The greatest political achievement of the 20th century was the establishment of the State of Israel": All she needs to do now is convince Mahmoud Abbas and a few Congresswomen and peace will descend on us.

2.22.2022 10:57pm - Sharing a heartwarming congratulation from my alma matter, the Technion It reminded me of my student days 60 yrs ago, and of my debts to all the excellent professors who stirred me into science. One notable change: In my time, male/female ratio was 30:1

2.22.2022 10:21pm - (Replying to @joe_sill and @PHuenermund) Your logic seems solid, but I am not sure what "other notions of confounding" exist that are not erroneous.

2.22.2022 5:20pm - (Replying to @soboleffspaces) This is not a trivial question which even the best regression textbooks fail to fully address.

2.22.2022 5:11pm - The #Bookofwhy goes through several such oversights. The most common one is enshrined in the word "malaria" (Mal Air) The less known is the death of Captain Scott at the South Pole, attributed to scurvy which, despite James Lind's observations was misunderstood till the 1930's.

2.22.2022 4:32pm - Belated congratulations to our colleagues at UCI, recently elected Fellows of the American Academy for the Advancement of Science (AAAS): "Automated Reasoning" and "Machine Learning" seem to be leading the show.

2.22.2022 3:58pm - (1/2) I think we made more enemies with this paragraph than any other we ever authored. Glad you found it truthful. "Assumptions are self destructive in their honesty. The more explicit the assumption, the more criticism it invites, for it tends to trigger a richer space of
2.22.2022 3:58pm - (2/2) alternative scenarios in which the assumption may fail. Researchers prefer therefore to declare threats in public and make assumptions in private."

2.22.2022 1:03am - Thanks for posting this interesting paper on "explanatory learning". I'm not sure though how this kind of explanation covers "causal explanations". The former seem to capture conditions that justify a "label" in prediction, rather than an event causally responsible for another.

2.22.2022 8:41am - (Replying to @noranta4 and @DavidDeutschOxf) The answer to "why I made this prediction" is different from "why an event happened". I predicted Rain because the barometer fell, yet the barometer falling is not why there will be rain.

2.22.2022 12:31am - `We Can't Shut Off One of Our Identities': Jewish SUNY Students Decry 'Exclusion' From Sexual Assault Awareness Group Over Zionism
2.22.2022 12:31am - Moral bankruptcy starts with clubs; today it's a separate club for Zionophobes, tomorrow its separate club for KKK. President Christian erred by calling Zioinism a "viewpoint", not "identity," and Jewish groups erred by decrying "antisemitism" rather than Zionophobia. @SAFECUNY

2.21.2022 10:07pm - It's now public - no return! I'll be speaking on Friday at a UC-Irvine Seminar, on: "The Science of Causal and Effect: From Deep Learning to Deep Understanding". For details, see I love the title - so much to say there; especially on "deep understanding".

2.21.2022 9:50pm - Thank you for offering @JVGIndyVoter . Our Foundation webpage is, which summarizes our activities and leads you to the donate button, or, here is the donate page directly

2.21.2022 9:20pm - That was an immense event, which I am watching now for the first time, showing what one high school kid can do for his high school peers. It makes me think of my neighborhood school and ask: What are they doing to make grades 6-12 part of "the army of change"?

2.21.2022 8:50pm - (Replying to @banazir @UncertaintyInAI and 3 others) Thanks for showing it.

2.21.2022 6:04pm - (Replying to @soboleffspaces and @ProfLaurenRoss) This is very interesting. Please preserve this issue of Nature.

2.21.2022 8:00am - In memory of his humane spirit, Daniel Pearl Foundation donates $20,000 to Journalists for Human Rights' Afghanist...
2.21.2022 8:00am - "To save hundreds of journalists from a similar fate, and give those journalists new life, new hopes, and freedom to pursue truth and report it."

2.21.2022 7:10am - (1/2) It was on Feb. 21, 2002, 20 yrs ago, that a team of FBI agents knocked on the door and asked us to sit down. "Bad?" we asked - "Yes". "Are you sure?" - "Yes". The only thing I remember next was lifting my hand up and down, up and down, marveling how human biology can continue
2.21.2022 7:10am - (2/2) to work, when everything else is dead. Up and down,... Up and down,... when everything is dead -- brain, heart, self, ...-- all frozen dead.

2.21.2022 1:42am - (Replying to @TomD_Santiago @ylecun and @TMoldwin) We now have a Turing test of "having a very clear notion of whatever " = teach it to a dumb robot.

2.21.2022 1:21am - Thanks God, the Causal Revolution has been spared this fait.

2.21.2022 1:13am - Why do you have to lie? We have been asking this question to BDS cronies since they came to town. Answer: truth and lies are one big soup of weaponry for a Zionophobe. @amnesty , too, has found the soup tasty and profitable.

2.21.2022 12:21am - (Replying to @Perperuna2) Once you postulate a physical mechanism, you need to encode your postulate in some symbolic form, and make it interpretable and usable to a robot. This is, in a nutshell what causal inference (or SCM) is all about.

2.20.2022 2:27pm - (Replying to @HillelNeuer @hrw and @KenRoth) Where do they get their education?

2.20.2022 1:26pm - (Replying to @IamManuell @danintheory and @ylecun) The argument turned from "discussed" to a "nightmare" in computer science, where you must prevent a robot from attempting to increase the mass m by increasing the force f, misguided by m=f/a. It's hard to appreciate how much we take for granted, till we try it on a stupid robot.

2.20.2022 10:28am - (Replying to @SPilleron) You'll never be the same.

2.20.2022 4:06am - (Replying to @GaelVaroquaux @ylecun and @TMoldwin) In SCM, causality reigns w/o intervention; the latter is tangential to the former. The essential relationship is "listening", not intervening.

2.20.2022 1:04am - (Replying to @yudapearl @dePolavieja and 2 others) It is the "listening" relationship that introduces causal asymmetry to SCM, not agency. It is the "listening" relationship that is missing from classical physics as well as QM.

2.20.2022 12:37am - (Replying to @dePolavieja @behaviOrganisms and @TMoldwin) To clarify, in my mathematical language (SCM), it is Nature that is doing its thing, not an agent. The First Law deals with a society of listeners, not actors, and uses the notation "Y_x = y" to mean: Y would be y had X been x.

2.20.2022 12:29am - (1/2) With the tension in Ukraine mounting, you probably can't wait to find out what ever happened to the "faculty uprising" at USC. This article contains a fairly comprehensive analysis of the general issues involved: It led to a petition demanding equal
2.20.2022 12:29am - (2/2) protection (from harassments) for Zionist students and will probably be discussed at the next meeting of USC's new "Advisory Committee on Jewish Life". Let's hope. @AnneHerzberg14

2.19.2022 11:07pm - (Replying to @drenerbas @rravi and 2 others) I refrain from discussing entropy, especially in information theory, because it is a probabilistic notion, which I am trying to separate from causal notions. I mentioned it once in #Bookofwhy, as an example of irreversible emergent property that holds in the macro world only.

2.19.2022 10:59pm - (Replying to @Perperuna2) I used "physicist" merely to distinguish it from "physics" which is prone to evoke all kind of speculations on the nature of the universe, time-space curvature and multiple worlds which are intriguing in themselves, but have nothing to do with the way toddlers kick balls.

2.19.2022 5:34pm - Regardless of what you think about #Ukraine or #Iran I find myself wondering: How did they use to make leaders when I was young?

2.19.2022 4:33pm - (Replying to @Monomorphic @skdh and 19 others) Yet, if asked, every "average person" that I know would swear he/she has the option of touching or not touching his/her nose when, in fact, touching/no-touching was already predetermined by the state of that person's software.

2.19.2022 12:07pm - (1/2) You don't need to go to CPT to argue that "deep down" there is no such thing as causality. Newtonian mechanics is also governed by algebraic equalities, so it is symmetric, telling us that f=ma is as valid as a=f/m. Where causality comes in is when we venture to model the
2.19.2022 12:07pm - (2/3) physicist, rather than physics. Here, we find an organism that perceives the world as a society of cause-effect relationships, in which the equality sign stands for an assignment operator a:=f/m. At this point we need decide who our users are, physics or physicists; namely,
2.19.2022 12:07pm - (3/3) with whom we wish to interface, who is the consumer of our reasoning apparatus, hence, in what language we ought to cast our explanations. The rest is causation, as in

2.19.2022 5:47am - (1/2) Philosophers have been quite reluctant to abandon Lewis metaphysical definition of counterfactuals in favor of the structural semantics of SCM. This paper is different: It attempts to define counterfactuals and actions with disjunctive antecedents eg,
2.19.2022 5:47am - (2/2) eg, "do (A or B)" as in "Paint the wall red or blue". I've discussed the problem in, and noted that extra information is needed. It's still not clear to me if Rosella&Sprenger paper agrees with my assessment, or how to compute E[Y|do(red or blue)]

2.19.2022 - 4:59am - (1/2) What makes Yossi Beilin attempt to revive the stale idea of "confederation) that practically everyone in the Israeli peace camp has given up long ago (see why My theory, he is speaking to American audience who is refusing to internalize the fact that
2.19.2022 - 4:59am - Palestinian eliminationism is DNA deep and is making super effort to maintain the appearance of "peace-aiming discussions" while ignoring the giant elephant in the room. Worth pointing to, next time Beilin speaks at your university. @EinatWilf @CotlerWunsh @NGOmonitor

2.18.2022 9:08pm - The root of this Western mentality must be studied by top behavioral pathologists. Is it gullible siding with perceived underdogs? Is it resistance to seeing a successful, democratic nation state? Is it resistance to seeing Jews in control of their destiny? Citation anyone?

2.18.2022 8:31pm - The root of this Western mentality must be studied by top behavioral pathologists. Is it gullible siding with perceived underdogs? Is it resistance to seeing a successful, democratic nation state? Is it resistance to seeing Jews in control of their destiny? Citation anyone?

2.18.2022 12:05am - (1/2) (Replying to @patrick_baud @SSS_music and 18 others) This is your Robot speaking. Yes, from everything I know about my making, I was compelled by my program to speak as I spoke. But, miraculously, I did not feel compelled, because I have a strange filter programmed into me that labels some actions "compelled" and others "free".
2.18.2022 12:16am - (2/3) (Replying to @yudapearl @patrick_baud and 19 others) The algorithm that governs the labeling takes into account all kind of inputs, eg, whether other options were consideted, whether consequences was considered, what assumptions were considered in the deliberation, etc. Once we discover the labeling algorithm, the ancient
2.18.2022 12:19am - (3/3) (Replying to @yudapearl @patrick_baud and 19 others) "scandal of free will" will be resolved, and we would be able to address questions such as "responsibility" "punishment" and "education".

2.17.2022 11:41pm - University Lecture by Israeli Chemist Latest Target of Antisemitic 'Zoombombing' at Campus Events
2.17.2022 11:41pm - "Zoombombing", a nasty new phenomenon, soon to hit your university! Reason: failure of all players to name the new virus. They keep calling it "antisemitic" (i.e., "look how hated we are") instead of "Zionophobic" (i.e., "look how despicable our haters are"). A Jewish syndrome.

2.17.2022 10:46pm - Time for self-congratulation! Highest number of @SloanFoudation awards go to my UCLA colleagues. We must be doing something right, despite my bickering.

2.17.2022 10:23pm - (1/2) (Replying to @Malinasaval @LAPressClub and @Variety) Mazal Tov, Malina, on winning the @LAPressClub award. Your article unveiled to me an experience I never had - growing up Jewish in America. Though I have been here 62 yrs, I was always shielded from this experience by my earlier 23 years, where I was a "normal Jew", mentored
2.17.2022 10:33pm - (Replying to @yudapearl @Malinasaval and 2 others) by Biblical figures and Ben Gurion, more than by Hollywood and its Woody Alans. What Hollywood needs now is an authentic movie about the period 1917-1948 in Israel, when the concept of a "normal Jew" was forged, put into test, and emerged.

2.17.2022 9:51pm - If this parent had children at UCLA he would have exploded twice over.

2.17.2022 4:11pm - Hamas must be listening to my academic colleagues at UCLA. Where else could they find such juicy "occupation-aggression-oppression" jargon.

2.17.2022 2:08am - (Replying to @irinarish @ilyasut and 2 others) I don't know much about LLM, but I know that this sentence "statistics do amount to understanding" was proven wrong mathematically if one insists on "understanding" to embrace "answering counterfactual question" (Rung 3).

2.17.2022 2:01am - (Replying to @ArletOttens @DavidDeutschOxf and 15 others) Still, the definition we choose for "free will" can't be totally arbitrary, divorced from our intuitive understanding of what fw is. We can't define say:" free will = playing tic-tac-toe". So what is it?

2.17.2022 1:54am - (Replying to @WiringTheBrain @BobbyAzarian and 15 others) Do you mean "all is not pre-determined" in the human mind? or "all is not pre-determined" in the software of a man-made robot?

2.17.2022 1:44am - (Replying to @DavidDeutschOxf @BobbyAzarian and 15 others) I fully agree. An AGI, if successful, would be a decisive proof that there is no contradiction between "All is pre-determined" and "Permission is always granted". But do we really need to wait for full AGI? Can't we formulate the principles by which it can be done? We should.

2.16.2022 10:36pm - (Replying to @BobbyAzarian @mpigliucci and 15 others) Related, and somewhat more ambitious question: What would make us concede that the ancient problem of free-will is SOLVED? {All is pre-determined, yet permission is always granted. —Maimonides (Moshe Ben Maimon) (1138–1204

2.16.2022 10:26pm - (Replying to @ArletOttens @erikphoel and 15 others) I'm wondering about this question too. From my viewpoint, the missing part in all compatibilistic theories is that they haven't simulated their ideas on a deterministic machine to demonstrate how compatibility works.

2.16.2022 12:55pm - An illuminating Podcast with Gary Marcus and Sean Carroll on AI, AGI, ML, DL and many topics often raised on this Twitter channel:

2.16.2022 12:58am - (Replying to @ArletOttens @philipcball and 13 others) The more I think about it, the more I tend to agree with you; a chess-playing machine could be a good laboratory/metaphor for playing with ideas about what free will is, and at what level of discourse it has survival value. Thanks.

2.15.2022 10:58pm - (Replying to @ArletOttens @philipcball and 13 others) According to your definition, every chess-playing machine has free will. Hard to swallow.

2.15.2022 5:45pm - (Replying to @UCLAengineering @UCLA and @FundacionBBVA) Thanks for posting. I didn't realize @UCLA is on tweeter. Please join our discussion on Free Will, consciousness, and other engineering issues.

2.15.2022 5:25pm - (Replying to @philipcball @NeuroMinded and 12 others) It's hard to believe that the sensation/illusion of free will has no survival value and, once we understand what that value is, we should be able to capture it on a deterministic machine. Thus, it is more than a catch-all antonym to determinism.

2.15.2022 10:19am - (Replying to @adam_lynton @ylecun and @ilyasut) I prefer to focus on those components of consciousness that have survival value, such as those that contribute to cooperation and communication among agents. If we succeed to emulate those components only on a digital machine we would be better off, leaving the rest to futurists.

2.15.2022 8:48am - Those who understand the Middle East to its core, also understand that symbolic gestures like this are more than "symbolic," and that peace will descend upon Israel-Palestine when (and only when) Hatikvah is played in Ramallah.

2.15.2022 8:34am - (Replying to @adam_lynton @ylecun and @ilyasut) As far as I know we do not have an agreed Turing test for consciousness, except, of course, systems that act and communicate as though they have consciousness. Here, my faithful guideline is: faking it is having it, because it is practically impossible to fake w/o having.

2.15.2022 6:07am - (Replying to @JohnAntonakis @metrics52 and @angrist_noam) Agree, I should try much harder. At the same time you and compatriots should realize by now that if I am taking the risk of shaking the temple, the temple needs reexamination.

2.15.2022 5:55am - (Replying to @instrumenthull) You will never find me saying "econ don't know how to solve this DAG". What you will find me saying is: "econ don't know how to solve this problem which they themselves declare to be important." Please check my record.

2.15.2022 5:46am - (Replying to @JMannhart @ylecun and @ilyasut) Good point. However, since we do not have a formal definition of c, all we can do is sift through the pile of qualities philosophers have identified that go with c, and ask whether these could be attained without some understanding (of the self). We do understand "understanding".

2.15.2022 5:30am - (Replying to @JohnAntonakis @metrics52 and @angrist_noam) Let's get to the substance of it. Can you take a quick look at the examples discussed in the paper and report back whether or not they add an important new dimension to the "causal science" as you know it, the one that was nailed long time ago. Thanks.

2.14.2022 8:18pm - (1/2) Goodness Gracious, SSRN informs us that our paper, "A CRASH COURSE IN GOOD AND BAD CONTROLS", was listed on SSRN's Top Ten download list, downloaded 1,553 times. I'm not surprise, of course, since most regressionists and economists fumble on these
2.14.2022 8:18pm - (2/2) sort of problems. More encouraging, Carlos has now posted R code with numerical simulation for all examples:…, which should prove all theoretical claims valid by simulation. This may convince perhaps Angrist and Imbens to convert to causal science.

2.14.2022 11:53pm - Here is an experience you never had. You listen to Beethoven Moonlight Sonata while driving on a Spanish highway, and a system tells you how a Bayesian Network works, and how it calculates probabilities of accidents where and how you are driving. Try it:

2.14.2022 11:29pm - (Replying to @ChrSzegedy @ylecun and @ilyasut) Because we hear some conscious systems saying: "Gee, I can't understand how I could do it to you, you have always been nice to me...."

2.14.2022 8:00pm - (Replying to @instrumenthull) I recognize the napkin (it belonged to David Freedman), but not the solution - it was much more elegant on David's napkin.

2.14.2022 7:50pm - (Replying to @newiins @ylecun and @ilyasut) Feynman differentiation between knowing and understanding calls for causal explanation. We know that air heats up when compressed. But this, together with PV=RT are still not enough, until we ask what makes an air molecule speed up upon collision with a moving wall.

2.14.2022 4:03pm - (Replying to @quantadan @totteh and @fhuszar) Not much consciousness is required for a system to conclude: "If soldier-1 refrains from shooting the prisoner would still be dead" (#Bookofwhy) or for Thucydides to conclude: "Without an earthquake I do not see how such an accident could happen."

2.14.2022 3:15pm - What comes first? I just answered it in my reply to @ylecun:… Consciousness requires understanding (part of) yourself, and you can't do it if you dont have the facility to understand an external domain i.e., answer question at all 3-levels of the Ladder .

2.14.2022 8:18am - On this happy birthday, we should not forget what Herzl said in the First Zionist Congress: “Zionism is a homecoming to the Jewish fold even before it becomes a homecoming to the Jewish land.”

2.14.2022 2:48am - (Replying to @fhuszar) Sorry for taking it seriously. In our world of "everything goes" one tends to lose the distinction between a joke and a serious face.

2.14.2022 2:42am - (Replying to @HillelNeuer) She is amazing. Where would an intelligent being, a graduate of a law school, get the idea that "race" has anything to do with the Israel-Palestinian conflict? Separation on the basis of hostile behavior - yes, but "race"? I would double check her Law degree.

2.13.2022 8:11am - (1/2) We are in receipt of J. Angrist (revised) Nobel Prize Lecture:… Anyone familiar with the science of cause and effect would probably wonder: "Is this the current state of knowledge in economics?". I'm sure many readers are ready to say: "No way!".
2.13.2022 8:11am - (2/2) My comments on the so-called "experimentalist" approach to economics are as valid today as they were in 2015, and 2020 The shift from model-driven to question-driven works has gone to far, the latter withers w/o the former.

2.13.2022 6:30am - Anyone who is concerned with the crime of apartheid, as it was institutionalized and practiced in South Africa and other countries to these days, till emptied of content by @AmnestyUK to become a mark of a thoughtless herd of dummies, should write to @AmnestyUK for ruining it.

2.13.2022 5:35am - Adolf Eichmann was the only person sentenced to death, and executed in Israel. His trial and its profound ramifications to our era are discussed in this event:… co-sponsored by the Daniel Pearl Education Center in East Brunswick N. J.

2.13.2022 5:14am - The old breed of BDS-dummies is quickly being replaced by a new mutant, called "amnesty-dummies". I can see them already roaming the green pastures of US campuses, breathing heavily: "apartheid," "apartheid" ...

2.13.2022 4:57am - Q. What's the difference between "apartheid" and "resistance"?
A. The beholder

2.13.2022 4:33am - (Replying to @soboleffspaces) I mean identifiability is an asymptotic notion, it only tells you what to expect when the number of sample increases indefinitely. It does not tell you how fast you can approach that limit or what the best way is of doing so.

2.13.2022 4:26am - (Replying to @gottfriedmath @artistexyz and 2 others) Precisely.

2.12.2022 11:36pm - The identifiability of PNS, PN, PS is a property of the assumptions made, not of the method we use. If the assumptions dictate bounds, then no smart method can turn them into point estimates. I suspects the probabilities you get will turn into bounds as sample size increases.

2.12.2022 11:24pm - The most poetic way to start the week.

2.12.2022 11:18pm - Anyone who wants to appreciate the blessing of counterfactual thinking should take a good look at this photo of Corbyn and imagine this entity occupying 10 Downing Street. Counterfactuals like this restore my faith in divine supervision.

2.11.2022 6:52pm - (Replying to @AJCGlobal and @TheRightsForum) It's time to stop wailing: "Look what they are doing to us" and start planning: "What can we do to universities that comply with the listing?" @DavidHarrisAJC

2.11.2022 6:35pm - (Replying to @Algemeiner and @SAFECUNY) It's time to stop wailing "look what they are doing to us" and start an all out offense on the racist elements of their ideology.

2.11.2022 4:10pm - (1/2) Here's footnote 10 (Causality p.216): I've presented this example to well over a hundred econometrics students and faculty across the United States. Respondents had no problem answering question 2, ONE was able to solve question 1, and NONE managed to answer question 3.
2.11.2022 4:10pm - (2/2) Chapter 5 (Section 5.1) suggests an explanation, and Section 11.5.4 a more recent assessment based on Heckman and Vytlacil (2007). It is now 20 years since I conducted this personal experiment, lots has changed in econometrics. Can any economist solve the problem? @causalinf

2.11.2022 3:562pm - (Replying to @conjugateprior @causalinf and 4 others) On page 215-7 of Causality, you can find a cyclic structural model, describing classical economic relationships between Price, Demand, Income etc. It is solved using the Fundamental Law of causal inference. Footnote 10 asks why econometrics students have difficulties doing it.

2.11.2022 7:46am - (Replying to @Springer866 @JCDecaux_UK and 2 others It's not @Amnesty, it's Qatar.

2.11.2022 7:40am - Here is the most depressing revelation of Amnesty's apartheid I've seen this week, many similar snake-heads will follow soon. Definitional, authoritative, scholarly, scientific, factual, almost as factual as a subway map -- stealthily penetrating the mindset of the dummy.Quote Tweet

2.11.2022 4:23am - (Replying to @causalinf @LaDelgadoP and 3 others) This is the UK edition, where the decided to draw causal graphs inside the letters.

2.11.2022 4:17am - This is a good overview of the 2nd Edition of Causality (2009). I don't know where I got the energy to write a separate review for each field that requested it. The explicit comparison to PO should have settled the equivalence once for all - it didn't. Highly recommended.

2.10.2022 9:35pm - (1/2) Readers concerned with the philosophy of economics may find this paper to be of interest… It articulates what generations of economists thought about counterfactuals in decisions. I do not share the author's "pluralism", given our current theory of
2.10.2022 9:35pm - (2/2) counterfactuals, which appears fairly complete and satisfactory.

2.10.2022 3:13pm - (Replying to @Zuasibel @davidmdraiman and @jdforward) This method of attacks has already taken hold, from London bus stations to US campuses, where Zionophobes now claim: "Get all apartheid centers out of our campus: Hillel, Israel study centers, classes taught by Israeli scholars, etc. We want an "apartheid-free" campus."

2.10.2022 12:43pm - The Trial of Adolf Eichmann: Documentary Viewing and Discussion
2.10.2022 12:43pm - Glad this event is sponsored by both Rutgers University (my alma mater 1965) and the Daniel Pearl Education Center, East Brunswick.

2.10.2022 11:45am - The communication people at BBVA were good enough to add the Israeli flag next to my name here:…. I now feel wholesome, having paid tribute to my college professors who stirred me into science, and to my Israeli colleagues who continue the tradition.

2.10.2022 10:57am - Amnesty's Israel report has dangerous implications — for Jews and Pale via @jdforward
2.10.2022 10:57am - @SharonNazarian underestimates the damage @AmnestyUK's "apartheid" has inflicted on the moral compass of the 21st century. It's libelous toxicity exceeds that of all swastikas ever painted on U.S. walls and pavements, for it comes from supposedly more respectable sources.

2.10.2022 10:16am - We are underestimating the damage @AmnestyUK's "apartheid" has inflicted on the moral compass of the 21st century. It's libelous toxicity exceeds that of all swastikas ever painted on U.S. walls and pavements, for it comes from supposedly more respectable sources.

2.10.2022 4:10am - 20 year old wine becomes tastier when celebrated by @_MiguelHernan at the @amJEpi anniversary.

2.10.2022 3:31am - This is great news. If 800 economists downloaded "Causal Machine Learning & Business Decision Making" the field must be awakening.

2.9.2022 11:12pm - Is this a hoax? Are these the same @amnesty and @amnestyusa that only a week ago flooded our streets with gasoline? Are they calling now for a better fire engine? Can someone ask them (politely) to clean their act first?

2.9.2022 10:36pm - (Replying to @UtkarshaniJ) I've skimmed through the paper with the hope for finding a formal definition of CausalKG, similar to the definitions we have for CBN and SCM (eg Causality, Ch 1), but I could not find any. Have I missed it? Specifically, what do the nodes and arrows represent?

2.9.2022 4:47am - They just sent me an English citation of the Award: Now I can see what they liked about my work, and whether they got it right.

2.9.2022 3:18am - I am extremely proud and humble to have been selected a recipient of the 2020 BBVA Frontier of Science Award. I hope this recognition would make causality research visible beyond its standard boundaries, and I thank the nominators and committee for recognizing these potentials.

2.8.2022 11:27pm - (1/2) More on the death of Sir David Cox. David was at home on the day of his death, together with his wife Joyce and his daughter Joan. He had gone upstairs to look for his cell phone and then he fell from the top of the long and narrow staircase. Neither Joyce, Joan, nor the
2.8.2022 11:27pm - (1/2) ambulance crew could get him back to lasting consciousness. The medical report indicated the comforting news: David "was not in any pain at all”. I hope this information helps all of us better accept this sad event. (Related to me by Professor Nanny Wermuth)

2.8.2022 10:47pm - @KenRoth , I am hardly a government agent, yet I dare accuse you of degrading the human rights movement to become an icon of deception and disreputation -- you are guilty of a breach of public trust.

2.8.2022 10:00pm - (Replying to @robberg15 and @isaacdecastrog) I would try the @ADL . In the old days, they used to send someone to speak with the principal and make sure the person responsible gets reprimanded, on record. But that was when the offense was against Jews, not sure where Israelis rank on #ADL priorities these days. I would try.

2.8.2022 9:37pm - (Replying to @hitlabjason) Interesting metaphor. The brain breaks the rules at night to escape from local minima, then restores them in the morning, and check if the new solutions are better than the old one. Interesting.

2.8.2022 3:56pm - The problem is not that these students complained about "safety", but that the professor took them seriously, probably unaware of their tactics, and actually cancelled "Wonder Woman" and replaced it with another score. The music never sounded "safer".

2.8.2022 2:09pm - Great to see from above places I used to visit either bare foot, or in army boots,

2.8.2022 9:27am - Knowledge vs. influence: "We are the land where every covid crackpot says, “I'm doing my own research” — which usually consists of googling the views of other crackpots...Lack of knowledge is practically a prerequisite for influence. "

2.8.2022 1:35am - If you think this mentality is limited to Kuwait, you are innocently wrong. A student from the department of Music at UCLA told me last year that a class discussion on the music score of "Wonder Woman" was cancelled when a Palestinian student complained he doesn't feel SAFE.

2.8.2022 1:11aM - (Replying to @artistexyz) Thanks for the alert.

2.7.2022 6:14am - Rebellious ML students who took time to study causal inference might wish to look into this company that is looking to hire research scientists to lead their expanding operations.

2.7.2022 2:01am - Kuwait bans upcoming movie starring Gal Gadot,
2.7.2022 2:01am - Oh, how I wish Kuwait would give me the honor of banning my books and articles, so that I, too, will be labeled "former soldier in the Zionist occupying army" -- a noble title which I equally deserve, given my 1953-1956 service. @GalGadot #coexistence

2.6.2022 10:46pm - My most painful disappointment is seeing colleagues at my university legitimizing the barbarity of Hamas. It started here: and is still going on here: Hamas counts on their support when celebrating death of a holocaust survivor.

2.6.2022 10:26pm - Another metaphysical plan for resolving the Israeli-Palestinian conflict destined to be rejected by both sides; Palestinians, b/c they can't stomach a permanent sovereign Israel no matter how "confederated", and Israelis, b/c they perfectly understand how Palestinians feel.

2.6.2022 8:27pm - To understand how brutal and inhuman holocaust denial is, one must listen to this testimony of Shoshana Kahana who survived Auschwitz and came to Israel in 1946. She is the mother of UCI Professor Rina Dechter, author of "Constraints Processing" (2003).

2.6.2022 6:43pm - (Replying to @StandWithUs and @michaeldickson) Breaking down what they are saying ain't enough. The fact that they rally hundreds of people to this chanting tells us that we have not been using the right strategy to expose their moral deformity. Its time for SWU to adopt the word Zionophobia, which highlights this deformity.

2.6.2022 7:53am - (Replying to @DubinPamela) One thing power does care -- losing power.

2.6.2022 4:02am - (Replying to @lackboys3 @SAFECUNY and 9 others) It's the only vaccine that targets the virus.

2.6.2022 3:56am - (1/3) To wake up the establishments we need to present them with a thoughtful plan that would challenge their hegemony over community resources. It should say: "Here are 2 dozen groups of young activists that have sprung up during your slumber. They are now prepared to unite and
2.6.2022 3:56am - (2/3) claim ownership over public trust and community resources." To be credible, the plan should designate a fearless experienced leader, and a respectable "war cabinet" that would jolt our leadership into a major reassessment of their outdated strategies and
2.6.2022 3:56am - (3/3) corner them into deciding if they wish to remain in a position of leadership. @GilTroy @YKleinHalevi @EinatWilf @CotlerWunsh
2.6.2022 3:56am jj

2.6.2022 2:21am - (Replying to @soboleffspaces @attilacsordas and 3 others) No need to sweat over the fancy word "recursive". In the context of SCM, recursive refers to a set of equations with no cycles, also going by the rubrics "no feedback" or "no simultaneity".

2.6.2022 2:00am - (Replying to @RMCarpiano @DrTomasAragon and 5 others) Not clear what the "silly issue" was, or what is evidence for what.

2.6.2022 1:55am - (Replying to @DrTomasAragon @RMCarpiano and 5 others) I dare to fear that, if the matching used is not scrutinized by DAGs it is likely to introduce bias.

2.6.2022 1:36am - Zionophobia Is Discrimination via @JewishPress
2.6.2022 1:36am - (1/2) Like the majority of Jews, I am secular and Zionist. My take on Zionophobia is therefore different from the one presented in this paper; it is based on the claim that religion has no monopoly on man's identity. Still, I join the author in urging writers and speakers to use
2.6.2022 1:36am - the fighting word "Zionophobia" (rather than "antisemitism") in all efforts to achieve: (1) respect for Jewish students on campus and (2) security for the Jewish people worldwide. (see @lackboys3 @artstop @GilTroy @cpechet @ShimonLevit @EricJGreenberg

2.5.2022 4:22am - (1/2) I'm not ready to join @Eve_Barlow in lamenting the future of American Jewry. It needs an awakening Herzl, not a lamenting Achad Haam. To witness, @ADL did revise its outdated definition of racism. Exhibit-2, Jewish establishment was shaken up at USC by 65 professors who
2.5.2022 4:22am - decided "enough is enough" when Zionophobia broke out American Jewish establishment is bewildered and sluggish, true; it needs constructive guidance and forceful jolts from observers who are not handcuffed by fund-raising and institutional inertial forces.

2.5.2022 3:17am - (1/2) A new paper comparing Rubin and Pearl's approaches, in J. Econ. and Phil. I agree with the main conclusion: "The differences between the frameworks are best illuminated not by appeal to their causal semantics, but rather reflect pragmatic modelling choices." I interpret this
2.5.2022 3:17am - (1/2) to mean: "If researchers in one camp tend to be sloppier in some step of the analysis, it does not make their theoretical framework any different. IOW, non-sloppy researchers starting w same assumptions will reach the same conclusions regardless of which camp they feel loyal to."

2.4.2022 8:49pm - (Replying to @beenwrekt The #Bookofwhy goes through several such oversights. The most common one is enshrined in the word "malaria" (Mal Air) The less known is the death of Captain Scott at the South Pole, attributed to scurvy which, despite James Lind's observations was misunderstood till the 1930's.

2.4.2022 12:34am - (1/2) Friends called my attention to a new paper in Nachrichten aus der Chemie which describes in horrific colors how "cancel culture" is encroaching on scientific research, not merely in hiring and funding policies, but also in censoring scientific texts.
2.4.2022 12:34am - I wasn't aware, for example, that "normal distribution" could be offensive to some (many?) readers who do not see themselves as normal. I hope this madness does not creep from Chemistry to AI or CI. Note, it's the offended feeling that rules, not the author's intention.

2.3.2022 4:00pm - Amazed by the magic power of anniversaries, I confess that today was the first time that I watched @peterhimmelman Ted Talk: Sharing with readers, before we go back to science and other mundane matters.

2.3.2022 11:15am - @DovWaxman 's argument is a painful product of Jewish tendency to calibrate every evil in terms of its antisemitic content, and of Jewish leadership's reluctance to adapt Zionophobia as a new measure of racism that currently endangers the future of the Jewish people.
2.3.2022 11:15am - The Zionophobic (hence racist) character of @amnestiUK report stems from its failure to acknowledge the fact that Israel is facing a war-waving neighbor that is openly and totally dedicated to its destruction, hence need be judged by rules governing defensive wars.

2.3.2022 1:23am - Getting it Right in Defining Racism by @JGreenblattADL
2.3.2022 1:23am - Happy to see @JGreenblattADL revising ADL's definition of Racism. The key phrase is "other forms of identity-based hate" which now includes "anti-Zionism" and "anti-gays". The next step is for ADL to adopt the word "Zionophobia," to make its new definition crisp and visible.

2.3.2022 12:49am - @EinatWilf is at her best in this video. My only reservation of the debate is its giving a Zionophobe like @mehdirhasan the honorary title "antisemite" - he does not deserve it, he is not a product of uneducate parents; he knows where it leads, and pursues it cold bloodedly.

2.3.2022 12:14am - (Replying to @Plank93621370 and @Ostrov_A) Proclaiming that a Jewish state is "a fact" does not negate the proclaimer's intent to promote its destruction. What I am calling for is on @AmnestyUS to recognize the right of Israel to distinguish those who seek her destruction from others; such distinction is NOT apartheid.

2.2.2022 10:16pm - (Replying to @Ostrov_A) Why not? Hatikvah Bat Shnot Alpayim!

2.2.2022 9:22pm - If we stopped to count all the trumpets condemning antisemitism we would have no enemies left.

2.2.2022 9:08pm - If Amnesty's report does not suggest the dismantling of Israel, as this guy in the video claims, they can easily issue a clarifying addendum, saying: "Israel is the historical homeland of the Jewish people, and its existence is a moral imperative". I'm waiting.

2.2.2022 8:22pm - The causal structure of the Arab-Israeli conflict in a nutshell: As long as you make me "temporary", I cannot make my defense "temporary". No one could make it clearer that @EinatWilf

2.2.2022 4:24pm - (Replying to @kkuzmich) That's precisely the danger in @amnestyUK making trumped up charges; it makes even innocent bystanders believe in evil spirits where none exists.

2.2.2022 6:35am - Thank you @WilliamPesek and @peterhimmelman for reposting these moving articles. Daniel's spirit is still moving souls here and there around the planet.

2.2.2022 3:35am - (Replying to @profnfenton @gottfriedmath and 5 others) Nice demonstration. However, note that you have assumed a functional relationship (linear) between Y and its parents. Moreover you are learning the priors of the invisible exogeneous variable U, not the visible exogeneous variables in the BN. These place your analysis within SCM.

2.1.2022 8:38pm - Sharing a couple of articles that appeared today, the 20-yrs anniversary of Danny's murder:
2.1.2022 8:38pm - (1/2) [no text]

2.1.2022 12:03pm - (Replying to @Ostrov_A) Amazing! Reminds me of the days I used to do book-tours and sign books. Where/when was it?

2.1.2022 11:30am - In 2013, David Broza wrote a song to Danny, based on the Lion's Den lyrics. He performs it here: Thank you, @DavidBroza

2.1.2022 5:45am - [jp retweeted] On this 20th anniversary of Daniel Pearl's murder, pls take a moment to honor his legacy by embracing his dream that a common humanity amongst peoples, countries & continents will someday prevail. I am certain Gabriel likes his music @yudapearl

2.1.2022 3:30am - ADL Condemns Amnesty International's Latest Effort to Demonize Israel via @ADL
2.1.2022 3:30am - @JGreenblattADL @Ostrov_A @EinatWilf @AJCGlobal @HillelNeuer @NGOmonitor @KhaledAbuToameh @bandlersbanter

2.1.2022 1:28am - (Replying to @_fernando_rosas) Actually it was the thrill of my life, not hard at all. Imagine finding a flashlight and using it to discover many wallets that people have lost over the years in the darkness. The disappointment came later, when people could not remember loosing those wallets.

2.1.2022 12:50am - An Amnesty International report labels Israel an "apartheid" state in a biased report that the Biden Administration should condemn as false. via @WSJOpinion
2.1.2022 12:50am - (1/3) The WSJ, as well as other condemnations of Amnesty International that I've read, neglects to mention the most important factor missing from the report: ARAB REJECTIONISM. Reflected in media, education, and mosque sermons, rejectionism compels Israel to protect its existence
2.1.2022 12:50am - (2/3) seriously, sometimes sternly, perhaps even excessively, yet always more humanely than any country in the history of mankind that had to endure 75 years of continuous besiegement, deligitimization, waves of violence and open existential threats by its neighbors. No other
2.1.2022 12:50am - (3/3) such country exists. Criminalizing the creation of a country that survived 75 years of rejectionism amounts to rewarding the latter and threatening coexistence everywhere.

2.1.2022 12:07am - (1/ ) Not an easy question. It is somewhat surprising for me to realize that #Bookofwhy may be one of only few books to cultivate scientific thinking. (Perhaps because it tackles a topic that has been neglected by most standard disciplines.) I was going to recommend Jan Hacking's
2.1.2022 12:07am - (2/ ) "Taming of Chance," which traces the development of probability from an intuitive to scientific concept. Also, Thomas Kuhn's "The Structure of Scientific Revolutions" is a good cultivator (see What we need is a book that would lead the reader gently
2.1.2022 12:07am - from practical research questions to their formalization (in terms models of reality) to their algorithmic solutions. Perhaps a future edition of #Bookofwhy would tailor it better for this purpose, though Primer is not far from it.

1.31.2022 2:52pm - If you were to include a 3rd option: "The data" it would have garnered 2K votes from the 3K PhD's that "data science" departments are producing each year. The other 1K would say: Causality?

1.31.2022 11:44am - Feb. 1 marks 20 years since our son, Danny, was murdered in Karachi, Pakistan. We will be lighting the traditional "soul candle" and say Kaddish in his memory. Readers noted that the link to my poem "The Lion Den" is obstructed by The New Yorker's paywall, so, I am sharing
1.31.2022 11:44am - a fresh link: If you want to join us in "Kaddish para Daniel," please click on and whisper the holy words "Yitgadal V'Yitkadash Shmei Raba" -- they have shaken the gates of heaven many times before.

1.31.2022 10:02am - (Replying to @AsraNomani and @mehdirhasan) I would never honor a Zionophobe with the title "anti-Semite". The latter is a victim of unfortunate upbringing, alcoholic parents or God knows what. The former are calculating eliminationists; they know the genocidal consequences of their agenda and pursue it nevertheless.

1.31.2022 9:49am - For @mehdirhasan the world is prism clear: When Israel makes compromises in pursuit of peace, it's "hypocrisy"; when Arabs reject compromises in pursuit of war, it's "resistance".

1.31.2022 9:03am - (Replying to @FJnyc @KimmelmanLouis and 2 others) @mehdirhasan has never hidden his Zionophobic mentality. In one of his Oxford debates he stated explicitly (rephrased): "There are many ethnic group that do not have states, Jews should join them". The music of Hatikva made his stomach turn, for it renders the genocidal dream of
1.31.2022 9:23am - (Replying to @yudapearl @FJnyc and 3 others) stateless Jews less likely, and Zionophobes feed on that dream. Not because they are antisemitic, God forbid, but simply because they innocently know nothing about Jewish peoplehood and were fed Zionophobic mother milk that makes some nations lesser nations than others.

1.31.2022 3:47am - (Replying to @giovbriganti) From my corner of the wood the revolution seems to be progressing rather slow; do you see any reason why #Bookofwhy would not be studied in Med schools? I promise to donate all royalties to Covid-19 research.

1.31.2022 1:40am - (Replying to @zevkalman and @jonathanliew) I am a student of symbolic logic. I knew this peace is a REAL PEACE the moment I heard a UAE orchestra play the Israeli anthem Hatikvah (confessedly, I shed a tear). I now know that the rest will come, and easily so -- V'Im Lo Machar Az Mochorotyim.

1.31.2022 1:07am - A welcome introduction of Bayes Networks in the field of mental disorders. I am surprised to learn that the field has thus far been dominated by partial correlation networks, quite a distance removed from the underlying mechanisms. .

1.30.2022 5:03am - Say what you will about this visit: extravaganza, diplomatic-gesture, PR-show, etc,. but as one who grew up in an era when such a visit was inconceivable, I feel truly touched, even moved. And if our washing expert @jonathanliew decides it's "peace washing", I'd be doubly moved.

1.30.2022 11:57pm - It is refreshing to listen to an educated Muslim woman discussing antisemitism and Islamophobia, days after hearing Linda Sarsour say the latter is systemic, the former is not. Dr. Qanta Ahmed is the author of "In the land of invisible women" .

1.30.2022 5:14am- (Replying to @Moshepupik1 and @jonathanliew) Conspiracy theories tend to metastasize.

1.30.2022 3:59am - (Replying to @gottfriedmath @soboleffspaces and 4 others) If you express formally the counterfactual question asked (about Fred's would be life expectancy) you will see that it differs from what is computed using CBN. This means that there are more than one SCM that satisfy the initial CBN probabilities, giving different PN values.

1.30.2022 2:58am - (1/2) My attention was drawn to an article on Campus Antisemitism which I recommend for students a faculty entrapped in this battle ground: It properly puts anti-Zionism at the center of the problem, not antisemitism. However,
1.30.2022 2:58am - (2/2) it is still hung-up in yesterday's debates on when anti-Zionism spills over into antisemitism, as if the latter is the the only basis for the former's illegitimacy, instead of viewing the former as morally despicable on its own right.

1.29.2022 10:02pm - (1/2) Keen observation! It appears that @profnfenton computes a Level-3 quantity, PN, from CBN, which is a Level-2 model; something we know is impossible (see Causality p. 39). Where is the catch? The computation applies the twin-network procedure to a CBN, as if it was a SCM.
1.29.2022 10:02pm - (2/2) Doing so may be valid under some assumptions (eg monotonicity) but not others. The bias introduced can be assessed by comparing the counterfactual expression of PN with the do-expression actually computed.

1.29.2022 9:32pm - Any definition of "racism" that excludes Zionophobia must exclude Islamophobia; religion does not have a monopoly on man's identity. @JGreenblattADL

1.29.2022 8:08pm - We can't let Holocaust Memorial Day pass without listening to one Yiddish Song. About 14 min. into this video: we can hear Oyfn Pripetchik, a song that my mother sang when I was 2-3-4 years old. A treasure of a lost civilization.

1.29.2022 7:20pm - Intersectionality is based on the logic that all good causes are good, hence linked by goodness. Unfortunately, intersectionalists tend to conflate it with the logic of "If one of my causes is good, they all must be - I am incapable of promoting an ugly cause".

1.29.2022 6:54pm - Sister Linda, Sister Linda. I don't know much about antisemitism, but I can assure you that Zionophobia is very much systemic in my University. Some Zionophobes are so systemic, they even said: "You can't be a Feminist if you are a Zionist", remember?

1.29.2022 6:30pm - Looking at the leukemia example, it occurred to me that this is undeniably "real life" data [F&L, 1990], so, it should pacify all skeptics (mostly economists) who vow to remain unconvinced unless they can touch "real-life data".

1.29.2022 4:36pm - Yes, it is amazing! So, I am retweeting for readers who would like to use it in Covid-19 and other situations, as we did in Causality p. 301 to show that there is a 37.5% chance that a child who died from leukemia would have survived absent nuclear radiation. While ACE = 0.000162

1.29.2022 3:58pm - (Replying to @soboleffspaces and @gottfriedmath) Monotonicity is enough for this to hold. See Causality Theorem 9.2.15.

1.29.2022 3:01pm - (Replying to @Perperuna2) Bias is not the issue because we are doing regression, not causation. The issue is efficiency.

1.29.2022 2:52pm - (1/2) As a personal tribute to Sir David Cox, here is a toy problem that he solved in 1960 and which has kept me awake many night since. Needed: the regression coefficient R_yx for a Markov chain X-->Z-->Y. Guess which is better, to estimate R_zx and R_yz separately, then multiply
1.29.2022 2:52pm - their estimates? Or estimate R_yx directly? Cox has shown that the former is better (yielding a lower asymptotic variance). Here is why, how, and what we can learn from it in the age of Data Science: Thanks for the inspiration, Sir David.

1.29.2022 2:30pm - This CACM article supplements the "Seven Tools", and marks an important step towards mainstreaming CI. The next step is getting ML researchers to ask: "What should the world be like for our NN algorithms to answer our research questions." Getting there!

1.29.2022 8:23am - (Replying to @BenMFreeman @Eve_Barlow and @ADL) I agree with @BenMFreeman . Skin color is but one among several immutable characteristics of a group that would qualify hatred as "racism". Zionophobia, for example, earns its racist character by virtue of attacking an ingrained component of a people's identity. Please correct.

1.29.2022 6:26am - (Replying to @causalinf) Do you see the history of economics differently than I do? In particular, do you agree with my answer to "What Kept the Cowles Commission at Bay?" From teaching viewpoint, any benefit in going to an era where "average causal effect" was not defined?

1.29.2022 6:08am - (Replying to @symmo1969 and @jeremycorbyn) Beg to disagree. I've been following @jeremycorbyn since he became a Zionophobe. Israel is not an "invented nation" but a symbol of normalcy to its people, and Corbyn has expressed unqualified support to organizations who proclaim publicly their intention to eradicate Israel.

1.29.2022 5:50am - (Replying to @Scholars4Peace and @SAFECUNY) Depressing and well documented report by @ADL . Unfortunately, it ends with another "Oy Vey"; no "war cabinet" and no plan for counterattack. The word Zionophobia (or equivalent) does not appear even once. Things will turn around when we learn to spell Zionophobia. @JGreenblattADL

1.29.2022 2:07am - Belated congratulations to Jennifer Chayes @jenniferchayes for receiving the 2020 ACM Distinguished Service Award. It is a personal thrill for me, because Jennifer was that mysterious lady who woke me up in 2012 to tell me about winning the Turing Award. I first thought she
1.29.2022 2:07am - was pulling my leg, so I asked her to spell "Turing" - she passed. We are since teasing each other with unimaginable congratulations, but this one is real and most deserved, given what she has accomplished: A warm welcome, Jennifer to our Twitter family.

1.28.2022 3:55pm - (1/2) Sad news. I was just notified that (Sir) David Cox has died on January 18, at age 97. David was kind enough to invite me to a Statistics Conference (Florence, 1993) when I was still making my first steps in Causality, and his last email (March 2020)
1.28.2022 3:55pm - congratulates me on my election as Honarary Gellow of the RSS. I tried to interview him for #Bookofwhy, but he said: "Causality? I'll tell you what I think, but don't quote me." I am respecting his wish. I'll miss his honesty, brightness and kindness.

1.28.2022 3:16pm - I am reminded that January 27 is the birthday of Lewis Carroll, author of Alice in Wonderland, Why do I mention it? Because he also wrote "The game of Logic" (1887), proposing graphical methods for solving logical problems, improving on Venn's Diagrams.

1.28.2022 11:56am - (Replying to @ZachThorwirth and @hima_lakkaraju) I havn't seen it. Many suggestions, please send me details on Thanks

1.28.2022 1:46am - This is one of many occasions in which Corbyn identified with the terrorists responsible for the Munich massacre or with their holy cause. After all, as Mayor Livingston said, terrorists have no tanks, so they must do what they do best, ... also known as the "Corbyn doctrine"

1.28.2022 1:23am - (Replying to @SwissRexAG and @hcburger1) So why is it so hard to summarize here, on twitter, in a few words. E.g., Cointegration is a phenomena in the data/model/world which involves X,Y and Z, and makes it difficult to estimate/interpret/infer W or S, since the assumption of A, B is violated...

1.28.2022 1:13am - (1/2) I should have said: "ONE purpose, of causal inference." Another purpose is, of course, to identify what assumptions about the world are sufficient (or necessary) for estimating causal effects when RCTs are infeasible. Still another purpose is to assist trialists in deciding
1.28.2022 1:13am - whether some RCT can answer the research question asked and, if so, how. Personalized decisions, by enlarge, involve questions that cannot be answered from RCT data, unless augmented with observational studies. I hope the trialists at ISPOR got convinced.

1.28.2022 12:37am - (1/2) Having read some of the replies to @jeremycorbyn 's tweets on Holocaust Memorial, I am totally disoriented: Was it a scam, a hoax, or the real Jeremy Corbyn opting to speak on this day. Can anyone verify?
1.28.2022 12:37am - (1/2) Another phenomena: All who decry Corbyn's hypocrisy accuse him of antisemitism which he, of course, denies. None points to his obs. animosity to Israel which he can't, and won't deny. Has he ever acknowledged the human face of 7 million human beings begging for a day of normalcy?

1.27.2022 11:37pm - (Replying to @RevInPaint @paulm1402 and 2 others) [retweeted 1.27.2022 11:14am] @jeremycorbyn , if you had any respect for those who perished in the Holocaust, you would spare their surviving families the agony of listening to you on this memorial day, a man who devoted his career to demonizing Israel -- the embodiment of their dreams and dignity.

1.27.2022 11:31pm - (Replying to @AnnJohn30914404 @jeremycorbyn and @IslingtonBC) [retweeted 1.27.2022 11:14am] @jeremycorbyn , if you had any respect for those who perished in the Holocaust, you would spare their surviving families the agony of listening to you on this memorial day, a man who devoted his career to demonizing Israel -- the embodiment of their dreams and dignity.

1.27.2022 11:14pm - @jeremycorbyn , if you had any respect for those who perished in the Holocaust, you would spare their surviving families the agony of listening to you on this memorial day, a man who devoted his career to demonizing Israel -- the embodiment of their dreams and dignity.

1.27.2022 6:50am - (Replying to @symmo1969 and @jeremycorbyn) Populist clichés. You would never (repeat: never) find me use the term "antisemitic" in any discussion of Israel. The term "Zionophobic" serves much better to expose the bigotry and genocidal aims of those who are dehumanizing a people under the cover of "criticism of Israel".

1.27.2022 6:33pm - Rev. Nelson should resign. It was indeed with friends from the Wiesenthal Center that we went to Alabama to speak at the Presbyterian conference about peace in the ME. This was before this warm community got hijacked by incurable Zionophobes like Rev. Nelson. He should resign!!!

1.27.2022 4:03pm - (Replying to @hcburger1) I've heard this term so many times, in so many interpretations and misinterpretations that I could not refrain myself from asking: Could anyone explain what it is in less that 285 characters? Still waiting.

1.27.2022 3:44pm - (Replying to @bernhardsson) The purpose of causal inference is to make randomized experiments easier, by removing obstacles that make them so hard today. For example: sparse data, selection bias, reconciling two RCT's, etc. That's what this video is trying to show to trialists:

1.27.2022 3:23pm - Deep in their hearts, the ML folks know it too, yet, for some strange reason, they are going at it through the back-door (pun intended).

1.27.2022 3:12pm - (Replying to @andersagf) Wow, how I wish some of the so called "new- school" economists would read Haavelmo in the source. Wow, how many hours or torture would their students save. Wow, how much clarity would shine on their so called "psuedo experimental" exercise. More Wow's? See

1.27.2022 1:32pm - For me, "Never Again" means a safe Israel, accepted by its neighbors, and allowed to channel its unique vitality towards peaceful dreams, instead of existential defense. Retweeting my words at the UN, 2006,

1.27.2022 1:59am - Thank you @bernard_bolzano for "liking" this old Tweet of mine on Meta-analysis and EBM. I was about to double "like" it when it came up, only to find out it's mine - disappointment. I do not have any penetration to EBM circles, can anyone retweet it for them, please.

1.27.2022 1:48am - (Replying to @josephpapptheat @blakeflayton and 11 others) I'm getting "account suspended" for @IfJews . Twitter vows to protect all Zionophobes especially Jewish.

1.26.2022 12:57pm - (Replying to @ZachThorwirth and @hima_lakkaraju) Have I been on your program? What is the project about? Please email me details. Thanks.

1.26.2022 10:19pm - UC Regents Members Call UCLA Asian American Studies Dept. Statement Accusing Israel of “Yellow-Washing” “Inappropriate” via @jewishjournal
1.26.2022 10:19pm - (1/2) When UCLA department of Asian-American Studies (AAS) first posted their malicious nonsense on their website we appealed to the Chancellor, who directed us to the President, who deferred it to the Committee on Academic Freedom, who issued a "ruling" that they will be looking
1.26.2022 10:19pm - into it ... In the meantime the offensive post remained on UCLA website, inviting students and parents to see how idiotic, yet racist some faculty can be. Then one of us said: The system is broken, perhaps the Regents would care? We did, and they seemed to care. Future will tell!

1.26.2022 9:06pm - (Replying to @FrnkNlsn) Thanks for posting. Yes, David MacKay was a fun -loving, smart and cheerful fellow. He even discovered one of my algorithms which makes him super-smart. I'm going to look up his Chapter 27. Thank you.

1.26.2022 7:17pm - (Replying to @soboleffspaces and @ThomasAgoritsas) The theorem is stronger. It says that even if one wishes to ignore the advice of the backdoor criterion, the asymptotic bias of all those estimators would be the same. A rare case where ignorance does not pay.

1.26.2022 10:30am - Campus Double Standards and the Need to Protect Jewish and Zionist Students via @jewishjournal
1.26.2022 10:30am - Excellent article on the legalities in the USC scandal, and why the administration was hesitant to act properly. It's time for USC to recognize "identity-based Zionists" as a sub-category based on "national origin," and treat such students as any other minority group.

1.26.2022 9:49am - Presbyterian Clerk Doubles Down Against Israel, But His Critics Man the Barricades
1.26.2022 9:49am - To my friends in Presbyterians for Middle East Peace. I spoke to you in Birmingham Alabama in 2006, and I would cherish renewing our conversation about the ME conflict. I hope Rev. J. Herbert Nelson takes your recommendation and resigns. Presbyterians deserve a better leadership.

1.26.2022 9:20am - (Replying to @bIack_eye @michaeldickson and 2 others) Thanks for the correction. I tried to find the most plausible explanation for @jonathanliew obsession with uncovering a clandestine "washing" behind each of Israel's projects. A society that can do no good mirrors a journalist that can write no truth.

1.26.2022 4:05am - The only way to talk about Muslim antisemitism is to admit that it is highly contaminated with Zionophobia and to address the latter as a more dangerous variant of the two. Our reluctance to name this variant makes our fight impossible.

1.26.2022 2:24am - Even the most devout EBM evangelist, one who preaches "Data, Data and Only Data", would succumb to Science and admit that theory is inevitable when the need arises to combine the findings of two or more RCT's. @soboleffspaces @artistexyz @eliasbareinboim @causalinf @VC31415

1.26.2022 12:50am - For Marek Edelman, Every Day Was Holocaust Remembrance Day via @jewishjournal
1.26.2022 12:50am - Beg to differ. Remembrance in silence, even a vigilant one, betrays those we wish to remember. Silence in the face of current attacks on Israel, Zionism and resurrection espouses forgetfulness, not remembrance. See a different take on Remembrance Day:

1.25.2022 11:01pm - Remembrance, Vigilance, and Beyond -
1.25.2022 11:01pm - In November 2005, the United Nations General Assembly officially adopted January 27th as the International Day of Commemoration in Memory of the Victims of the Holocaust. I was invited to address the inauguration ceremony, January 27, 2006, and gave the speech below, in which
1.25.2022 11:01pm - the word "Zionophobia" and the dangers of Zionophobic hatred were spelled out in the context of holocaust remembrance. As we are preparing for Holocaust Commemoration Day, 2022, and as a tribute to those who perished under hate-based violence, I am retweeting my words of 2006.

1.25.2022 5:12pm - (Replying to @soboleffspaces) Yes. I start teaching the controlled direct effect and then show the difficulties that emerge: (1) It varies with the value of the mediator and (2) you can't define the indirect effect that way. Stuck! Now we know why we need NDE and NIE.

1.25.2022 4:57pm - The controlled direct effect is level 2, but the Natural direct and indirect effects are Level 3, because (in general) we cannot express the in do-expressions, namely, they are not estimable from CRTs.

1.25.2022 3:30pm - I've enjoyed it thoroughly, especially because it forced me to re-think how trialists could be convinced that CI is something they should study and teach. I used a new approach, and I think I convinced even the most skeptical EBM folks; at least this was how I felt. @UweSiebert9

1.25.2022 12:34am - (Replying to @artistexyz @RWJE_BA and 5 others) The question is not whether EBM should shun homeopathy, but whether it should embrace human anatomy, and how.

1.25.2022 12:24am - (Replying to @michaeldickson @jonathanliew and @guardian) @jonathanliew reminds us how the terms "sportwashing", "yellowwashing" "pinkwashing", all evolved from "brainwashing", of which he is a poster victim. Only a thoroughly brainwashed organism would cite conspiracy theories about Israel HR violations w/o questioning their source.

1.24.2022 10:48pm - I'll support any boycott initiative aiming to shift the burdens of "sign ups" from reviewers back to the publisher.

1.24.2022 10:29pm - Thanks for having me, Uwe. I am going over the slides now, the first of which reads: A Computer Scientist Meets Trialists or What can Causal Science tell Healthcare Decision-Making. It promises to be a lively discussion.

1.24.2022 10:20pm - (Replying to @soboleffspaces @artistexyz and 4 others) Are you implying that, before Eddy, say in the tradition of RA Fisher, expert opinion was accepted as another guide for medical decisions, and Eddy purged them out, to create what we now call Evidence-based medicine?

1.24.2022 1:33pm - "Years of lost life": An interesting new application of PNS, in addition to being a great example for teaching counterfactuals.

1.24.2022 6:50am - (Replying to @Medcifue @SPilleron and 2 others) Today, with the help of graphs or potential outcomes, even statisticians can define confounding; they couldn't before Rubin 1974. Even today, those 90% who do not use graphs cannot tell when confounding exists, b/c "ignorability" is too opaque for mortals.

1.24.2022 12:54am - (Replying to @SPilleron @Medcifue and 2 others) From my readings, statisticians never defined what they mean by "confounding", for a good reason: it cannot be defined in the language of statistics. Epidemiologists defined it in 1986 and operationalized in in 1999. See

1.23.2022 10:04pm - I wrote a poem for Danny called "The Lion Den" It ends: And when Daniel was lifted from the den, So the Bible tells us, No wound was found on him, Because he stood his ground 'Cause he stood our ground So the Bible tells us. (Daniel 6:28)

1.23.2022 7:58pm - (Replying to @ROBYN7685) Yes, he is 19, first year in college. A great solace.

1.23.2022 7:36pm - Opinion: Don't use Martin Luther King Jr. Day to spread anti-Israel hate. King was a Zionist.
1.23.2022 7:36pm - Evidently, MLK saw the looming dangers of Zionophobia as early as 1967. Remarkably, just a few months before his death, he talked admirably about Israel, and described it as "an oasis of friendship and democracy" (MLK, March 1968)

1.23.2022 3:35pm - (Replying to @KyleCranmer) On collision between the two.

1.23.2022 11:23am - Horrible Day. Evil runs amok when God falls asleep on his watch!

1.23.2022 10:39am - (Replying to @soboleffspaces and @HenningStrandin) A slight correction to Lewis: The "only if" part of this property is true, not the "if" part. (Assuming c and e are variables). See Primer for examples of non-transitivity.

1.23.2022 10:12am - My faithful Twitter counter notes that the number of followers of our educational channel has reached 55K, an increase of 15K in the past year. Thank you all for keeping me honest and alert and giving this inquisitive community a sense that we are learning from each other.

1.23.2022 9:40am - (Replying to @wayneholmes) I do actually believe so. Whatever we know about our intelligence from centuries of philosophical, psychological, neurological studies will dwarf what we will learn about it from AI, in the next 3 decades.

1.22.2022 11:15pm - (Replying to @blakeflayton) Don't blame the Zionophobes. Jewish leaders, writers and activists have all given them the idea that Zionophobia is a less disgusting form of bigotry than anti-Semitism. E.g., Why should the latter can be disguised as the former and not the other way around.

1.22.2022 10:38pm - So what exactly happened there in Colleyville and its aftermath? The sharp words and insightful eyes of @BretStephen_NYT offer us a glimpse: He is too easy though on CAIR's mentality: "Zionist synagogues are your enemy"

1.22.2022 8:43pm - "Is AI Intelligent?" Gee, my reply came out both usable and punchy, so I'm sharing.

1.22.2022 8:21pm - (Replying to @wayneholmes) AI is the road to a higher form human intelligence -- an agent that, in addition to being intelligent also understands what makes her so.

1.22.2022 8:08pm - The Ziohophobic cesspools at CUNY campuses @CUNYkcc are overflowing. My only thought: “History will judge those who were silent as part of the problem and those who took action as part of the solution,”

1.22.2022 3:55pm - Pfizer CEO Albert Bourla awarded $1 million Genesis Prize for 2022 via @timesofisrael
1.22.2022 3:55pm - Hard to think of another person more deserving of the Genesis Prize for 2022.

1.22.2022 3:19pm - Honored to rally for Roe v. Wade with my friends at Zioness -- the only feminist movement that rhymes with Lioness. @ZionessMovement @zioness_lioness

1.22.2022 2:36pm - Another article hit my desk, asking: "Is AI Intelligence?" Remarkably, it takes seriously the proposition that "intelligence always relies on both an understanding of causal relationships and the competence of making practical use of it."

1.22.2022 5:43am - (Replying to @gottfriedmath) Not really. One cannot find in rung2 an answer to the question "What is the probability that it was the aspirin that cured my headache, not the TV program I watched". It's counterfactual because it invokes a world in which I did not take aspirin, conflicting with one that I did.

1.22.2022 5:18am - New arrival at my desk: A comprehensive survey of "Fairness in machine learning" from the perspectives of moral philosophy, sociology, and law.

1.21.2022 10:22pm -
1.21.2022 10:22pm - No need to condemn. All @CAIR needs to do is to take back Billoo's charge, state publicly that: "Zionist synagogues are NOT our enemy" and everyone would vow for your sincerely. Can you?

1.21.2022 9:42pm - (Replying to @artistexyz) Here is a simple derivation of ETT=P(y_0=1|x=1) The key is to write P(y_0=1) =P(y=1|do(x=0) as
P(y_0=1) = P(y_0=1| x=1) P(x=1) +
= ETT P(x=1) + P(y=1|X=0)(1-P(x=1)
This gives ETT in terms of L1 and L2 probabilities.

1.21.2022 9:24pm - (Replying to @artistexyz) I dont see how these derivational steps are justified. Boris post has good derivations for several L3 expressions.

1.21.2022 7:33pm - A good question to explore: To what extent has Ralph Bunch, in his super-urge to achieve cease fire in 1949, contributed (unintentionally) to the creation of the tallest obstacle to peace - UNRWA.

1.21.2022 5:58pm - Level 2 probabilities contain either Y_0 or Y_1, not both, nor anything contradicting the antecedent or the outcome. Consider ETT=P(Y_0=1| X=1) =P(y_x'|x) and recall how it is treated (for binary X) in Causality p. 396-7 P(Y_0=1|Y=0) is, likewise, Level 3.

1.21.2022 2:30pm - (Replying to @soboleffspaces @EpiEllie and 5 others) I find P(y_x, y'_x') to be more convenient in lengthy derivations, clearly showing that we are looking at quantities defined at Level 3 (not Level 2).

1.21.2022 2:21pm - Congratulations to @saramagliacane , and to the Journal of Approximate Reasoning for launching a #Causalinference area. This Journal used to be dominated by fuzzy logic and other non-probabilistic approaches. Glad it opens its gates to #Causalinference.

1.21.2022 1:08pm - (1/2) I believe much of the confusion in this thread stems from the fact that we use "counterfactuals" (and "unit-level") to mean two different things. (1) counterfactuals to the exclusion of level-2 and level-1 expressions, and (2) counterfactuals including level-2 and level-1.
1.21.2022 1:08pm - (2/2) To avoid confusion, I suggest we distinguish the two by using "level-3" for the exclusive interpretation and "counterfactuals" for the inclusive interpretation.

1.21.2022 12:13pm - (Replying to @HenningStrandin @ingorohlfing and 4 others) This is what I mean by being "unclear". Woodward defines all his concepts in terms of the causal graph (me 2), but does not define what the absence of an arrow means in the graph. Zero change in distribution is the wrong interpretation of arrow absence.

1.21.2022 12:03pm - ( Replying to @soboleffspaces @EpiEllie and 5 others) The equivalence: E{Y_x} = E{Y|do(X=x)} is true, but what is "we should stick"? Is it a restriction?

1.21.2022 10:36am - (Replying to @EpiEllie @AndersHuitfeldt and 5 others) The notational distinction represents an important substantive distinction: Those lucky PO expressions that can be written as do-expression represent causal quantities that are estimable from experimental studies. Those that can't, need functional assumptions; or only be bounded.

1.21.2022 4:54am - (Replying to @causalinf @HenningStrandin and 3 others) Identification is a separate issue from representation. The former is model dependent, the latter isn't. You cannot express PNS as a do-expression regardless of whether the model permits identification or not.

1.21.2022 4:50am - (Replying to @HenningStrandin @soboleffspaces and 3 others) Good question. "Counterfactuals " is a language within which many types of statements can be expressed, Eg,interventioinal, mediational, PNS, or PN... Some can be reduced to population properties (eg interventional) and some cannot (eg. PS).

1.21.2022 12:20am - (Replying to @soboleffspaces @causalinf and 2 others) Comment: Mediation is level-3, because you cannot express the magnitude of an Indirect Effect as a do-expression. See Graphically, you cannot remove the direct arrow by a do-surgery on a set of nodes.

1.20.2022 4:39pm - In memory of Ilan Halimi, I am reposting my Eulogy of 2006

1.20.2022 4:09pm - Leading UN Watchdog Says Facebook Has Dropped a Threat to Delete His Account Over Taliban Post
1.20.2022 4:09pm - This news made my day. @HillelNeuer is one of the few sane voices remaining at the U.N. Evidently, some Taliban-inspired employee of Facebook decided that Neuer's Tweet-joke was too offensive for the sensitive stomachs of his peers at @benandjerrys and threatened his account.

1.20.2022 12:44pm - Thank you, @ISPORorg for inviting me to speak to "trial folks" -- my first "target trial" -- "All my trials, Lord, soon be over." (Joan Baez, 1960's)

1.20.2022 12:32pm - Nice display of the hierarchy. One comment: In the example of Level III, I would say "Would the grass be wet had the sprinkler been off...". The subjunctive mode "had been" already entails the "making sure" operation.

1.19.2022 10:49pm - (Replying to @Nichogachet) I have heard it from many researchers, from all fields. You really have to be a masochist to psych yourself into "I could do without them", as some famous people do.

1.19.2022 10:42pm - (Replying to @MandyZong) Thanks for telling me. I wasn't able to attend, but will request a video.

1.19.2022 10:40pm - (Replying to @another_DataSci) DAGs are relevant to tasks like those described in the first chapter of #Bookofwhy, no more no less. See If you can take these tools and apply them to GR and QFT, you would benefit these two fields.

1.19.2022 10:29pm - (Replying to @joshuaafein and @drjgauthier) Whatever I add to your Tweet would sound like book-promotion, so I am tempted to keep quiet. On the other hand, I feel I should tell readers that I am very happy with the way the book came out - I just did.

1.19.2022 10:20pm - (Replying to @boredyannlecun) "Apparently some researchers are working on "electric trains", said one steam locomotive to another. "Meanwhile, you and I have changed the modern world." They did.

1.19.2022 10:05pm - (Replying to @scotmorrsn) I would not recommend this publisher to any potential author, they act as though they are out of business, except for one warehouse in Kamchatka. That warehouse should, God willing, ship the eager world a revised and corrected printing for 2021.

1.19.2022 8:43pm - To all causality-minded readers who have been using Primer in the classroom and had trouble solving any of the homework problems - rejoice! We are now making the Solutions Manual available to all: So, let wisdom light the darker corners of our world.

1.19.2022 6:08am - I have some comments, too long for Twitter. e.g., why talk about "treatment" in the "association" column? e.g., I would not resort to counterfactuals in the "effect of causes " column; causal Bayesian Networks are sufficient, based on intervention only. Minor points.

1.19.2022 5:31am - (Replying to @soboleffspaces) Right. But I consider these features to be corollaries of the basic miracle: path interception ==> conditional independence. Probability theory was good to us.

1.19.2022 5:24am - CAIR's dilemma: How to tell which Synagogues are off-limit and which are associated with Israel. Trouble is, all have the Star of David on their buildings, and they all pray: "give peace to the land of Israel, and perpetual joy to its inhabitants" -- normalizing colonialism.

1.18.2022 8:04pm - (Replying to @yabadabadoo1212 @RafAKazimi and 8 others) That's what Zionophobes think, yes. And that's why Zionophobia is a mental deformity.

1.18.2022 5:53pm - (1/2) "Why DAGs work" is a profound question that can be answered at several levels. Conceptually, DAG's work because (1) the logic governing counterfactuals is captured by the surgery operator (as shown in the First Law) and (2) the logic of conditional independence is captured
1.18.2022 5:53pm - by path intersection in graphs (as shown in the Second Law, and the theory of graphoids Both are gifts from heaven.

1.18.2022 3:03pm - (Replying to @MandyZong) This is the Zoom link I found:

1.18.2022 1:56pm - Tomorrow, Jan 19, 11am-12pm, my UCLA colleague, Kal Raustiala, will zoom-discuss the key role that Nobel Peace Prize winner Ralph Bunch played in the early days of Israel.

1.18.2022 6:43am - (Replying to @joanvjoseph) Can't see that. Placebo effect does not show on the Ladder of Causation, and I can think of many other reason whey DAGs work.

1.17.2022 4:36pm - (Replying to @anastasicamille) The British media has not received CAIR's 2021 guidelines:"Zionist synagogues are your enemy." It is also unaware that we are still waiting here, on Twitter, for @CAIR to rescind this statement, and that I've predicted they won't, b/c they can't.

1.17.2022 4:25pm - Happy Holiday. Rumor has it that the Ficus tree I planted on Tu Bishvat 1941, in the main street of Bnai Brak is still standing there. Can some insider verify it? Happy Holiday!

1.17.2022 1:13pm - Sharing Moshe @vardi 's Logic Day lecture "From Greek Paradoxes to Political Paradoxes" which made me fear (at the very end) that perhaps we should stop all works on "personalized medicine", "unit-level response" and what observational data reveal about you and me. Scary.

1.17.2022 12:31am - Cool indeed! It appears like personalized medicine is unraveling before our very eyes!

1.17.2022 12:24am - I came across this cute Tweet on Confounding Variables, which has attracted 6,355 'likes'. I immediately thought: God knows how many control-minded researchers could benefit from the "Crash Course in Good and Bad Controls". So, here it is, reposted:

1.16.2022 10:29am - (Replying to @RafAKazimi @swon26 and 7 others) This is the same "condemnation" style we heard from CAIR: "We love Jews", "We love their place of worship". Only "Zionist Synagogues are our enemies". Please challenge @RashidaTlaib to distance herself from the latter -- she can't! because the word Zionist makes Jews a people.

1.16.2022 9:03am - At the very minimum, "Zionophobia" would transform the nature of our narrative -- from seekers of protection to rightful claimants. It would no longer be dismissed with "Jewish paranoia," and the haters would be stripped of their "we love Jews" mantra.

1.16.2022 8:35am - The Brandeis connection should not escape the attention of Brandeis leadership. Every pathological graduate is a potential stain on the university reputation, for it reflects, however minimally, the kind of culture that was allowed to ferment on campus.

1.16.2022 8:23am - (Replying to @hmatejx) The semantics of feedback system is well defined in SCM, but, with the exception of linear systems, we lose the d-separation tool, so the analysis becomes harder.

1.16.2022 8:18am - The Jewish leadership, sorry to say, contributes to this deceitful tactic by not adapting the word "Zionophobia," so as to expose the deceit and highlights its moral pathology.

1.16.2022 12:32am - (Replying to @Nayborhoodbully @ZachLewis3187 and 5 others) I believe the world would forgive CAIR for past stupidities if they only state: "Zionist synagogues are NOT our enemies". As simple as that. But they won't, because they can't, they would lose their entire membership.

1.15.2022 9:44pm - Condemnations are easy. What's hard for @CAIRNational , in fact impossible, is to take back their guiding credo: "Zionist synagogues are your enemies" which, today, does not sound too ecumenical. I am still waiting...

1.15.2022 8:06pm - Confirmed! What a relief!

1.15.2022 7:26pm - (Replying to @TriciaRoth18 @CAIRNational and 3 others) That's because Jewish organizations have been charging them with antisemitism, which they can cheerfully deflect with empty condemnations. Try charging them with Zionophobia and watch their response... incredible watching ... Now try to convince @ADL & @AJC to use this word.

1.15.2022 7:06pm - (Replying to @MSchneebaum @isaacdecastrog and 5 others) For @lsarsour , @CAIRNational , and many other Zionophbes, hate towards Zionism, Israel and truth is but a noble cause of the "resistance." Holy Ghost! Thinking that @SenSanders designated her his "Middle East Advisor" blows my mind.

1.15.2022 6:38pm - To be taken seriously in its condemnation, @CAIRNational must address the statement by CAIR + Linda Sarsour,+ Zaharia Billoo: “Zionist synagogues are your enemies”. All it takes is to publicly rescind this statement. Can you? We are listening.. @bandlersbanter @HenMazzig @ADL

1.15.2022 6:18pm - @WajahatAli , If you are sincere in your plea against conspiracy theories, you must address the statement by CAIR + Linda Sarsour,+ Zaharia Billoo: “Zionist synagogues are your enemies”. Addressing means telling your readers what Zionist synagogues mean to @WajahatAli . Listening..

1.15.2022 4:26pm - A PhD from Brandeis University! In her trial, Afia Siddiqui demanded that all prospective jurors be DNA-tested and excluded from the jury at her trial "if they have a Zionist or Israeli background." I feel totally dejected.

1.15.2022 9:00am - (Replying to @quantitudepod and @tkaiser_science) Confessing: I peaked into the control variables episode, and enjoyed tremendously listening to hard problems that our crash course attempts to simplify: Please have a look and tell us if we succeeded.

1.15.2022 7:57am - A begged RT: Christianity wouldn't have become a world religion had its narrative stopped at the crucifixion rather than the resurrection. Holocaust museums, from which most Americans get their only glimpse of Jews, allocate zero floor-space to Israel - the Jewish resurrection.

1.15.2022 1:52am - Slight correction. I am not a signatory to the professors Letter, though I endorse every word of it. I am at UCLA, not USC. The latter rejected me when I was looking for a job. But I have never complained -- I actually deserved it.

1.15.2022 1:40am - Holocaust education has two huge blind spots: (1) The Arabs (+Brits) role in preventing Jews from escaping Europe. (2) The Yishuv role in building a state to receive the survivors.

1.15.2022 1:08am - (Replying to @DubinPamela) Donors and parents know this now, but, as I understand it, part of the reason for ignoring the issue, by both the administration and the leadership was to keep it out of parents/students knowledge, lest they get overly alarmed.

1.15.2022 1:02am - (Replying to @RishabhVaid8) I misunderstood you. The numbers were obviously fictitious, but the story describes two studies that could in principle be conducted with the outcomes described. Moreover, the numbers must be what they are if you buy the explanation of two requirements, English and Math.

1.14.2022 5:55pm - USC President Announces Measures to Combat Antisemitism on Campus via @jewishjournal
1.14.2022 5:55pm - (1/2) Correlation or Causation? What's common to the reactions of all three: @Hillel , @ADL , @AJC ? Ans. 1. None mentions the 65 professors who risked their jobs. 2. None mentions Israel or Zionism. 3. None suggests effective implementations of Folt's plans. My wild prediction:
1.14.2022 5:55pm - All three will be honored with tall seats at the newly established "Advisory Committee on Combating Antisemitism," while students and faculty continue to wonder: "Are we welcome on a campus we have helped build?" @DavidHarrisAJC , @JGreenblattADL , @HillelIntl , @PresidentFolt

1.14.2022 3:51pm - (Replying to @shripadt) I don't follow your argument. Why "all variables" if the graph tells you that a small set Z is sufficient? Unless you don't trust the graph. So, what do you trust? Don't tell me "data" !!

1.14.2022 2:42pm - What I love about this clip is how Arabs, the victims of E. Said "orientalism," turn around with authoritative face to re-define what their neighbors are.

1.14.2022 2:33pm - (Replying to @GalitPeleg) That coast of Jaffa has been stormy for thousands of years now, since Jonah took that boat trying to run away from his higher calling. Lesson: Listen to the waves and follow your calling.

1.14.2022 2:27pm - Six CUNY Professors Sue Faculty Union Over 'Antisemitic, Anti-Israel' Positions
1.14.2022 2:27pm - Any lawyers among our readers? I wonder if a student can sue a university for diverting part of his/her tuition to pursue political advocacy far outside of his/her department's educational mission, as specified in the university brochure. (I have two UCLA departments in mind).

1.14.2022 1:56pm - (Replying to @FJnyc) If we do not expose their exclusionary policies, using academic representation on their "Boards", they will eat us alive. DEI colonies are already consuming 20-30 million dollars (average) on each campus and they are growing like mushrooms.

1.14.2022 1:41pm - (Replying to @MatheusFacure @artistexyz and 3 others) When you say: " I would love to see a simple way of combining experimental and observational data" you need to specify what purpose you hope the combination to serve. Our purpose here: was to learn individual response.

1.14.2022 3:36am - (1/2) Two funny things about President Folt's Letter: (1) She completely ignores the Letter of the 65 professors who initiated the push-back and jolted her to issue a public statement and even mention the forbidden word Zionism twice. (Up from Zero throughout USC history)
1.14.2022 3:36am - (2/2) (2) In the newly created "Addressing Antisemitism" webpage, the "Frequently Asked Questions" are formulated by the administration staff, not by the public. The real "critically asked questions" are not in sight. This webpage should be made open to USC students and faculty.

1.13.2022 8:58pm - (1/2) My Initial response: Momumental achievement of 65 USC professors who, unlike the established campus leadership, did not try to hush-hush this Zionophobic incident. However, the professors' key question: "Are Zionists welcome on this campus?" is not answered in Folt's Letter.
1.13.2022 8:58pm - (2/2) Moreover, past "Advisory Committees" have served as cover-ups for inaction. Their effectiveness depends on whether Folt staffs them with professors from the 65, or with the usual hush-hushers. Ditto Jewish representation on DEI Board, usually reserved for hush-hushers.

1.13.2022 4:03pm - The #Bookofwhy was written as one big tutorial, but if you are looking for a more technical introduction, don't miss the Primer or any of the dozens review articles listed here: at your finger tips.

1.13.2022 12:18pm - (1/3) You hit it on the nail: Statistics 101 book from the future. This is precisely what I had in mind inwriting #Bookofwhy. So, here is an even simpler, more compelling example of why combining observational and experimental data should inform us about individuals behavior:
1.13.2022 12:18pm - (2/3) It's called "The Curse of Free-will and the Paradox of Inevitable Regret" Here, the reason that free choice turns out disastrous is familiar to all of us: Laziness. We also understand why the consequence of this tempting quality of human nature
1.13.2022 12:18pm - (3/3) can only be revealed by comparing observational and experimental data. The dictatorial nature of the latter proves its virtue by comparing the average performance of the oppressed to that of the un-oppressed (or un-disciplined). Any one study, alone, cannot do it.

1.13.2022 11:29am - (Replying to @soboleffspaces @artistexyz and 6 others) What's ESL?

1.13.2022 10:32am - (Replying to @yudapearl @zakkohane and 8 others) As the Talmud says: Anti-Zionism is never Antisemitism to an Antisemite, and Antisemitism is never Antisemitic to a Zionophobe.

1.13.2022 10:27am - (Replying to @zakkohane @DShaywitz and 5 others) This has nothing to do with antisemitism. It has to do with the fact that we don't have a word to label this kind of moral deformity, and Jewish organizations like Hartman Inst., @ADL , @AJC , even @bariweiss refrain from using the only word we do have: Zionophobia.

1.13.2022 10:03am - The link was broken for some readers, here is a good one:

1.13.2022 2:27am - Glad to see practical applications of the theory of transportability: which ML folks call "domain adaptation" or "transfer learning" @atafti , @gshmueli , @eliasbareinboim , #Bookofwhy,

1.13.2022 1:46am - (Replying to @gottfriedmath @artistexyz and 4 others) You are probably not "astray". I just do not understand the problem. Best to follow Scott's "preliminary example" to understand why RCT+observational data inform us about individual-level behavior.

1.13.2022 1:34am - Yes, it's true. I'll be speaking at the ISPOR Conference January 25, on: "selection bias, personalized treatment effect, fusion of data from several sources (observational and experimental studies), and causality in observational studies." #Bookofwhy

1.12.2022 11:05pm - (Replying to @FJnyc @KeshetGLBTJews and 13 others) We live in an era where "refusing service" is how we serve a noble cause. Rumor has it that McDonald refuses Big Mac (or was it Big Duck) to customers walking like pedophiles. The company could not find a more convincing way of projecting its virtue. It's still in business.

1.12.2022 10:37pm - (Replying to @artistexyz @soboleffspaces and 6 others) The title JudeasRx made me call may mail-order pharmacy and ask why my prescription has not arrived. "Six more business days"--the said. I wish I could enjoy your program, but the notation throws me off. Also, Is the data real or fictitious?

1.12.2022 10:12pm - I'd take it back. There is a simpler example here: to illustrate why the combination of experimental and observational data can inform us about individual behavior. It shows how a simple observation can rule out the model of "no effect on any individual."

1.11.2022 7:43pm - (Replying to @Shakeelr and @nntaleb) My eternal mantra: What is the research question?

1.11.2022 9:00am - Every time I see the cover of #Boodofwhy I am reminded that, yes, I wrote it (with Dana), and that, yes, it's a heck of a good book. Thanks for sparkling the present with good memories!

1.11.2022 8:57am - (Replying to @adamsafron @GeorgeLakoff and @sapinker) This is a great project for an ambitious linguist, to look in the evolution of the word "because" and try to correlate it with technological advances in various cultures.

1.11.2022 7:40am - I cannot think of a simpler example than this: And, if you want to dress it with stories about "firing squad", "Radiation & Leukemia" or legal liability, see Section 9.3 in Causality

1.11.2022 7:20am - (Replying to @deboerk07) translation = ?

1.11.2022 7:14am - (Replying to @artistexyz @MatheusFacure and 3 others) On the contrary. The greater the bias, the more informative observational data become towards narrowing the bounds.

1.11.2022 1:19am - (Replying to @jstreetdotorg and @AIPAC) To be honest about Israel's wrongdoings, and to write about it with some credibility, you need to do more than just join the many packs of howling hounds. You need to also call out the Industry of Lies. This is what @jstreetdotOrg , @InfNotNow, @NewIsraelFund rarely do.

1.11.2022 1:10am - (Replying to @djmayer18) Some Holocaust Museums do include an exhibit on life in the post-war DP camps. But few pursue their torturous voyage to Israel, and none asks how Israel was created to be able to receive them and provide them with home and new life.

1.11.2022 1:00am - (Replying to @HanaBalshin @sullydish and 2 others) To be honest about Israel's wrongdoings, and to write about it with some credibility, you need to do more than just join the many packs of howling hounds. You need to also call out the Industry of Lies. This is what @jstreetu , @InfNotNow, @NewIsralFund, @sullydish rarely do.

1.10.2022 10:38pm - Beautifully said! Was it me? Yes! So, here is a personal request to all followers: Please forward it to the Director of Holocaust Museum closest to you. Yes, the one who asked you for end-of-year donation. You can sign for me, or your own name, or Aristotle, but make them move!

1.10.2022 9:07pm - 'Jews of Color' category divides the Jewish people - opinion
1.10.2022 9:07pm - I'm not convinced that pushing the "Jews of Color" argument is a wise move. Israel has been a living proof that noble forces, stronger than skin-color can bond people together. Imitating the American obsession with racial division is madness.

1.10.2022 8:23pm - And to think that this entire Jezreel Valley was one big swamp only one hundred years ago (Nahalal was founded Nov 2021), full of malaria infected mosquitos, makes your heart pray with awe: What a miracle!

1.10.2022 2:43pm - Agree. It was a good talk. And I need to say it, else I'd be accused of hiding personal biases.

1.10.2022 3:39pm - (Replying to @rina31420183) ( Judea Pearl -- The Foundations of Causal Inference [The Book of WHY] WHY-19)

1.10.2022 3:13pm - Sidney Poitier is dead - sad!. I had the fortune of having dinner with him in 2008, but he didn't tell me the "Sunday Morning" story below.

1.10.2022 2:56pm - Agree with the choice of books. But, let's stop and think: these and other books have not stopped Andrew Sullivan @sullydish from pretending he never heard of the medical convoy of 1947 and the Arab invasion of 1948. A professional reporter gone blank -- how come?

1.10.2022 1:07pm - (Replying to @artistexyz @causalinf and 3 others) I love the picture, and I agree that "utility" is a better word than "benefit function".

1.10.2022 12:56p - (Replying to @CGische) Oh, No, have a look at this: 1.7.1 Counterfactuals in Linear Systems *

1.10.2022 12:40pm - (Replying to @vejaocentro) Not in covid context, sorry. But the best paper on "effect of treatment on the treated", IMHO, is this: the potentials of which are waiting to be discovered.

1.10.2022 12:32pm - This now leaves Econometrica the last flagship Journal whose pages haven't seen a causal graph, nor a do-operator. How about collaborating on modernizing econometrics?

1.10.2022 9:01am - (1/2) Very nice paper. Glad you managed to publish it in Psychometrika - the journal that rejected my paper three times, and made me feel guilty for depriving Quantitative Psychologists of ever learning graphical models. They should be thanking you for lifting them to modernity.
1.10.2022 9:01am - On the technical side, I believe the next step should be the analysis of counterfactuals which, in linear models, should yield neat solutions using matrix algebra together with the First Law Fusion would be another fruitful direction. Congratulations!

1.10.2022 8:07am - I had a different take on the difference between epidemiology and economics: The former are more advanced, in that they use graphs and are guided by the First and Second Laws of causal inference: Most economists are still in PO era

1.10.2022 7:59am - (Replying to @hangingnoodles) DAG's have several limits. (1) Being acyclic, unable to represent feedback -- SCM can do that. (2) Being propositional, unable to represent "object-property" relations, needing predicate calculus lift-up.

1.9.2022 11:11pm - (Replying to @offaltube) We are all guilty of a tendency to wishfully forget the obstacle to the 2nd, which is at least as formidable as the 1st. I was taught to fight for 2-state solution since I was 3 yrs old. We are still fighting, though less aggressively now that we understand the real obstacle.

1.9.2022 10:02pm - (1/2) There are two delusions, widespread in the West, that made Sullivan's questions sound so depressing: (1) The feasibility of 1-state solution if it weren't for the world fixation with the 2-state goal. (2) The feasibility of 2-state solution if it weren't for Israeli policies.
1.9.2022 10:02pm - The first would be fought to the death by 90% of Israelis who won't give up a self-rule earned after 2,000 years of bloody longing. The second would be fought to the death by 99% of Palestinians who can't betray their national DNA - the elimination of Israel.

1.9.2022 8:07pm - (Replying to @FJnyc) I was depressed too by Sullivan's ignorance and gullibility. But this is the kind of people we need to educate. That's why I found it instructive to see how YHK handled it; we need all kinds of bullets in our pouches.

1.9.2022 7:19pm - Strongly recommended!!! 30 minutes of sanity to clear you mind from the dust of populist slogans.

1.9.2022 3:30pm - (1/2) If I understand your questions correctly, they may be of general interest: a) Does "do calculus" assume "ceteris paribus" i.e, all else remaining fixed? Yes, with a caveat that "all else" needs a DAG to be well-defined; we certainly do not want to fix the effects of X .
1.9.2022 3:30pm - (2/2) b) are human causal models (eg implicit physics) graphical? Ans. In some sense, Yes. Human models are built upon the elementary notion of "who listens to whom," which defines a graph. On the other hand, unaided humans are incapable to processing these mental models
1.9.2022 3:30pm - in their heads, so as to answer elementary questions such as: what makes good controls different from bad controls? or, is the model testable? Only economists (joking) can do it in their heads; that's why they shun graphs, as explained here:

1.9.2022 12:44pm - (Replying to @wayneholmes) Trust, like information and dis-information, is an acquired commodity -- I'll be waiting patiently. As to the late Desmond Tutu, a person that sees "apartheid" in Israel is as trustworthy as those who see a stolen election in 2020 & no slavery in Sudan:

1.9.2022 11:57am - (1/2) @SabinaLeonelli , to understand relationships between "model" and "data" we need a theory of modeling, namely, SCM. There is truly no conceptual limit to the "design based approach" because, to delineate a "thing," we need to understand what the "thing" is. Champions of
1.9.2022 11:57am - (2/2) the "design-based approach" have thus far not explicated, in general, what the approach IS. In the specific cases where they did, as in the applications of IV, it turned out to be one limited corollary of Structural Economics. See, Section 4.

1.9.2022 5:03am - I find "model-based" and "model-averse" economists talking past each others. The former ask: "What scientific principles SHOULD guide economic research?" The latter respond with "What makes model-free studies POPULAR among model-averse economists." Same in

1.9.2022 4:21am - (Replying to @PHuenermund @KaiGehring1 and @causalinf) What's the Imbens-Deaton debate?

1.9.2022 4:07am - (Replying to @KaiGehring1 @causalinf and @PHuenermund) I agree with your addendum. Revising my reaction: I would add that "design" may also be choices made by a mental model which the designer can't make, or refrains from making explicit.

1.9.2022 3:55am - Initial reactions to Card: (1) I wish I could convince him to read #Bookofwhy (2)Put under a microscope, "design" may end up being: "choices made by intuition", so it's hardly a "methodology" nor an "approach". (3) Card reinforces my take, Section 4 of

1.9.2022 2:45am - I remember Peter Szolovits and @Tshortliffe very well. Your post compels me to retweet my personal recollections from that prehistoric period:

1.9.2022 2:41am - (Replying to @jwbelmon @NoahHaber and 2 others) I think you are craving for the word "mediation", unless there is more that you are craving for. I would check this survey

1.9.2022 1:02am - (Replying to @ompdamani and @VC31415) Agree. Given the intense debates regarding vax policies, and the availability of analytical tools to manage policy questions, it would be illuminating to apply the latter to the data available. I'll be glad to contribute my biases to any such exercise.

1.8.2022 7:35pm - Why is it that disinformation, well groomed and charmingly feathered, can propagate faster than truth itself and mesmerize even the best of minds, especially in academia? Does it take KGB-size budget to undo Moscow's fabrication of "Israel apartheid"???

1.8.2022 6:31pm - (Replying to @adababbage and @drob) I hope it turns into a non-spurious transformation.

1.8.2022 8:02am - On the positive side, the counterfactual analysis needed for settling debates on path-specific effects has become a second language in certain circles of epidemiology. Here is a gentle and comprehensive survey:

1.8.2022 5:14am - (Replying to @BethCarey12) As viewed in my lens, causality falls out of the world, and causal talk is an important source of information on how people view the world and what made them successful in navigating it. By all means.

1.8.2022 4:53am - Amazing that this debate started 200 years before Covid-19 and was not settled in the reign of data science.

1.7.2022 9:04pm - Biden administration remains 'deeply blind' to inherent flaws of UNRWA, say Mideast experts via @JNS_org
1.7.2022 9:04pm - Feeling good vs. doing good. How the US State Department cultivates the next generation of conflict in the Middle East.

1.7.2022 2:19pm - (Replying to @NeilFoxLaw and @blakeflayton) Zionism was born with the Exodus, Judaism at Mt. Sinai. Jewish-born Zionophobes argue that Jews who continue the historical dream of Return to Zion are causing them social-acceptance problems among non-Jewish bigots. I lose sleep over them; they are in bad shape indeed.

1.7.2022 12:54pm - (Replying to @SAFECUNY and @theamyalbertson) I would go even further and, given the opportunity, would respond with surprise and indignation: "Don't tell me you are, God forbid, a Zionophobe". An Al-Jazeera reporter once asked me (in Doha): "Are you a Zionist?" My answer: "Aren't you? Isn't everybody?". He never recovered.

1.7.2022 4:59am - Here is a most penetrating and informed iddle East analysis, both conflict and solution, in 27 minutes: The first to draw the logical conclusion of two facts: 1.Peace = Two States 2. Palestinian Neural Net = One (Pal.) State Conclusion: Peace = Rewire NN.

1.7.2022 4:39am - (Replying to @VC31415 and @soboleffspaces) Agree. Semantics is orthogonal to identifiability or observability.

1.7.2022 4:38am - (Replying to @VC31415 and @soboleffspaces) Agree. Semantics is orthogonal to identifiability.

1.7.2022 4:36am - (Replying to @VC31415 and @soboleffspaces) The meaning of an arrow depends on whether you are looking at a Causal Bayesian Network, Rung-2, or an SCM, Rung-3. See chapter 1 of Causality: Section 1.3 vs Section 1.4.

1.7.2022 4:27am - (Replying to @ntraft @_WEEXIAO and 12 others) Agree. The symbols vs. gradient-based debate does not address the more basic question of information content: Do the data available contain the information needed for estimating the query asked? The reason it's rarely asked in ML (unlike CI) is absence of a theory to answer it.

1.7.2022 4:15am - (Replying to @artistexyz @zaffama and 3 others) The purpose is to answer the corresponding questions under the stated observations. Agree, PN invokes only Y_0 and PS only Y_1.

1.7.2022 4:09am - (Replying to @zaffama @artistexyz and 3 others) do(x) is a controlled action, but not necessarily in randomized trial. A farmer asking: I wonder what my yield would be if I used fertilizer-1 vs. fertilizer-2 is asking E(Y|do(fertilizer-i))=?, while a farmer reported to be using fert-i will reap on average E(Y|fert-i).

1.6.2022 5:56pm - (Replying to @StandWithUs and @missmayim) They need a push, a tiny one, like the one jolted by @missmahim

1.6.2022 4:04pm - This may be a challenge to some. There are, however, other challenges, e.g., to examine whether the information available in the data is sufficient for certain tasks, regardless how cleverly one manipulates the data, be it with or without continuous differentiable functions.

1.6.2022 3:45pm - (Replying to @GaryMarcus @tejasdkulkarni and 12 others) The reason many think we do not need world models for scene perception is that, unlike causal reasoning, we do not have a mathematical theory of scene perception to prove them wrong.

1.6.2022 3:39pm - Gives me an idea: If more celebrities start boasting their Zionist identity on Wikipedia, Zionophobes should get the idea that they are but a passing deformity, like yesterday's smallpox.

1.4.2022 7:59pm - (Replying to @dylanarmbruste3) My professors heuristics incapsulate the idea that, if you want to know whether a task is doable with certain type of data, look not at how smart the analyst is, nor at how much data is collected, but at the input information. Certain tasks are provenly undoable w/o certain input

1.6.2022 1:23pm - I strongly suspect they have imported these Oryxes from Africa, quite recently, because I haven't seen them when I served in the southern Israel, at least not in such numbers.

1.6.2022 1:05pm - (Replying to @LekhtNaya) I wish I could click on a "double like".

1.6.2022 5:49am - (Replying to @zaffama @tdietterich and 2 others) The expression "had I not take aspirin" means "ceteris paribus", namely, everything else remaining unaltered, my medical history, my chronic migraine, including my urge to take aspirin each time I have a headache. And this is what Y_x' represents, as defined in the First Law.

1.5.2022 10:42pm - (Replying to @swarat @GaryMarcus and 12 others) Yes. This paper contrast the data-fitting vs. data-interpretation paradigms in data-science.

1.4.2022 9:16pm - Just watched, Learned a lot, Highly recommended!

1.4.2022 7:59pm - (Replying to @dylanarmbruste3) My professors heuristics incapsulate the idea that, if you want to know whether a task is doable with certain type of data, look not at how smart the analyst is, nor at how much data is collected, but at the input information. Certain tasks are provenly undoable w/o certain input

1.4.2022 7:43pm - (Replying to @roydanroy and @tdiettericha) It is not even a critique of DL, only of passive-data ML. Inference in RL is based on interventional data which brings us to Rung-2 of the Ladder (almost). Still not to Rung-3, the level required for deep understanding.

1.4.2022 5:19pm - (1/3) Here is the latest critique of pure deep-learning Another realization that "intelligence is more than statistics" and that data-fitting is just "“a fluent spouter of bullshit”, or “stochastic parrots.” The article is informative & well-written, however
1.4.2022 5:19pm - (2/3) I would have liked to see an acknowledgement that AGI is not as helpless as the blind spots of data-fitness would make us believe. The past 2 decades have given us a computational model of deep understanding that is free of such blind spots. I am referring, of course, to SCM
1.4.2022 5:19pm - (3/3) which, though limited to the causal corner of our wood, demonstrates nevertheless how the data-centric yokes of DL can be circumvented, and how human-level reasoning about physical reality can be encoded, algorithmicized, and put into novel practical usage. @cnaintelligence

1.4.2022 4:28pm - (Replying to @thegreatdukd) "Representation Learning" is more sophisticated causal discovery method which extracts directionalities from multiple environments. Like traditional causal discovery, the limits of its performance depend on the structure of the model-set that is equivalent to the one discovered.

1.4.2022 2:55pm - (Replying to @AiSimonThompson) I find it hard to be amazed. Imperfect fit is still a "fit". The input is a set of data points and the output is a function between variables. Fancy names such as "inductive bias", "regularization", "split data" do not change the input-output relation. Still amazed?

1.4.2022 2:13pm - (Replying to @soboleffspaces @PHuenermund and 2 others) "Direct Causal Effects" are always defined for a given granularity of details. Each arrow in a DAG represents Zillions of micro processes, including feedbacks, that transmit the effect of X on Y. By assuming that arrows are conditional probabilities we lose directionality.

1.4.2022 2:08pm - I made this remark 4 years ago, expecting a flood of counter arguments by DL advocates. So far, none convincing. We hear about "learning to fit" "learning to learn to fit" etc. etc. But my engineering professors warned us: Think input-output and you won't fall for circular traps.

1.4.2022 1:55pm - (Replying to @MeganTStevenson @EpidByDesign and @DrBobGoldberg) "Good Causal Inference Research" requires no more, no less, than what is absolutely required to solve the research problem stated. The proper use of calculus in physics requires the same substantive understanding and internal dialogues. The analogy CI = calculus is a good one.

1.4.2022 8:16am - (Replying to @Moshepupik1 @ShawnDreyfuss59 and 9 others) How else would you tell Peter Beinart (or @IfNotNowOrg ) : "Even if you do not hate Jews or Judaism, even if you are Jewish yourself, even if you lay Tfillin twice a day, you are still morally deformed". This is what "Zionophobia" says and "antisemitism" does not.

1.4.2022 7:04am - (1/2) (Replying to @EpidByDesign and @MeganTStevenson) Questions phrased as "What proportion of X are Y" are indeed purely predictive, purely statistical, necessitating no causal consideration. However, most questions in the social social and medical sciences are motivated and marketed as policy questions. Moreover, even purely
1.4.2022 7:12am - (Replying to @yudapearl @EpidByDesign and @MeganTStevenson) predictive questions meet obstacles that require causal thinking to be circumvented. Examples are: missing data, multiple data sources, transportability and more. Thus, what remains is traditional statistics, not much.

1.4.2022 6:30am - (Replying to @AiSimonThompson and @kareem_carr) There is a difference between "accounting for maladies" and "overcoming maladies". Theory can only tell you whether an "accounted for" malady can be overcome, and how. Moreover, it also tells us that no one can do better, including those who refuse to "account for" maladies.

1.4.2022 6:14am - (Replying to @DavidJonesBrain) The theory of personalized decision making does not negate a culture of personalized predictive features, the two should run side by side. Our post shows in fact how the former can guide the latter in its search for the most predictive feature combinations.

1.4.2022 1:28am - (1/2) For readers interested in Personalized Decision Making, we have updated our blog with a clearer and formal distinction between (1) the response of a specific individual in a population and (2) the response of a population resembling that individual.
1.4.2022 1:28am - (2/2) The former is defined in Eq. (3) which, unlike the latter, cannot be point-estimated by RCTs. This clarification should hopefully answer some of the questions raised by readers of the first version. @smueller

1.4.2022 12:16am - (1/ ) (Replying to @zioness_lioness @femmerrhoid and 9 others) @femmerrhoid is expressing aspirations that drive many readers on the left, e.g., "multiethnic federation put in place that respects the rights of all ethnicities". Unfortunately, when it comes to ME, such aspirations meet a basic logical contradiction: One ethnicity demands
1.4.2022 12:27am - (2/ ) (Replying to @yudapearl @zioness_lioness and 10 others) the right to deny another ethnicity the rights it claims for itself. Sounds circular? It is. So how do we disentangle it if we really want to be nice to everybody? The Zionophobic solution is simple: Eliminate one side and logical consistency is restored. The Zionist solution
1.4.2022 12:41am - (3/ ) (Replying to @yudapearl @zioness_lioness and 10 others) is different: assign priorities to demands. Specifically, value the right to self-determination above other rights (including the right to deny it to others) and continue from there. It's logical consistent, true, but, alas, we are no longer "nice to everybody". Specifically,
1.4.2022 12:57am - (4/4) (Replying to @yudapearl @zioness_lioness and 10 others) We are no longer nice to the Palestinians who, since 1922, claimed that Jewish self-determination is lower priority, and should yield to Palestinians right to sole ownership of the holy land. This, I think, is what @femmerrhoid decries so emotionally. My sympathies.

1.3.2022 1:59pm - My contention has been: A "science of data" is hardly a science unless it means "science of reality, as reflected through data" The taxonomy you've asked for is part of "Data Fusion" Note how nicely data sources are annotated.

1.3.2022 1:39pm - (Replying to @metamnisia) In the case of Israel, "Rights" is the relevant issue. Being the only state in existence that is attacked on the pretense that, though it exists, it has no right to exist, Israel insistence on "rights" is a litmus test of Palestinians sincerity of intents.

1.3.2022 10:45am - (Replying to @mistagogue) Rung-3 is the mathematical framework where imagination takes place.

1.3.2022 8:03am - Well put. Today, any administrator who agrees to meet to discuss antisemitism says, in effect, "Let me enter an endless debate as a cover for inaction". Sincere administrators have the courage to name the disease: "Zionophobia". I hope @LordJohnMann does not fall for this trap.

1.3.2022 6:43am - My interview is Wednesday, Jan. 5, 11:30am Pacific. See you!

1.3.2022 2:15am - Speaking about imagination and invention, here is a nice article, beautifully summarizing the spread of the major invention among our ancestors:

1.3.2022 12:27am - (Replying to @soboleffspaces @mygoodjeans and @nltmcd) My immediate questions: Does their taxonomy of conditionals correspond to anything we know from the theory of counterfactuals? Can the latter refine or disambiguate the taxonomy used by ELT folks?

1.3.2022 12:14am - Idea: Instead of preaching to Israelis how they ought to defend themselves, we spend our time showing them that what they see is not there, namely, that there exists ONE Palestinian leader, cleric, intellectual, or educator willing to accept a post-occupation Israel BY RIGHT.

1.3.2022 12:01am - (Replying to @SamuelJWeinberg @Ya3il8200 and 2 others) We are not talking Hamas and Iran, but about eternal Palestinian vows, permeating all walks of life, meticulously monitored by Israelis, to eliminate Israel by all available means. Given that this is what Israelis see, advising them how to defend themselves sounds a bit unfair.

1.2.2022 10:53pm - (Replying to @ccaballeroh10 and @soboleffspaces) Very interesting article, straight to my "to read" list (6 ft high).

1.2.2022 10:47pm - If one or more of X's parents is unobserved, you can't use the truncated factorization directly. The graph structure may permit us to replace X's parents by other sets of covariates, namely backdoor adjustment. See Primer Fig. 3.6

1.2.2022 10:27pm - (Replying to @SamuelJWeinberg @Ya3il8200 and @blakeflayton) Yes. But Arab persistent vows, publicly and covertly (mosques and education), to continue and seek Israel's demise once the occupation is lifted, makes lifting a bit harder than what appears from the populist slogans of BDS and their cronies at @IfNotNowOrg .

1.2.2022 10:02pm - This sweeping assertion came to me from reading Harari's "Sapiens", in 2015, but the more I study the Ladder of Causation and its ramifications, the more I believe the assertion is right, i.e, humans would not have reached such revolutionary dominion without Rung-3 of the Ladder.

1.2.2022 7:49pm - (Replying to @soboleffspaces @mygoodjeans and @nltmcd) Here you are doing the counterfactual thinking in your head, based on what we know about Archimedes. But formally, the sentence "If L then M" does not imply "L = false". Moreover, even "If L then M" & "L=false" together, do not imply "M=false". e.g, the earth could move by magic

1.2.2022 3:50pm - (Replying to @LahavHarkov and @HananyaNaftali) I am hearing reports that Chris Williamson @DerbyChrisW is a real person, not a Hamas operative.

1.2.2022 3:10pm - (1/2) I like this line: "Would you like us to interpret the Koran for you?". Not that it would make any difference to hard-core Zionophobes like Nomad and Avi, but it will make a big difference to hundreds, perhaps thousands of silent readers on this channel who are wondering:
1.2.2022 3:10pm - (2/2) "How should I respond to a Zionophobe pretending to know and define (for me, my friends, or child) what Zionism stands for. I know many such readers exist, because they write to me privately to continue and share my experience in exposing this new breed of inverted Orientalism.

1.2.2022 11:12am - Who said "data are profoundly dumb"? Who said numbers are too dry to capture a miracle?

1.2.2022 10:51am - (Replying to @soboleffspaces @nickchk and 5 others) What's his conclusion? I mechanism necessary? Does he deal with justification on a personal level too?

1.2.2022 10:44am - (Replying to @soboleffspaces and @nltmcd) One thing to keep in mind: Wolves are super hare-catchers, but they haven't invented a hare-trap; for this we need counterfactual thinking, rung-3.

1.2.2022 8:53am - (Replying to @soboleffspaces and @nltmcd) Fully agree!! Philosophers preoccupation with interpreting other philosophers is an indication of helplessness in solving problems of epistemology. Computer scientists cannot afford the luxury of such helplessness because knowledge must be represented and used.

1.2.2022 8:38am - (Replying to @UnSethled) Zionism, like any organic historical process does not need any defense, it is amoral imperative. I am defending the use of the word Zionophobia, to expose the moral deformity of those who attack Zionism.

1.2.2022 8:32am - (Replying to @DJMinded) Was Zion ever granted self-determination by its neighbors? Not to my recollection. I have been there all along, and can't recall a day of normalcy. Thus, the problem seems to revolve around denying Zion self-determination, both physically and rhetorically, since 1937.

1.2.2022 8:18am - (Replying to @UnSethled) If you only knew how many writers are making careers out of proving that "anti-Islam", "anti-Islamism", "anti-Gay" or "anti-women" is rational and necessary, you would be more careful with the words "rational and necessary".

1.2.2022 8:02am - (Replying to @UnSethled @Nomad_Olive and 9 others) I don't believe you are lol. Most likely you are expressing discomfort facing a mirror after years of basking in a bubble of self-righteousness. The next step is to vilify others as "terrorists", "shameful" and other populist slogans. The next is another look at the mirror.

1.2.2022 7:38am - (Replying to @UnSethled) Zionism, like any organic historical process does not need any defense, it is amoral imperative. I am defending the use of the word Zionophobia, to expose the moral deformity of those who attack Zionism.

1.2.2022 7:21am - (1/2) This argument, that the word Zionophobia, by sounding like Islamophobia, conflates animosity towards a people with prejudice against religion was voiced by several writers, friends and foes. My answer: Religion has no monopoly on individual or collective identity.
1.2.2022 7:21am - (2/2) Zionism, as a realization of a people's aspiration for self determination, historical revival and restored normalcy is no less deserving of reverence and protection from discrimination or vilification than any ancient religion, its rituals or symbols.

1.2.2022 5:53am - (1/2) Another phenomena to notice among anti-Zionists upon hearing the word "Zionophobia" for the first time. They can't bear the thought that anyone would consider them less than morally pure, let alone racists. Shocked into disbelief, they immediately seek group therapy
1.2.2022 5:53am - (2/2) in the form of foul language, slogan chanting, or populist slurs and hyperboles. It must be profoundly traumatic to face a mirror after years of basking in a bubble of self-righteousness.

1.1.2022 8:00pm - (1/2) Zionophobes cannot bear the thought that their logic is based on the racist premise of E. Said's Orientalism, i.e., that they alone are indigenous natives, hence everyone else must be a "colonizer", forever barred from even describing the native culture or aspirations.
1.1.2022 8:00pm - (2/2) They alone have the right to define others, to decide who is a "colonizer" and to tell right from wrong. Some choose "power" to be the defining criterion of a "colonizer", thus making the powerless French minority in Algeria "colonized," and the ruling Algerians "colonizers".

1.1.2022 6:10pm - (Replying to @soboleffspaces) As a computer scientist, I couldn't leave the puzzle where Hume and Lewis left it, without asking: How is it that people form consensus about counterfactuals which are (by Def) unobserved? Equivalently, what do we need to feed a robot so it answers counterfactual questions ?

1.1.2022 1:02pm - I wouldn't put it that bluntly. First, there two Humes: (1) Causation = Covariation and (2) Causation = Counterfactuals. Second, I object to Hume-1 (like everyone else). Third, I agree with Hume-2 but go further to ask "Where do counterfactuals come from? & how are they stored?" This sweeping assertion came to me from reading Harari's "Sapiens", in 2015, but the more I study the Ladder of Causation and its ramifications, the more I believe the assertion is right, i.e, humans would not have reached such revolutionary dominion without Rung-3 of the Ladder.

1.1.2022 11:50am - (Replying to @Hewett1Tim) "Generate" from judgment or from data?

1.1.2022 3:48am - (Replying to @MairavZ) This statement ("you can't fight antisemitism with Zionism,") unveils ignorance of both, contempt for the latter and vindication of the former. 65 USC professors have just proven the opposite: The moral force of Zionism makes antisemitism doubly shameful

12.31.2021 11:27pm - Happy New Year to Everyone. May the new year brings peace to the world love to our hearts and wisdom to our brains. Shanna Tova.

12.31.2021 11:42pm - (Replying to @TaliaRinger) It's the mood that counts. Shana Tovah.

12.31.2021 11:24pm - Glad we have such moments of reflection, where we amplify blessings, suppress adversities, and compress it all into a 2 min watch. This must be the scientific definition of optimism.

12.31.2021 4:01pm - For Math-AI folks, I will be interviewed by Stephen Wolfram at the 2022 SIAIM (Intl. Symposium on Artificial Intelligence and Mathematics) (scroll to bottom of page). It's a heavenly gift for my 85th birthday, for which I am most honored and grateful.

12.31.2021 12:29pm - USC administration's response to anti-Semitism comes under the microscope via @JNS_org

12.31.2021 12:29pm - (1/3) Finally, the real issues underlying the USC scandal are brought under the microscope. This may be my last Tweet this year, so, may that it sparks and spreads an academic pushback on many US campuses, fueled by the principle that:
12.31.2021 12:29pm - (2/3) "University administrators unwilling to publicly name a problem are not in a position to solve it". Corollary 1: Working with Jewish organizations behind the scene, forming "task forces" or "working groups" and erecting "anti-hate" cathedrals are but cover-ups for inaction.
12.31.2021 12:29pm - (3/3) Corollary-2: EDI (Equity Diversity and Inclusion) offices lacking Jewish-Zionist representation operate contrary to the principle of "Diversity & Inclusion" and are totally useless until reformed. Corollary-3: Kudos to 65 USC professors who started this process. Happy New Year!

12.31.2021 9:57am - Even if we exonerate BBC from the charge of antisemitism, what remains unforgivable is their hiring of a Zionophobic editor who succeeded, perhaps covertly to distort reality in the style of BDS operatives.

12.30.2021 8:36pm - Anuradha Mittal, given her upbringing, deserves a variant of my recent tweet: Those who compare the Palestinian war to eliminate Israel with anti-colonialism are dishonoring the latter, inflaming the former, and prolonging the suffering of two peace-deserving peoples.

12.30.2021 8:07pm - (1/2) In my prism, we do have epistemological theories already, at least embryos of such theories, and we are far from "fumbling in the dark". The benefit of philosophical discussions would come to light by showing ONE (not more!) question that philosophers identified which cannot
12.30.2021 8:07pm - (2/2) be answered with current computational models. Alternatively, showing that lingering philosophical debates can be resolved by computational models would extremely valuable to both communities. I hope you pursue one of these two avenues in your dissertation research.

12.30.2021 7:43pm (Replying to @EinatWilf) Don't forget your admirers in Golat America; Dorshei Shlomech V'Hem Yetter Adaraich,

12.30.2021 7:35pm - Those who compare the Palestinian war to eliminate Israel with the war against apartheid are dishonoring the latter, inflaming the former, and prolonging the suffering of both peoples.

12.30.2021 5:38pm - Aside from self-congratulation and self-calibration, end-of-year tallies help students gauge if they are getting into thriving or declining fields. In our case, I'm delighted to assure them of the former, and to welcome all 15K new followers to this educational channel.

12.30.2021 4:10pm - 2021 all-time records of CI books:
1. Probabilistic Reasoning has reached 30,000 citations:
2. Causality reached 20,000 citations, and all time record of 2,050 in 2021
3. #Bookofwhy, all time record of 662

12.30.2021 3:29pm - Moment Magazine tells us that the 6th most read story in 2021 was a fruit of this year's Daniel Pearl Investigative Journalism Initiative: A stunning story by Tom Gjelten into the Chinese government treatment of the Uyghur people

12.30.2021 1:07pm - (1/2) (Replying to @legalytical and @jon_ckhan) Interesting questions. We all know that lawyers have had hard time both presenting statistical evidence in court as well as understanding the strenth of evidence, not to mention combining evidence. The question arises, would causal analysis make things easier or harder?
12.30.2021 1:13pm - (2/2) (Replying to @yudapearl @legalytical and @jon_ckhan) An argument in favor of "easier": Our mind is a causal, not statistical machine, so, having to process statistical relations is hard. Argument in favor of "harder": Most lawyers take a class or two in statistics, and none in causal analysis so, even the easy looks hard.

12.30.2021 1:17am - (Replying to @MarcSosnick) Michael Chayim, I don't believe in God, but I believe in prayers. So I am going to send you a man-made earthly prayer, full of good hopes for your fast recovery. Be well, and stay the course Mi Haish HeChafets Chayim Ohev Yamim, Lirot tov JP

12.29.2021 9:08pm - I want to believe that many liberal students among our readers will be empowered by this opinion piece by Elior Amar. It is part of the recent Zionist pushback on college campuses, and it contains several gems that need be kept in mind and made widely known. Hats off, Elior.

12.29.2021 7:11pm - (Replying to @soboleffspaces) Wider audience? Perhaps the pharma industry is a wide-enough audience to satisfy Paul's vision.

12.29.2021 7:01pm - (Replying to @augustwester) I have been tempted several times to join the Popperian discussions but got discouraged upon realizing that I am too biased by my new religion that clarity of discussion requires computational models of notions such as "knowledge", "facts", "ought" "theories" and "conjectures".

12.29.2021 6:28pm - I have an visionary offer to propose to @tala666 , the killer-dreamer. I volunteer to burn in hell through my entire afterlife, if you and your people agree to let us, Israelis, live peacefully a few years in our humble earthly lives. Deal?

12.29.2021 3:28pm - (Replying to @timoreilly) Any link to George Saunders' story?

12.29.2021 1:10pm - The Bayesian notion of updating priors has become so entrenched in our mind that it appears to echo every act of knowledge updating. Careful sniffing of details, however, reveals that "priors" differ from "knowledge", see

12.29.2021 12:49pm - (Replying to @jaredlander @hillel and 3 others) It is not the scarcity that impedes us but what the energy we must spend and games we must play to fight over those resources.

12.29.2021 12:42pm - I'm not sure statistical literacy entails the ability to recognize which claims are or aren't of a statistical nature. I (and u) know some highly literate statisticians who claim that everything is statistical in nature. (Not much different from data-centric ML paradigm.)

12.29.2021 12:26pm - (Replying to @itsrainingdata) The mapping is on page 232 of Causality. The equivalence is based on Halpern complete maximization, and I remember a Twitt by @analisereal that gives the proof in less than 285 characters.

12.29.2021 10:57am - (Replying to @timoreilly) I admit ignorance of Tse's theory of free will while writing #Bookofwhy. It seems that Tse's theory had a rather mild reception (93 citations on Google Scholar). More importantly, I believe what free will needs now is a computational, rather neurological basis, as outlined p.358.

12.29.2021 10:31am - (Replying to @jaredlander @hillel and 3 others) Theoretically, the more voices the better, if they are coordinated, and if some thinking goes into effective division of labor. My observation: the many voices we currently have compete over same resources and, unintentionally, render each other ineffective. Volumes to tell...

12.28.2021 11:22pm - (Replying to @RaphMikeyTurtle @_charadams_ and @Evolutionistrue) You the first person I know who can't predict the result (selection bias).

12.28.2021 8:13pm - Naive question: Isn't @Hillel doing all this fighting already? In their end-of-year message they claim they do. So do @AJC , and @ADL , and SWU and @SSI and more... How many organizations do we need?

12.28.2021 4:10pm - UN launches unprecedented open-ended investigation of Israel via @Evolutionistrue
12.28.2021 4:10pm - I have offered one scientific explanation here but I'm afraid this psychosis will remain the focus of intense research for generations of pathologists, therapists, psychiatrists and rare-disease specialists. It's Beyond Science -- COVID-22

12.28.2021 3:51pm - There is another explanation to the UN obsession with Israel bashing: post-nationalism. Most UN delegates, brought up on Marxist ideologies cannot forgive Israel for proving their textbooks wrong; a national homeland and a strong successful democracy must be guilty of something.

12.28.2021 2:12pm - @USC @PresidentFolt made it to Wiesenthal Center's list of 2021 top ten antisemites. w/ @JVP and @CAIRNational I wish there was a separate list of top ten Zionophobes, because Folt's main wrongdoing is refusing to spell Zionism:

12.28.2021 11:38am - (Replying to @LonesSmith and @JNS_org) Those who use Tutu's death to trumpet the myth that the Palestinian war to eliminate Israel is morally equivalent to fighting apartheid in South Africa are doing more damage to his legacy than those who exonerate him as an innocent victim of misguided intersectionality.

12.28.2021 11:28am - (Replying to @LonesSmith and @JNS_org) Those who use Tutu's death to trumpet the myth that the Palestinian war to eliminate Israel is morally equivalent to fighting apartheid in South Africa are doing more damage to his legacy than those who exonerate him as an innocent victim of misguided intersectionality.

12.28.2021 10:41am - (Replying to @DrBobGoldberg and @EmmanuelDetrini) Oh, that makes sense. The Seven Tools, admittedly, were assembled to counter Stigler's "Seven Pillars of Statistical Wisdom" which I have found to be lacking one tiny wisdom - causality.

12.28.2021 9:50am - Jokes aside! This is basically how @USC student Jasmeen describes her state of victimhood and "un-safety",; a petition to protect her "safety" is being circulated.

12.28.2021 9:28am - (Replying to @acastroaraujo) If you have a clear, educated and good-faith definition of "intersectionality," please share, and we can seek another explanation for what made Desmond Butu speak so inappropriately against Israel.

12.28.2021 9:14am - (Replying to @LonesSmith and @JNS_org) When a Peace Prize winner dies, and the slander machine of BDS starts trumpeting his most inappropriate (if not de-humanizing) utterances against a being-denied society, someone should defend him and remind people that even a most decent person can be lured into such blunders.

12.28.2021 8:31am - (Replying to @lukasvermeer) A theory is what connects among the experiments, eg, it drives one to ask: Given the disappointing result just obtained, I wish we could conduct the following experiment....

12.28.2021 8:10am - (Replying to @DrBobGoldberg and @EmmanuelDetrini) What are the 7 simple rules of causation? I usually name only 2, the First and Second Laws.

12.28.2021 8:06am - (Replying to @acastroaraujo) If it wasn't intersectionality, what explains how a humanist like Archbishop Tutu could be lured into a path of blindness and immorality?

12.28.2021 12:37am - (Replying to @LonesSmith and @JNS_org) Everyone I know in Israel is concerned about this misery. Yet for a humanist like Desmond Tutu to demonize 8 million human beings, while ignoring the openly eliminationist plans of their neighbors, takes more than an innocent blind spot.

12.27.2021 10:45pm - (Replying to @LonesSmith) I hope you agree that even a wonderful man can fall prey to deceptions; Tobin's articles documents Tutu's mistakes and details their painful consequences.

12.27.2021 8:15pm - Archbishop Tutu and the disturbing power of intersectionality via @JNS_org
12.27.2021 8:15pm - The siren songs of intersectionality can lure even a decent man like Archbishop Tutu into a path of blindness and immorality.

12.27.2021 8:05pm - (Replying to @DebyNavarroR @rivkacjoseph and @lsarsour) Dear Deby, you made my day. If you say that Boric can "actually think", I would like to send him a copy of #Bookofwhy. Spanish translation.

12.27.2021 9:38am - (Replying to @rkoenker @jmwooldridge and 2 others) This is why we have counterfactuals, with the help of which we can launch an "Undoing Project" and undo every confusion that do-calculus has left undone.

12.27.2021 1:22am - (Replying to @AustenCF and @FreeformMethod) You may have other Jewish features, but some anti-Zionist Jews are indeed "by birth only". So, go ahead, define yourself the way you want but, please, let Zionists define themselves as they want, separated from you, and do not say "As a Jew" because you are not speaking for them.

12.26.2021 11:47pm - (1/5) As the year draws to a close, many are looking back on the moments from 2021 that gave them hope and encouragement. Kenneth Markus lists his 10 Most Inspiring Moments in the fight against #antisemitism: I would like to add another, missed by
12.26.2021 11:47pm - (2/5) the media, suppressed by our leaders, and hush-hushed by university administrators. Yes, I'm coming back to the USC scandal, and the 65 of its top professors who scored a huge victory through this letter: Here they defined in effect a new
12.26.2021 11:47pm - (3/5) minority group on US campuses: Identity Zionism, defined not by victimhood and need of protection, but by its inspiring nation-building experience and ideas, its academic excellence, and its unique contributions to campus life. To appreciate the magnitude of the victory,
12.26.2021 11:47pm - (4/5) please scroll down the Letter to "Reactions", and note the 26 items of media coverage. EVERY USC student now knows about the outspoken 65 professors who broke through heavy walls of blackout, and told students: "We have your back, we won't let Zionophobes hijack our campus."
12.26.2021 11:47pm - (5/5) Another fallout of this publicity lies in cornering the @USC administration to stop hiding behind the vagueness of "antisemitism" and state a clear, unambiguous position on Zionist students, as articulated here: The next two weeks should be critical.

12.26.2021 9:53pm - (Replying to @AustenCF and @FreeformMethod) I thought you were complaining that Zionists attempt to speak for all Jews. I replied: "quite the opposite", Zionists try to separate themselves from non-Zionist Jews, and speak for themselves only; begging: "Please allow us to define ourselves."

12.26.2021 8:22pm - (Replying to @cas_wognum and @UUBeta) Stay the course.

12.26.2021 8:18pm - (Replying to @AustenCF and @FreeformMethod) I said "quite the opposite", namely, you were incorrect.

12.26.2021 8:14pm - (Replying to @cas_wognum and @UUBeta) Thanks for alerting me to this course. I hate to generalize, but I find Europeans less constrained by dogmas, tradition, and authority, hence more likely to advance causal inference in the future.

12.26.2021 6:17pm - This would be easy to determine. People who buy into the First Law understand where counterfactuals come from (see Those who don't, walk in the dark, like Sobel and Imbens, assuming that potential outcomes and ignorability assumptions are God given.

12.26.2021 5:54pm - Sobel's course is rather dogmatic. Unless Sobel has converted to modernity, the course is likely to be harmful in its religious aversion to graphs, to structural models, and to modern way of thinking. See why:

12.26.2021 3:35pm - Jason Roy is a co-recipient of the 2021 ASA Causality in Statistical Education Award:, so his course should be illuminating. I forgot however if he buys into the First Law and the Ladder of Causation. Please remind me.

12.26.2021 3:24pm - (1/2) Another inspirational birthday - 85 yrs to the Israeli Philharmonic Orchestra (IPO) As a 3-month old baby, I missed the opening concert w Arturo Toscanini, 1936, but I remember Isaac Stern in Mendelson's Concerto (1949) and Itzhak Perlman narrating
12.26.2021 3:24pm - (2/2) Peter & the Wolf. Every October, since 2002, Maestro Zubin Mehta has dedicated one IPO concert to the Daniel Pearl World Music Days. We'll forever salute his vision, artistry and humanity, while admiring the IPO for coloring our history with music, excellence and meaning.

12.26.2021 1:47pm - Quite the opposite. We, Zionists, are laboring days and nights to distinguish "Judaism" the religion from "Jewishness" the peoplehood, from "yesterday Jews" or "Jews by birth only". We promise to classify you in the latter, if you promise never to say: "As a Jew". @FreeformMethod

12.26.2021 8:08am - (Replying to @JooTube @JCCWatch and 4 others) These days we urgently need ONE sort of ammunition -- bullets cast in "Zionophobia". Any day that passes while these well-meaning organizations are refraining from using this word is a day wasted, and time is running out.

12.26.2021 7:54am - What She (@lsarsour) meant is obvious: "We (Palestinians) are determined to deny any Jewish connection to the history of the Holy Land, so as to make ignorant people believe we are indigenous and they are NOT." I've challenged her to say the words "equally indigenous"- she can't!

12.25.2021 11:27pm - (Replying to @dpovey1) From everything I've read about Jesus, he had a rather low tolerance for hypocrites, thus I think even his infinite patience would snap at Linda Sarsour's latest deceit.

12.25.2021 10:46pm - A thoughtful essay by @YehudaKurtzer on the ideological basis for continued Jewish-Muslim dialogue. This was indeed the basis for the Daniel Pearl Muslim-Jewish dialogue that Professor Akbar Ahmed and I have been conducting 2002-2012. Perhaps 2022 may be a good time for renewal?

12.25.2021 10:04pm - For historical accuracy, Jesus would have sued Linda Sarsour for defamation over the label "Palestinian Jewish". At his time, the Philistines were still regarded as the sworn enemies of the Jewish people, foreign invaders and irredeemable pagans; un-fakers of Linda's Christmas.

12.25.2021 7:39pm - Christmas needs more reasons to celebrate, and every new reason adds to those already celebrated. BTW, I've been fortunate to see the original of Newton's portrait below, at the Malibu home of my late friend and book-collector Al Seckel.

12.25.2021 6:04pm - Linguistic analysis of EDI publications is but one test of a Zionophobic mindset. Another is the exclusion of Jewish representation from EDI Boards & Councils. Crying out loud: Requests by @USC students to add a Jewish student to the Provost EDI Council have NOT been granted.

12.25.2021 4:42pm - I'm happy the media has focused on her "kill every Zionist" and not on "curse the Jews". This is what makes the @USC circus so unique and hopeful -- to defend the most abused, abandoned & clearly defined minority on campus, by ending diversionary debates on what antisemitism is.

12.25.2021 4:02pm - (Replying to @soboleffspaces @smueller and @Ang_UCLA) Yet, speaking counterfactuals, I can't help but ask: How would William James have phrased it, had he lived today, equipped with the theories and languages we have to understand and distinguish the many abstractions of data, data-analysis, and data-generation processes.

12.25.2021 12:13pm - Readers watching the USC circus as a laboratory for the future of higher education would find recent updates to be intriguing. The final section "Reactions" on now lists 26 media articles and letter exchanges about an incident meant to be suppressed. Why?

12.25.2021 11:14am - The puzzle of counterfactual is, not that we can think about "roads not taken" but that, with the little we know about the world, we swiftly form a consensus about where those roads would take us. @EmilyReider #Bookofwhy See

12.25.2021 2:56am - The media is still mesmerized by the vulgarity of @USC student Jasmeen, which has reached new height this week. They seem surprised that, viewed from Palestinian bubble of sainthood, vulgarity spells "resistance", and death threats amount to "intifada". It's a big bubble.

12.24.2021 10:23pm - I believe Clarence did use a DAG, since his "roads not taken" were not "manipulable", hence their effects would be "ill-defined" in the potential outcome framework. Since George Bailey had no problem seeing these effects, he must have used same mind-DAGs as Clarence.

12.24.2021 10:06pm - (Replying to @soboleffspaces @Ang_UCLA and 8 others) I can't find SAP, Lemma 5, r488,etc. sorry.

12.24.2021 9:58pm - (Replying to @soboleffspaces @smueller and 8 others) Before we answer this question we need to agree on what "treatment effect" we consider. The Individual Causal Effect ICE(u) = Y(1,u)-Y(0,u) is the same by definition, regardless of how probabilities are assessed, since individual u sees the same hospital & nurses in both trials

12.24.2021 9:45pm - (Replying to @GalitPeleg) Happy to hear that Haifa has not changed since my Technion years, 1957-1960. My home-town, Bnei-Brak did change. We had about 4-6 Christmas trees in 1960, in private homes -- I don't think I could find one today. They even changed the name of Herzl Street, to show contempt.

12.24.2021 8:51pm - It has been a while since I could identify with every word spoken by an Israeli Foreign Minister. To all my Christian friends worldwide, @yairlapid words are my words. Merry Christmas!! Feliz Navidad !! Feliz Natal!!

12.24.2021 4:47am - (Replying to @stephensenn @soboleffspaces and 7 others) Surely examples exist. My question was:if placebo effects vary from study to study, what can be said about treatment effectiveness on the population that eventually adopts the treatment once all studies are done?

12.24.2021 4:19am - (Replying to @soboleffspaces @smueller and 7 others) Last email I received from Steyer he was still trying to mine causality from probabilities. Is the book a change of mind?

12.24.2021 4:11am - (Replying to @smueller @soboleffspaces and 7 others) The only case I know of where P(y_x|z) would not be causal effect is when z is affected by X, as exemplified in Primer for z mediator. Gender seems Kosher to me.

12.24.2021 1:40am - (Replying to @roberto_limongi and @connormcmk) I wouldn't be s o cruel, though it would be helpful to view causality as an uprising, not a supplement to the teachings of traditional statistics.

12.24.2021 1:25am - Haifa, where I spent my college years was always the stronghold of Zionist socialism. Yet the multicultural spirit of the holiday season was genuine and over reaching.

12.24.2021 1:11am - Hilarious , if it wasn't so sad ~ two zionophobic camps germinating from same misguided jewish education

12.24.2021 12:46am - Sorry to see Billoo gone.seriously . She gave honest picture of the mindset of every CAIR member. Now we are back to listen to disingenuous leaders who say one thing to the camera and another to their members. We'll miss u Billoo.

12.23.2021 2:52pm - (Replying to @stephensenn @soboleffspaces and 6 others) I'm curious to learn what “concurrent control” is. Plus, if something varies from study to study, what measures can be taken to insure that effects estimated under RCT applies to the target population under normal use of treatment?

12.23.2021 2:32pm - (Replying to @stephensenn @soboleffspaces and 6 others) The key take_away is: “What counts is more than body counts.”

12.23.2021 10:36am - (Replying to @connormcmk) I wish I had. Advice: stay on Twitter and ask. Your questions will ( eventually) be answered.

12.23.2021 10:25am - (Replying to @kareem_carr) Causal inference, however, is not a separat e attempt, but the conceptual thread behind all mentioned attempts, some successful and some less, to enrich stat w something it could not handle

12.23.2021 10:07am - (Replying to @soboleffspaces and @smueller) Module some pathological cases, amen!

12.23.2021 2:59am - Hats off! @SecBlinken and @yairlapid

12.23.2021 2:53am - This Zionist group deserves your support ~ follow them and retweet. They don't have the typical Hang ups of seeking acceptance among zionophobes, nor among donors or yesterday Jews.

12.23.2021 2:40am - Historical statement by Tony blinken, recognizing the transformational potentials of the #AbrahamAccords, defying the zionophobic squad, and pushing for vigorous expansion of its influence. Hats off, Tony!

12.23.2021 1:52am - I I don't know about you but as someone who was brought up on the instinct of siding with the underdog I don't know how anyone can refrain from defending Israel in times like these.

12.23.2021 1:39am - True followers of Christ stand with oppressed people everywhere, including Israelis, who were denied a day of normalcy for the past 73 years. Pray for peace and coexistence.

12.23.2021 1:12am - Two years ago I started asking anyone I could what “target trial” is. Today, i am still asking.

12.23.2021 12:51am - I think what you call “reinventing” are in facts attempts to liberate it from its data-centric handcuffs, as defined by Pearson (1911) and fisher (1925).

12.23.2021 12:34am - Correction: Antizionism=zionophobia=racism Why go the round~about way if u can go straight to the point? Recall: antiSemitism is a cover for inaction to universities, and diversionary debate to our enemies

12.22.2021 7:38am - There is some progress in this suggestion to offer Palestinian students mental health consultation, but it still falls short of decriminalizing USC elite professors.

12.22.2021 7:26am - Qatar enormous success in corrupting the minds of our universities should worry our port authorities.

12.22.2021 7:21am - (Replying to @lukasvermeer) Experimentation without theory is craftsmanship.

12.22.2021 7:13am - Public indictment of Kenneth Roth

12.22.2021 7:00am - (Replying to @CombatASemitism @CampusFairness and @USC) Too late and missing the point. Focus on how president Folt pretends she can't hear what USC faculty are asking

12.22.2021 6:48am - Show me a decent person who isn't

12.20.2021 11:25am - (Replying to @stephensenn @soboleffspaces and @smueller) I'm heading to vacation, so I can't participate in this discussion, sorry. If I could, I would have asked, again and again, what is the research question? What types of data are available, etc. Bye

12.20.2021 5:53am - So where is Chile heading? Towards Venezuela? Or towards Canada?

12.20.2021 5:44am - (Replying to @RichardGrenell and @AdamMilstein) The anatomy of the USC scandal is more scandalous than what's portrayed in the media. The main issue is @PresidentFolt refusal to decriminalize its Zionist faculty.

12.20.2021 4:25am - Palestinians know the soldiers will not touch them as long as they don't get violent. They know that Israeli security professionals are not going to hurt them. They would be equally brave to confront a soda bottle, or their own brother wearing a uniform he was told to wear.

12.20.2021 12:22am - (Replying to @DrBobGoldberg and @AishaAriella) Much worse. The wicked child in the Haggadah says: Hey, what is this that you are doing? @AishaArielle says: I despise, not only what you are doing, but who you are, where you came from, your fears and hopes, and I'm committed to smearing your character wherever I go. Much worse!

12.19.2021 11:31pm - As promised, @smueller just posted a revised and clear exposition of our theory of "individual response" on the Data Methods platform. Looking forward to a continuing discussion @stephensenn

12.19.2021 10:12pm - Law Professors Condemn NYU Legal Journal's Israel Boycott, Warning of 'Political Censorship'
12.19.2021 10:12pm - Another cotton bullet. BDS should be condemned, not because it stands contrary to free exchange of ideas, but because it is genocidal in aims and fallacious in practice.

12.19.2021 7:51pm - I bet she is dead honest, and truly "couldn't bear to look at them". Funny, when I call this deformity Zionophobia, colleagues tell me "you are exaggerating, these people are offensive but not mentally ill". Listen to @AishaAriella and tell me if I am exaggerating.

12.19.2021 12:48pm - (Replying to @MatthewNLondon @questionsin2014 and 7 others) I don't know much about it. But you can get a sense of its editor's quality from the opening statement of Azoulay: "When I lived in the Zionist colony in Palestine, Israel.." The rest can be typed by a computer program. "When I lived in the American colony in Mexico, California..

12.19.2021 12:08pm - (Replying to @MatthewNLondon @questionsin2014 and 7 others) They are brainwashing students only when the Boston Review gives them an unearned platform -- her story is boring and repetitive and incoherent and self-centered ... -- if it wasn't for its vitriolic hate towards Israel, it wouldn't have made the pages of Boston Review

12.19.2021 11:45am - (1/2) (Replying to @questionsin2014 @MatthewNLondon and 7 others) I don't see a point in getting nervous at a convoluted account of a self-hating Jew who evidently went through some traumatic identity crisis and is now blaming it on Israel. My wife went through similar transitions and never stopped blessing Israel for restoring her dignity
12.19.2021 11:56am - (2/2) (Replying to @yudapearl @questionsin2014 and 8 others) and Jewish identity in a country she considered her homeland, after escaping from Iraq. And here is an account of a Moroccan Jew, proud of Israel, w/ no hang-ups or identity crisis: For every complaining unfit there are thousands of more inspiring fits

12.19.2021 7:43am - And I can't get over the following puzzle: How come no one asks "what's the definition of Islamophobia?" Users of the term assume that the word is self-defined, self-evident, obvious and axiomatic. Why is Zionophobia not self-defined? Do religions have a monopoly on identity?

12.19.2021 7:24am - Glad you asked, the definition is simple: Zionophobia -- An obsessive animosity toward the idea of a homeland for the Jewish people. I have defined it in more details here:, and For a recent debate:

12.19.2021 7:08am - Zionophobic mentality at its best! And she will continue to spew this hate in the name of noble causes so long as the people she criminalizes ( @ADL , @AJC , @HillelIntl) do not dare tell her how disgusting Zionophobia is. They'd rather fight with cotton bullets named antisemitism

12.19.2021 6:40am - I rarely salute military operations. But in this case, knowing the pain of the victim's family and the super-human efforts it took to catch the perpetrators, I salute the @IDF with all my heart. Techezakna Yedey Kol Acheinu...

12.19.2021 6:30am - Agree with @lewbel . But what promotes this commonly held belief? Two factors: (1) The temptation to ride the ML hype, and (2) Ignorance of the barriers defined by the Ladder of Causation.

12.19.2021 6:20am - 50 years? I bet on 5-10 years.

12.19.2021 6:17am - So why is it so hard for @DavidHarrisAJC and @AJCGlobal etc tec. to spell the word Zionophobia? Zionophobes like CAIR's Zahra Billoo are benefitting from the hesitation, and no one dares telling them how ugly they are. Why?

12.19.2021 6:03am - (Replying to @stephensenn @soboleffspaces and 5 others) Unfortunately, I am not verse in the language of "cross-over" and "parallel group". What is the research question, the assumptions, the data-sources, the feasible interventions? My weakness, I confess, but we need a common language to communicate.

12.19.2021 5:52am - (Replying to @LuisaBorrell) Promising? Do you have an alternative way of escaping the impossible mess of non-causal fairness?

12.19.2021 5:46am - (Replying to @stephensenn @soboleffspaces and 5 others) This is precisely what the science of causation provides us -- a coherent symbiosis of counterfactual assumptions, passive observations, interventional experimentation and counterfactual conclusions. Take away the science, and the 4 components can't communicate with each other.

12.19.2021 5:35am - I like the article, thanks for posting. But I would model "systemic racism" as a moderator, not a mediator. Here is another article introducing causal models to the ML discussion of fairness:

12.19.2021 5:17am - (Replying to @stephensenn @f2harrell and 4 others) Before 1975, people were intuitive counterfactual thinkers, lacking logic to protect them from incoherence and unable to tie their thoughts to empirical data -- very much like gamblers before Cardano.

12.19.2021 5:11am - (Replying to @soboleffspaces @stephensenn and 5 others) There were many "Belling the Cat" skeptics when Leonardo predicted that humans will be able to fly some day. The science of feasibility comes first, a pilot model second (Wright Brothers) and hordes of left-behind skeptics third.

12.18.2021 8:28pm - Another news outlet joins the USC circus, missing the crucial issue: 65 top professors are asking if, as Zionists, they are welcome on a campus whose reputation they've labored to build Yet @PresidentFolt plays deft - she can't spell Zionism.

12.18.2021 7:54pm - My apology was formulated on October 11, 1947, when Azzam Pasha (Secretary General of the Arab League) declared: "a war of extermination and momentous massacre which will be spoken of like the Mongolian massacre and the Crusades." I was only 11, but my parents took him seriously.

12.17.2021 3:54pm - True, but is that the only reason to fight Zionophobia? Suppose non-Zionist Jews grew a horn to make them recognizable and acceptable to Zionophobes. Would that make Zionophobia less disgusting?

12.17.2021 4:23pm - (Replying to @HL327 @stephensenn and 6 others) My impression was that the blog seeks to understand how population averages (which your example shows) can inform individual level response, eg, Joe would benefit from this treatment and Charley would die iff he takes it.

12.17.2021 4:16pm - (Replying to @blakeflayton) I am reading the nasty replies you are getting from Zionophoic bigots, and I can't help thinking it's because you are too polite; you label them "anti-Zionist", as if there is a grain of legitimacy in their deformity. Try using "Zionophobia" and see how they turn self-loathing.

12.17.2021 3:54pm - True, but is that the only reason to fight Zionophobia? Suppose non-Zionist Jews grew a horn to make them recognizable and acceptable to Zionophobes. Would that make Zionophobia less disgusting?

12.17.2021 3:24pm - (Replying to @stephensenn @soboleffspaces and 5 others) Nice example. Can we write this in some mathematical notation? Still, this is a statement about correlations, not causation.

12.17.2021 3:16pm - (Replying to @stephensenn @soboleffspaces and @PavlosMsaouel) And I am still craving to learn one thing they could say about causality aside from the usual: E(Y|treatment) -E(Y|control) = E(Y(1)-Y(0)) (Even here I had to use PO notation.) Please tell us the next thing they could say, just a tiny increment.

12.17.2021 3:04pm - (Replying to @soboleffspaces @stephensenn and @PavlosMsaouel) Not really. Gamblers had the calculus of proportions to guide them, which made Pascal and Fermat solve hard problems, like expected length of a game, way before Kolmogoroff.

12.17.2021 12:35pm - (Replying to @stephensenn @soboleffspaces and 6 others) I am craving for one lesson from the theory of experiment, expressed as a "principle" or "theorem", not as a link to Nelder 1965.

12.17.2021 12:32pm - (Replying to @soboleffspaces @stephensenn and 6 others) Not me, sorry. Because, based on 60 yrs experience, I can predict the result: Reading papers on causation written in the standard language of statistics would end up a waist of time and a painful disappointment. Even the concept of "confounding" would not be defined.

12.17.2021 12:21pm - (Replying to @stephensenn @soboleffspaces and 6 others) I am yearning for principles, not algorithms. How could Nelder deal (mathematically) with causation in 1965, 10 years before Rubin's PO language? Sending me to 1965 does not help. We are in 2021 when, fortunately, we can communicate principles in Twitter's efficiency.

12.17.2021 12:12pm - (1/2) A mini-breakthrough in @USC scandal and a monumental victory to higher education! USC's President Carol Folt has issued a PUBLIC statement mentioning the word "ZIONOPHOBIA" (Take note, UC President Drake, it ain't radioactive).
12.17.2021 12:12pm - (2/2) Why is it merely "mini-breakthrough", and not "grand breakthrough"? Because Zionophobia is mentioned as part of the faculty protest, not part of the President's concern. Most importantly, the 60+ professors still do not know if, as Zionists, they are welcome on their campus.

12.17.2021 12:12pm - (1/2) A mini-breakthrough in @USC scandal and a monumental victory to higher education! USC's President Carol Folt has issued a PUBLIC statement mentioning the word "ZIONOPHOBIA" (Take note, UC President Drake, it ain't radioactive).
12.17.2021 12:12pm - (2/2) Why is it merely "mini-breakthrough", and not "grand breakthrough"? Because Zionophobia is mentioned as part of the faculty protest, not part of the President's concern. Most importantly, the 60+ professors still do not know if, as Zionists, they are welcome on their campus.

12.17.2021 10:43am - (Replying to @soboleffspaces @stephensenn and 5 others) If you are smart, you can do all these in your head and never know you are missing anything. It's only us, mortals, that feel the need to mathematize cause and effects.

12.17.2021 10:36am - (Replying to @MichaelRJonas) My definition of "understanding" is the ability to answer questions on all 3 levels of the Ladder. RCT's reside on Rung-2, unable to handle explanations and individual response.

12.17.2021 10:23am - (Replying to @soboleffspaces @AndrewPGrieve and 5 others) This is my difficulty too in reading DOE literature. It speaks "effects" "influence" "blocking" "cause" "covariation" but has no math language to express these relationships and their nuances, save for the language of covariation -- all causal considerations are done in the mind.

12.17.2021 10:13am - (Replying to @poulastyab) Start with #Bookofwhy, then sail through these: *

12.17.2021 10:10am - (Replying to @stephensenn @soboleffspaces and 5 others) I am thrilled by the prospect of integrating causal and DOE matters but my stumbling block has been absence of a language. For example, do we have a mathematical expression for "blocking", or, what should the world be like for a given DOE to be more effective than another?

12.17.2021 9:59am - What a beautiful quote by Leonardo, which is probably censored in "evidence based" circles who live and die by RCT's and take no "understanding" as evidence or aspiration.

12.17.2021 2:12am - This is what AI needs to spur the next AI Spring. If this happens, perhaps we won't need to move to Timbuktu after all. Recall:

12.17.2021 2:03am - (Replying to @patrickc @Stanford and 2 others) HMM... So perhaps we don't need to move to Timbuktu after all: Perhaps Arc Institute will spur the next AI Spring?

12.17.2021 1:13am - (Replying to @JudahAriGross) @YonahLiberman truly believes that the Zionists vs. Bundists debate has not ended, and hopes that the demise of Israel will prove the latter right. Here is why @IfJews @elderofziyon

12.16.2021 6:19pm - (Replying to @SAFECUNY) Kol Hakavod.

12.16.2021 6:17pm - Another article about the @USC coverup scandal. Again, everyone is enchanted by that saintly Palestinian student who just wanted to "kill every m--f--g Zionist," instead of dealing with the issue: 60+ elite USC professors ask if they are welcome on their own campus. No answer!

12.16.2021 4:54pm - We will be glad to participate in the Data Methods discussion on "individual response" and (1) Concretize the description of the example, (2) Explain why we believe it brings personalized medicine closer to reality and (3) Deal with the practical issues raised by @stephensenn .

12.16.2021 4:42pm - (1/2) If I had unlimited funds, my investment decision would involve other factors besides short-term financial rewards. For example, I'd be interested in being remembered as one who outsmarted all other investors, by predicting that all ML will eventually (5-10 yrs) be driven by
12.16.2021 4:42pm - (2/2) causal knowledge, perhaps even by causal calculus. So, while the rest of the world was busy building cathedrals for data-centric ML, I established a Center for Causal AI, at the university of Timbuctu, which eventually became the MIT+Microsoft of human-level AI.

12.16.2021 1:58am - Sharing my inspiration for the day: Daniel Pearl Magnet High School production of 2021 World Music Day: Enjoy:

12.16.2021 1:39am - Some say, and I'm inclined to agree, that recent tweets from the new Bundists (eg @IfNotNowOrg ) are but gasping attempts to rise from the dustbin of history and shake their guilt over betraying their European brethren. Israel's demise would prove them right - so they hope.

12.15.2021 9:05pm - (Replying to @JLinAffect) Good question. (1) Deep understanding, namely, ability to answer questions from all three rungs of the ladder (2) Social Intelligence, namely, ability to understand (ie, answer such questions about) other agents,

12.15.2021 7:55pm - Here is another avoidance statement from USC President @PresidentFolt , speaking "Jewish" and "antisemitism" to 60 top faculty who scream "Zionophobia". And the circus continues: Even the @latimes missed this avoidance acrobatics

12.15.2021 7:23pm - The circus at USC continues; its top faculty speaks Zionophobia and USC President @PresidentFolt answers with "antisemitism", as if she is deft or can't spell or doesn't really believe Zionist students are welcome on USC campus ???

12.15.2021 2:37pm - (Replying to @stephensenn @RWJE_BA and 6 others) Yes, this called "consistency", ie., a person who dies if avoided a drug by choice would also die if assigned placebo in the trial. Without consistency there is no theory to counterfactuals , see Sometimes written Y = X Y(1) + (1-X) Y(0)

12.15.2021 11:07am - "Discrimination" is a legal term. But Zionophobia is also an ugly and socially unacceptable moral deformity. This must be emphasized, especially in conversations with Zionophobes.

12.15.2021 10:57am - (Replying to @stephensenn @RWJE_BA and 6 others) I do not understand what changes from one trial to another. Are the arrows along the direction of time?

12.15.2021 7:32am - (Replying to @stephensenn @AndrewPGrieve and 5 others) Granted that every RCT is marred by hundreds of imperfections, it is still the holy grail of experimental science. Now, in order to understand what additional information an observational study can add, we skip the imperfections and consider (provisionally) an ideal RCT.

12.15.2021 6:16am - (Replying to @stephensenn @AndrewPGrieve and 5 others) Isn't this what we call "selection bias" ?

12.15.2021 6:14am - (Replying to @stephensenn @AndrewPGrieve and 5 others) The assumption of exchangeability between treatment and control groups is key to any experimental trial. So why should it be any different when comparing experimental and observational studies, conducted at the same time, on the same population.

12.15.2021 5:57am - (Replying to @stephensenn @AndrewPGrieve and 5 others) Yes, this is indeed the only assumption underlying our analysis, ie, that the participants in both studies are selected randomly from THE SAME population. If the RCT suffers (as usual) from selection bias, the results should be modified to correct (if possible) for such bias.

12.15.2021 5:47am - A thoughtful essay by Eric Mandel @MepinOrg on Israel and American Jews. One of a handfuls to recognize that "the strength of the connection to Israel is as much about our survival as it is about Israel's."

12.15.2021 5:28am - (Replying to @f2harrell @stephensenn and 4 others) I hope @smueller , my co-author, takes your invitation seriously and learns to navigate the streets of

12.15.2021 5:18am - (Replying to @stephensenn @JasonConnorPhD and 4 others) Yes, the 28% refers to CACE which is defined as the difference between the observed rates (Eq. 2).

12.15.2021 5:11am - (Replying to @stephensenn @AndrewPGrieve and 5 others) You can enforce compliance if the treatment is given in the hospital, under the supervision of a tough nurse. No one refused to enter the trial, if selected. Refusals occurred only among those choosing freely.

12.15.2021 5:04am - (Replying to @stephensenn @JasonConnorPhD and 4 others) Your interpretation of the observational study is correct. These are actual observed rates. Same as in the experimental study, except that here we interpret the rates as "effects". Drug refusers, fearing side-effects, simply refrain from taking anything.

12.15.2021 4:48am - (Replying to @JasonConnorPhD @stephensenn and 4 others) The observational study is described by the (fictitious) proportion of recovery in each subgroup.

12.15.2021 4:45am - (Replying to @stephensenn @AndrewPGrieve and 5 others) Good point.

12.15.2021 4:43am - (Replying to @stephensenn @AndrewPGrieve and 5 others) Clarifying: Para 1&2 100% compliance is enforced. Para 3, no prescription needed, the drug is available to all over the counter. This should be stated explicitly, agree.

12.15.2021 4:35am - Regarding whether "true qualitative interactions" exist, rare or not, here is a way of detecting their existence in the population, even when we do not have a name for the factor that causes them: This is another blessing of the new calculus.

12.15.2021 4:20am - (Replying to @stephensenn @AndrewPGrieve and 5 others) What kind of details do you want beside proportions? The more concrete version we have is a movie showing how the bounds vary with all possible combinations of observational and experimental data. I cannot think of any missing detail (except of course the sample size = 8 million)

12.15.2021 4:00am - (Replying to @stephensenn @AndrewPGrieve and 6 others) The data, of course is fictitious, but I dont see any reason for it being "rare", at least not "rarer" than the existence of some covariate-treatment interaction. Nor do I see any vulnerability to confounding, effect modifiers, or other modeling assumptions. It's assumption-free

12.15.2021 3:45am - (1/2) (Replying to @stephensenn @RWJE_BA and 6 others) I have difficulty understanding your objections. Do you doubt the possibility of significant differences (in response) between men and women? Do you doubt the possibility of any factor capable of producing such differences? Or do you doubt the possibility that new ideas
12.15.2021 3:53am - (Replying to @yudapearl @stephensenn and 6 others) can emerge, based on a new calculus, unknown to most statisticians, and highly relevant to health care and personalized decision making? At times it seems that the latter taints your objections. I hope this is not the case.

12.15.2021 3:36am - Hot off the press from @mitpress : THE HANDBOOK OF RATIONALITY: . #CogSci #Philosophy #Psychology #Rationality I have a chapter there:, but neighboring chapters are no less exciting.

12.15.2021 1:36am - I havn't changed my mind since I first saw this article, over a year ago. This may be either stubbornness or truth loving, pick your choice.

12.15.2021 1:23am - (Replying to @RWJE_BA @stephensenn and 6 others) Sorry, I can only respond to sound arguments. Moving on.

12.14.2021 8:25pm - (Replying to @Vashishtrv) Some readers said the Ladder of Causation is the Double Helix of causal inferences. I don't know if I would go that far, but it surely protects many researchers from falling into the countless potholes awaiting the uninitiated.

12.14.2021 4:20pm - (Replying to @stephensenn @RWJE_BA and 6 others) The data is given in terms of percentages, what else do you need? The number of patients in each group? Ans. 8 million in each. Their names and addresses? Please be specific.

12.14.2021 3:37pm - (Replying to @jos_b_mahoney) This is precisely what we are looking for, a Chomsky-like nativist picture for causal theories. What is innate and what is learned.

12.14.2021 3:34pm - (Replying to @stephensenn @RWJE_BA and 6 others) Blog posting is delayed due to my assistant surgery, sorry. But I still need to know what kind of justification you require. The mathematics is given, the assumptions are none. What else would you like?

12.14.2021 3:19pm - (Replying to @stephensenn @RWJE_BA and 6 others) Given all the noise Twitted around, I lost track of what the debate is about. What would you like to be justified, the conclusion? the method? the assumptions? What type of justification would satisfy you?

12.14.2021 3:09pm - We heard yesterday (Why21 workshop) from Alison Gopnik how 2-year olds outpace anyone's expectations in terms of number of trials needed for correct causal theory formation. I suggest investing in children-making, instead of ML startups.

12.14.2021 2:50pm - (Replying to @soboleffspaces @RWJE_BA and 6 others) Just in case anyone is curious about the history of statistics, there was no method of covariate adjustment before 1993. Hand-waving, yes, but no systematic method that is correct. Reason: confounding is not a statistical concept, see

12.14.2021 2:40pm - To all teachers, students and fans of Primer, good news from Wiley: "Corrected copies of the book will be available for purchase in 6 weeks' time." Watch for the statement "Reprint with Revisions 2021" on page iv. Enjoy!

12.14.2021 2:18pm - (Replying to @RWJE_BA @PavlosMsaouel and 6 others) Tell us more about what "all the people know". Assuming you are one of those people, please tell us what you know about causation, e.g., which homework in Primer you can solve correctly, or whether you can write down mathematically "It was the aspirin that cured my headache".

12.14.2021 2:02pm - (Replying to @RWJE_BA @PavlosMsaouel and 6 others) Read the paper, but havn't found any new argument over the established reasons for preferring RCT. Your key question to RCT priests should be: Are you familiar with the causal revolution? Not sure about these authors. I've asked Deaton and Cartwright same:

12.14.2021 9:18am - (Replying to @Noahpinion @MaxGurewitz and 2 others) Great! Was he able to express (mathematically): "The rooster crow does not cause the sun to rise"? 20% of economists could expressed it in the 1980's, and 0.1 % of statisticians.

12.14.2021 9:13am - (1/2) @Noahpinion . Thomas Kuhn would have had a ball observing first hand how statisticians won't let go of outdated thinking to embrace modern analysis. To the ultimate surprise of all who resist the idea of combining RCT's with observational studies -- rejoice! it involves
12.14.2021 9:13am - no counterfactual assumptions whatsoever, so it cannot be criticized for making unwarranted claims. It is based merely on universal counterfactual logic which is as solid as commonsense herself. Do I hear a cheer? Any converts to the revolution?

12.14.2021 9:09am - (Replying to @RWJE_BA @stephensenn and 6 others) Correct. Statisticians recognized the difference between statistical and causal thinking, got scared of the latter, declared it taboo and, now, when it is on everyone's plate, they claim:" statisticians knew it all along - at least 80 years."

12.14.2021 8:21am - (Replying to @stephensenn @f2harrell and 5 others) Claims about differences between women and men were made in the Garden of Eden, way before statisticians began noticing them. My claims are not about men and women but about combining RCT with observational studies. Any reference to such analysis by a statistician? Very curious!!

12.14.2021 8:10am - (Replying to @stephensenn @soboleffspaces and 5 others) Of course statisticians knew of the important difference between causal analysis and prediction. Yet, respectfully, they have done almost nothing about the former, see "Causal Thinking in the Twilight Zone" on Cochran

12.14.2021 7:40am - (Replying to @f2harrell @smueller and 5 others) More than similarity of sounds, what is needed is a bold abandonment of conventional thinking and a hard look at the new science of cause and effect. I wouldn't be spending time on Twitter if this result was a corollary of standard theories of experimental design. It ain't.

12.14.2021 7:23am - (Replying to @f2harrell @stephensenn and 5 others) The observational dataset plays no role in conventional RCT studies, or in IV studies. It can play critical role if properly analyzed and properly combined with RCT studies. But this requires unconventional methodology which I have hoped you would embrace with glee. It's worth it

12.14.2021 7:12am - (Replying to @DataFj) Hiccups of "empirics-first" hypes occur when "theory-first" practices stagnate or disappoint the public. In the case of economics, the stagnation resulted from broad dogmatic rejection of the causal revolution, w perhaps one exception: the late Hal White

12.14.2021 6:22am - (1/2) (Replying to @f2harrell @smueller and 5 others) Traditional decision making indeed finds no use for "wait a bit to see if the pt takes the drug". Our analysis shows that, sadly, lots of useful information is lost by ignoring the "wait and see" study. Critical differences between men and women are lost, together with
12.14.2021 6:28am - (2/2) (Replying to @yudapearl @f2harrell and 6 others) the belief that traditional decision making cannot improve (perhaps even revolutionized) by modern analysis.

12.14.2021 6:09am - (Replying to @MaxGurewitz @tdietterich and 2 others) The subfield of "Causal discovery" came into being after non-statisticians realized how terribly at odds causal modeling is with statistics, and after they discovered how causal assumptions, inarticulable in the language of statistics impose constraints on the data.

12.14.2021 5:51am - (Replying to @Noahpinion @MaxGurewitz and 2 others) If you never thought stat was at odds with causation, try to explain the absence of "cause" in the index of your favorite stat textbook. Or the inability of your favorite stat professor to write down mathematically: "The rooster crow does not cause the sun to rise". Try it!

12.14.2021 5:30am - (1/2) From my perspective of Kuhnian philosophy, the economic "empirics-first" hype is a temporary fad that will soon be begging to be rescued by its abandoned "theory-first" mother. See & The Kuhnian paradigm shift in Statistic
12.14.2021 5:30am - (2/3) has not been "modeling-->prediction" (both being prediction-aimed) but the transition "prediction---> causation", going contrary to Fisher's definition of stat (1925) as "the study of methods of the reduction of data." See Still, the most profound
12.14.2021 5:30am - (3/3) paradigm shift in modern times has been the traumatic discovery in machine learning of the limits of its own data-centric methods, as noted by @tdietterich . The Ladder of Causation highlights these limits in no ambiguous terms, #Bookofwhy.

12.13.2021 7:58pm - If you are a PhD applicant ready to join the Causal Revolution, don't miss this opportunity.

12.12.2021 5:56pm - (Replying to @NelsonDaleSmith and @FiveThirtyEight) Having scanned the paper, I must agree with @NelsonDaleSmith ; the causal vocabulary used could stand improvement in both clarity and consistency. Relations such as "causal association", "due to" etc could be explicated in graphs together with their supportive assumptions.

12.12.2021 3:48pm - (Replying to @ChristosArgyrop @stephensenn and 4 others) The question is whether NNT is deficient because it was never properly formulated, or because there is no demand for patient's judgment as to what risk he/she is willing to take. If there is a demand, the question is how to describe to a lay-patient the risks-benefits involved.

12.12.2021 1:01pm - (Replying to @holdmytowel and @soboleffspaces) This difference is precisely what we can quantify. I am talking of course on what CAN be done, not on HOW it would be implemented by machine learning folks, only a tiny fraction of whom understand what CAN be done -- see their curriculum.

12.12.2021 11:53am - (Replying to @f2harrell @PavlosMsaouel and 4 others) If you buy my religion that (1) people are counterfactual machines and (2) If you want information from them, cast the question in counterfactual language, then it follows that we could do something in this direction today, that we have the language, and can tie it to data.

12.12.2021 11:39am - (Replying to @soboleffspaces and @smueller) I know about Holland pessimism, but did not realize Mill dealt with individualized cause and effect. Do you have the page number in his "System of Logic"? I have the first edition (1843) on my shelf, arm's reach.

12.12.2021 7:58am - (Replying to @f2harrell @guyharvey70 and 4 others) If you are referring to then, no, the RCT data were properly analyzed. The discrepancy with "real world data" is due to allowing men and women to decide, on their own, whether to take the drug. ie, it was not randomized.

12.12.2021 7:36am - (Replying to @guyharvey70 @soboleffspaces and 4 others) Let's pursue your request! I see that you are asking for a measure that an individual patient can understand, to help make a personal decision. Statistics cannot provide it from population data. Please take a look at ; isn't it what you need?

12.12.2021 6:08am - (Replying to @stephensenn @f2harrell and 3 others) I am back to my bottom line: Do you think that: (1) physicians were justified in trying to find a criterion based on counterfactual intuitions? (2) We can fix today what they failed to do in 1980? (3) The quantity 1/PNS plays a role in the fixing?

12.12.2021 5:29am - (Replying to @f2harrell @stephensenn and 3 others) How can we say NNT is deficient if we haven't defined it? I believe the EBM folks tried to capture individual decision making but lacked a language to do so; all they had was the language of RCT. Bottom line: Do you believe we can rectify their confusion with today's tools?

12.12.2021 5:08am - (Replying to @stephensenn) I dont see equations with P(y_x) in them, but you may have used different notation for Rung-3 quantities. If you did, please flush them out. Bottom line: Do you agree that 1/PNS was what the NNT folks tried to formulate all along?

12.12.2021 4:54am - (Replying to @WernervanWyk2) Where can we find the entire text?

12.12.2021 4:49am - (Replying to @soboleffspaces @stephensenn and 3 others) I believe it is 1/PNS which generations of NNT enthusiasts have been trying to capture mathematically, but did not have the language to express. Someone should write a note to the EBM folks and tell them that things have changed and, today, we can do it right. Will they listen?

12.12.2021 4:35am - These critiques of NNT were all conducted in either conceptual language (English) or statistical language (Rungs 1 and 2). Today, with the advent of Rung-3 language, coupled with the bounds on probabilities of causation, we should be able to do it right.

12.11.2021 11:51pm - (Replying to @AndrewPGrieve @PavlosMsaouel and 3 others) Great paper! Surveying the literature, why NNT ain't easier to comprehend and why it ain't correct as defined. The latter is easy to see: NNT is laded with counterfactual notions and defined in term of RCT findings: Bingo! Impossible! A new survey is due, using Rung-3 language.

12.11.2021 10:49pm - Precious words, about truth and journalism, which I always wish to transport to science as well. No, we do not have authoritarian governments that silence us, but we have other forces that fragment us into non-communicating bubbles, keeping truth at bay.

12.11.2021 10:09pm - (Replying to @zioness_lioness and @ChancellorCUNY) @ChancellorCUNY is as bewildered as other Chancellors across America - they have never dealt with hate disguised in noble vocabulary. What he is waiting for is a draft statement that (1) he can legally make and (2) will end the BDS circus once for all. See

12.11.2021 9:36pm - Watching this map at this "embassy" and these two Presidents, I begin to understand why our sages insisted on: "A good neighbor is a true blessing." Some countries have it, some do not.

12.11.2021 9:02pm - (Replying to @soboleffspaces and @artistexyz) A more remarkable fact: PN and PS are also identifiable under monotonicity if we also have data from non-experimental studies. (Causality Theorem 9.2.15). It is so remarkable that (the late) Stephen Fienberg shook his head in disapproval throughout my lecture at CMU.

12.11.2021 8:41pm - Amazing program!!! Thanks @eliasbareinboim for putting all this together across paradigms!!!

12.11.2021 8:32pm - (Replying to @artistexyz and @soboleffspaces) It implies the converse, b/c the equality "defines" the linguistic term "exogeneity". It is interesting to trace the econometric literature of the 20th Century and watch (w some amusement) how they agonized over the latter w/o the do-operator. Some still do (Imbens etal, 2015).

12.11.2021 8:19pm - (Replying to @soboleffspaces) Monotonicity is not experimental condition; it is Rung-3, saying Y(1,u)>= Y(0,u) for all u. It cannot be ascertained from RCT data alone, because it asserts inequality "for every individual u". This is one example where the Rung 2 vs. Rung 3 distinction is important.

12.11.2021 8:04pm - (Replying to @artistexyz and @soboleffspaces) Identifiability is the ability to convert P(y_x) into any probabilistic expression, not necessarily P(y|x). For example, P(y_x)=E_z P(y|x,z), or P(y_x) = front door formula, or P(y_x) = Napkin formula, or whatever do-calculus authorizes. Exogeneity is just P(y_x)=P(y|x)

12.11.2021 3:15pm - Opinion: Don't be fooled. Zionism is an Indigenous rights movement and being anti-Zionist is antisemitic.
12.11.2021 3:15pm - Zionophobes want Zionism to be "debatable", while Zionists want Zionophobia (like Islamophobia) to be despicable. This opinion piece by Micha Danzig adds new dimensions to the debate, and anchors it in the court of San Diego Unified School District Board of Education.

12.11.2021 1:58pm - This silent video speaks to us louder than the loudest television presentation I've seen.

12.11.2021 12:56pm - Not sure I can agree with the logic of this argument. Whatever knowledge one brings to prove the failing of an "easy way of screening" can in principle be incorporated in a "better way of screening", thus making screening both easy and better.

12.11.2021 7:21am - I am the last to base "Jewishness" or "indigeneity" on genetic lineage (see, but since our neighbors do, this finding might be of value.

12.11.2021 6:34am - (Replying to @hangingnoodles) Taking liberty to complete the quote: "Whatever economists are up to, it isn't like physics modeling, and it's certainly unlike statistical modeling."

12.11.2021 6:30am - (1/2) Here is another review of the causal revolution, by a statistician/economist who refuses to accept the Ladder of Causation: and tries hard to embed causal inference in parametrized statistical models. Readers who understand this embedding may explain
12.11.2021 6:30am - (2/2) to us how interventional (and counterfactual) questions are answered in this framework, if at all. The quotes from #Bookofwhy will, I hope, entice readers of J. of Economic Surveys to examine the causal revolution from a logical perspective, liberated from statistical dogmas.

12.11.2021 4:50am - (Replying to @RWJE_BA) If you write a well-reasoned critique of @Bookofwhy, I promise to add it to the other 38 reviews of the book, all listed here:

12.11.2021 4:25am - Billoo and @Cair are totally misguided about Islamophobia. Zionists expose and oppose Zionophobes, like @Billoo , Ilan Omar and Rashida Tlaib, and this has nothing to do with Islamophobia. It has to do with their eliminationist animosity to the idea of a Jewish homeland.

12.11.2021 3:57am - The relationship between the individual causal effect ICE(u) and the probability of necessity and sufficiency PNS (see for definitions) should read: PNS = P(ICE(u)=1) Thanks for pointing out this needed clarification.

12.11.2021 2:53am - (Replying to @KunalTi00053352 and @ccaprani) Sure, but this requires many experiments, and even these only get you to Rung-2 of the Ladder. Causal discovery from non-experimental data is much weaker, yielding larger sets of observationally-equivalent graphs.

12.10.2021 2:21pm - Overwhelming Number of Diversity Officers at US Colleges Hold Anti-Israel Views, Study Says
12.10.2021 2:21pm - Has anyone calculated what percentage of students' tuition goes to finance the high salaries, enormous staff and lavish offices of the Diversity hoax in our universities? I call it "hoax" because I've been trying to get them to listen since 2015.

12.10.2021 1:59pm - (Replying to @tdietterich) But Moses kept talking to God, which is the greatest confounder known to mankind. This was obviously an awfully biased RCT.

12.10.2021 1:51pm - Among the 23 authors commenting on "American Jews and Israel": I am the only one suggesting that Jews should split apart: Strange, I have not received any hate mail yet . Evidently, people are ready for it -- Shabbat Shalom,

12.10.2021 1:12pm - "If our conception of causal effects had anything to do with randomized experiments, the latter would have been invented 500 years before Fisher." --- #Bookofwhy (page 135) Addendum: And God would not have spoken to us before 1925.

12.10.2021 11:15am - (Replying to @RWJE_BA) From a purely statistical viewpoint, there is no paradox in Simpson's reversal, so there is nothing to "interpret". Yet even a pure statistician like Simpson noticed that the "sensible interpretation" does not make sense, see Statisticians R humans.

12.10.2021 10:00am -
12.10.2021 10:00am - This press-release is a perfect example where "Zionophobia" should replace "antisemitism". The latter appears 14 times charging CAIR with offenses they would deflect with "We love Jews". Replaced with Zionophobic and they can't deny any; they can't say "We love independent Jews."

12.10.2021 4:15am - Let me explain. I used to think that @jvplive objects to a Jewish homeland while @IfNotNow agrees to a Jewish homeland as long as it does not defend itself. The recent @IfNotNow posters however make doubt my distinction - they are as bad as JVP's or CAIR's, perhaps worse.

12.10.2021 2:17am - I don't get it. What exactly is the difference between @IfNotNowOrg , @jvplive , @CAIRNational , and Muslim Brotherhood, not counting the "J" in the former two?

12.10.2021 2:01am - Agree. Simpson's paradox is the best proof that data alone cannot distinguish right from wrong. Still, getting ML folks to internalize it is a different story, perhaps more challenging than getting economists to use the First Law.

12.9.2021 10:31pmm - Luckily, the causal revolution has replaced the question of "convincing" with the question of "correctness": What do you need to do to make a matching estimator correct? Now, since matching is the same as "conditioning", the question is solved. #Bookofwhy p.273-5 gives examples.

12.9.2021 9:42pm - (Replying to @artistexyz) Sorry for being away. But Scott explains it without appealing to monotonicity.

12.9.2021 2:15pm - For those who are not alarmed by UCLA's latest turn in the politicization of higher education, let me retweet this captivating article "The Peril of Politicizing Science" in the Journal of Physical Chemistry It has already earned 66,000 downloads.

12.9.2021 10:51am - The chants at Charlottesville are disgusting, but they did not surprise anyone who knew Charlottesville. The tyranny of "woke" is a totally new fire on my campus, consuming every piece of commonsense and, what's worse, is echoed in the Halls of the US Congress. Scary.

12.9.2021 7:13am - (Replying to @DJMinded @ADL and 2 others) Having problem with policies is never anti-Semitic, and you would rarely find me using this word now that we have a more specific term: Zionophobia. I would also check my information, in case it is generated by Zionophobic sources, say Al-Jazeera, CAIR spokesmen, and some media.

12.8.2021 9:53pm - Smart, passionate and searching people are honestly unaware of the fact that their sense of individual identity has been manipulated by a woke collectivism. They need to be awaken by smart, searching and respectable people who can tell them: You have been duped by Zionophobes.

12.8.2021 9:22pm - (1/2) The Forward's piece about Zionophobia has been cleanrf of typos; it now makes more sense: One remark about the "Jerusalem Declaration of Antisemitism" which Alon Confino helped organize. What this thoughtless "Declaration" has accomplished one thing:
12.8.2021 9:22pm - (2/2) It has armed anti-Semites with an unassailable argument: "Even Jews do not agree on what anti-Semitism is, so how can you call anyone anti-Semite?". The "scholar" who authored this declaration is hoping to emasculate Zionophobia as he has done to anti-Semitism. No chance!

12.8.2021 8:30pm - (Replying to @alex__gut) (1) It's not illegal for an employer to just "recommend" who to vote for or what document to sign. (2) No, I recommend no committees, just individual commonsense. (3) Yes, apolitical. Especially, when they know nothing about the subject matter, rashly following populist slogans.

12.8.2021 7:34pm - (Replying to @alex__gut) Good question, which touches on Russell's self-referential paradoxes -- we certainly do not to include anti-inclusionists . However, to be practical, I do not want my direct boss to tell me how to vote, or whose ideology is most abhorrent. Allow me to use my reasoning software.

12.8.2021 6:18pm - (1/3) Disappointing news for those of us who believed in the de-politicization of higher education. Remember the Open Letter a UCLA Faculty Committee sent to UC President Drake? . A new ruling was issued last week, articulating
12.8.2021 6:18pm - (2/3) "an overriding principle that departments should not be precluded from issuing or endorsing political statements." In other words, each academic department may now decide what political party to endorse, which country to delegitimize and which people to demonize.
12.8.2021 6:18pm - (3/3) It's a new reality, where students who used to look up to their departments as pillars of inclusion and support will be told "you are not exactly welcome here" or "you may actually not fit anywhere on this campus". And junior faculty will need to decide: bow or perish. Sad!

12.8.2021 1:30pm (1/ ) Here is a paper that crossed my screen, dealing with an important problem of combining experimental and observational data, yet I can't understand why it assumes that the latter is always overestimating. Moreover, the problem addressed is selection bias
12.8.2021 1:03pm (2/2) in the RCT, for which we have non-parametric solutions e.g., Do the two methods converge? Hard for me to tell.

12.8.2021 12:09pm - Palestinians outraged by sperm-smuggling drama film
12.8.2021 12:09pm - An interesting drama of Palestinian sperm smuggling in a Jordanian movie. Evidently, even the Jordanians are incapable of understanding the Palestinian narrative and how protective it is of the divine status of their super-heroes, the prisoners.

12.7.2021 8:06pm - (Replying to @david_sontag) My students find the idea of "variables listening to each other" more natural than "potential outcome" - you dont even need "intervention" to grasp it. Then, seeing that everything is derivable from "listening", including potential outcomes, causes of effect, etc. is real thrill.

12.7.2021 7:52pm - (Replying to @killerkristi) Glad you are starting with "what is you actual goal?" but I assume you are a new follower of this channel because I have been trumpeting my goal since time immemorial: " Two States for Two People, Equally Legitimate and Equally Indigenous." Plus, start with the latter.

12.7.2021 7:15pm - (Replying to @killerkristi) No government is supported when you call someone a "morally impaired Zionophobe", you merely assert the right of Jews to a homeland and the moral deformity of those who deny them that right. "Governments" are in the minds of careless scholars who crafted the JDA Definition

12.7.2021 6:24pm - (1/3) Zionophobia will now be debated and used by readers of The Forward, a positive move to becoming the ugliest word in town. Alon Confino, the last scholar to be quoted in this article, is totally mistaken about pivotal role that pre-political Zionism plays in Jewish identity.
12.7.2021 6:24pm - (2/3) My grandfather prayed three times a day "He'll return us in sovereignty to our country" (Birkat Hamazon). The dream of "Return to Zion" has been the engine behind Jewish hopes and resilience throughout the generations. To label it "political ideology" is to confess ignorance
12.7.2021 6:24pm - (3/3) of Jewish history and to deny the collective memories that make Jews a people today. Balfour understood it better than our "new historians" and Hanukah proves him right An immutable component of identity

12.7.2021 11:31am - (Replying to @DJMinded @ADL and 2 others) Everyone is against "imposing apartheid" - that's easy. But occasionally I tend to forgive what appears to be "apartheid", for example, how they separate American from non-American passengers in airport terminals, or vaccinated from non-vaccinated customers in theatres. Awful!

12.7.2021 9:07am - I don't find it stunning at all. If Jews do not defend their identity from Zionophobic bashing no one would do it for them. Jews will gain respect only when Zionophobia becomes the ugliest word in town; it will, when @ADL , @AJC , @HillelIntl etc. start spelling it.

12.7.2021 6:20am - Just listened to a lecture by @david_sontag on Causal Inference and thought to myself: How much easier it would have been for MIT students had Sontag used the First Law to explain what {Y(0), Y(1)} are and what "ignorability" is. Thoughts for 2022 !

12.7.2021 4:48am - The media was not the only one in a haste to blame Israel. Two academic departments at UCLA have been fooled by the haste and decided to incriminate their colleagues and students for imaginary crimes that never took place. For details, see:

12.7.2021 4:33am - Zionists are behind Islamophobia, US police brutality - CAIR official
12.7.2021 4:33am - Amazing! No one has indicted CAIR for being behind Zionophobia, not even @ADL , so I'll do it now: @CAIRNational , you are a major contributor to a moral deformity called Zionophobia. It started in 2001 when your officials blamed Israel for 9/11, and continues today w/ your Billoo.

12.7.2021 4:10am - (Replying to @EinatWilf and @Georgetown) Great and comprehensive reading list on the history of Zionism. I would only add Avraham Mapu's "Ahavat Zion" (1853) which preceded Gordon's "Wake my people" by 10 years, and spread throughout the diaspora (including Sephardic communities) like brush fire.

12.7.2021 3:57am - I stand with Alexandra Lulka for refusing to state that every evil has some virtue. Some evils are plain WRONG, and school board officials are paid to expose them. I stand with Alexandra Lulka!

12.7.2021 3:16am - (1/2) Ernst Mach (1838-1916) is often remembered through his quote: "There is no cause nor effect in nature ; nature has but an individual existence ; nature simply is." (Science of Mechanics 1893:483). Recent interpretations of Mach should therefore be of
12.7.2021 3:16am - (1/2) interest to students of causal inference. As a student of AGI, however, I do not buy into Mach's pessimism. Even if Nature was cruelly opaque to us, AI should nevertheless try to interpret it causally -- its mission is to emulate us, not Nature.

12.6.2021 6:57pm - It seems that the USC cover-up scandal has hit the media: I can't understand however why they all focus on poor Jasmeen, instead of USC's mishandling of the 60 faculty Letter. The poor saint merely wanted to "normalize the language of the resistance."

12.6.2021 1:40am - (Replying to @richard_landes) USC's cover-up has been so successful, that students like Jasmeen did not know that most professors at @USCViterbi are Zionists ( hence, "colonial oppressors" etc., worthy of f**g death) Her infantile attack on Zionists turned out to be an attack on @USC 's academic excellence.

12.6.2021 5:32pm - CUNY Law Student Gov't Passes Pro-BDS Resolution Targeting Groups Like Hillel via @jewishjournal
12.6.2021 5:32pm - I have a hunch that these infantile tantrum attacks on Zionist students will continue unabated, until lovers of Israel, @HillelIntl , @ADL , @SWU etc. will learn to spell the word Zionophobia. @bandlersbanter

12.6.2021 4:57pm - USC Defends Response as 60 Faculty Members Call for `Rebuke' Over Student's Violently Antisemitic Tweets
12.6.2021 4:57pm - Hiding under a private response is one way USC tries to keep campus Zionophobia from public eye. Taking cover under generic anti-hate programs that deliberately avoid Zionophobia is another. For specifics, see @Algemeiner @jdforward @bandlersbanter @USC

12.5.2021 8:58am - (1/2) The USC scandal offers a perfect example of how yesterday's fighters against antisemitism offer a cover for universities inaction on contemporary antisemitism. In her reply to the 60 professors' Letter, USC President Folt mentions the Shoa Foundation
12.5.2021 8:58am - (2/2) and Dean Varun Soni as two groups "examining antisemitism on our campus". As you can expect, these groups have yet to spell "Zionism" in their writings, and none has invited ANY of the 60 professors to join the "examination" process. A perfect cover up for continued inaction.

12.5.2021 6:49am - Unfortunately, this unity of purpose all but disappeared when it comes to fighting contemporary antisemitism, i.e., Zionophobia. Some of our yesterday's fighters refrain from even spelling the word.

12.5.2021 6:34am - "Age" is the first question they ask you when you check into a hospital emergency room. Luckily, they haven't heard about the manipulability debate within the PO framework, as the one described here

12.4.2021 12:12pm - (Replying to @garybkatz @RozRothstein and 4 others) Not so! Universities are already catering to Qatar to replace Jewish philanthropy. What they still cannot do is replace Jewish professors with Qatari professors; that is why they are hoping that the former would be fooled by empty words of "fighting antisemitism".

12.4.2021 9:14am - (Replying to @DrewLevy @ESMDcan123 and @PavlosMsaouel) Wondering: How many regression analysts would agree with this statement: "A well-formulated regression model ultimately tells a causal story " which, based on my readings, would render most regression models "NOT well-formulated", since they have been denied any causal reading.

12.4.2021 9:03am - (Replying to @konohas_itachi1) If you have a fully specified SCM then, regardless of how you obtained it, it is sufficient, using the First Law, for answering all interventional and counterfactual questions that one can possibly dream. (including, I hope, ITE -- what is it?)

12.4.2021 8:04am - (1/4) Is it dumbness or deliberate blindness that prevents USC officials from listening to their students and faculty? Death threats were disseminated against Zionists. Incriminating statements were made against the very being of Israel. 60 distinguished professors are pleading
12.4.2021 8:04am - (2/4) with USC leadership to explicitly de-criminalize Zionist and Israeli identities [quoting from their Letter]: "Most importantly, Jewish, Zionist, and Israeli students, as well as those who support the right of the State of Israel to exist need to hear from our leaders that
12.4.2021 8:04am - (3/4) they are welcome on our campus." Yet, stunningly, in their response, USC leaders blatantly and meticulously refrain from spelling out the words "Zionist" and "Israel", leaving thousands of students, faculty, staff, potential students, parents of USC students, and the
12.4.2021 8:04am - (4/4) community at large wondering: Does "Israel" have a spelling? Are Zionist welcome on USC campus? Are Israel-supporting professors who have devoted their professional lives to an institution welcome on a campus they have helped build?

12.3.2021 4:13am - As I was looking for another song to lift our spirit on this 6th night of Hanukah, I bumped into this one and decided to share it. 12,000 voices from a country that swings stormily between "The Sweet and the Bitter," yet leaves zero room for pessimism.

12.3.2021 2:19am - (Replying to @RomanFolw) The subject of research in CI is almost the same: "how does the human mind process causality", except that we take the word "process" seriously, and insist on modeling it algorithmically, going from input to output, rather than hand-wavingly.

12.3.2021 2:09am - (Replying to @PophamFrank and @deevybee) Beautifully said! And, sadly, these circles are growing.

12.3.2021 2:05am - (Replying to @FionaMak70) James ward, from what I can tell, correctly identified Simpson's paradox as a proof that established methodologies cannot rely solely on correlations. I do not see why belief and confirmation biases should get in the way if one ventures to model them properly in SCM.

12.3.2021 1:32am - I taught "control theory" in 1970-1972 and little could I imagine that, one day, control theorists would find Causal Graphs to be beneficial, as this paper shows: The key feature they find useful is the "Single Door Criterion" (see

12.3.2021 12:52am - This paper has crossed my screen:, and, naturally, the title "Why do Things Happen?" made me think of #Bookofwhy and ask: Can "Attribution Theory" benefit from modern methods of causal inference? More audaciously, how do they manage without CI? No idea.

12.2.2021 11:41pm - (Replying to @xenophilictb and @drgabe1) I am willing to accept this, more authentic definition of the moral deformity called "Zionophobia": A Zionophobe is a person who equates a homeland for the Jewish people with Nazism. Brilliant!

12.2.2021 5:52pm - That quote was written a year ago. We down now to 2-4 years

12.2.2021 5:00pm - How many milliseconds does it take for an average reader to figure out what "Islamophobia" stands for? Why should it take more to figure out what "Zionophobia" is? The words says it all; what is being hated and why it is a moral deformity.

12.2.2021 1:27pm - (Replying to @schraubd and @ArnoRosenfeld) A painful indictment of left-leaning folks. I would complain.

12.2.2021 1:03pm - (Replying to @drgabe1) "Criminal" is a word reserved for Israel. Our leadership is merely "breaching public trust"; they don't get it that it is their duty to de-criminalize their elite faculty, as well as their students.

12.2.2021 12:54pm - Glad you see the power of "Zionophobia". Please use it in your writings, because 90% of your readers loose debates by applying "antisemitism" to "ugly anti-Zionism", thus allowing their adversaries to shift the debate into "even Jews do not agree on what antisemitism is".

12.2.2021 9:56am - UCLA situation may be worse. While the Zionophobes have not reached USC's level of "kill f**g Zionists" mentality, they nevertheless managed to highjack a whole department (Asian-American-Studies) and post their poison on University official website, see

12.2.2021 7:32am - (Replying to @konohas_itachi1) The First Law is applied to a fully specified SCM models, which cannot be obtained from data; it needs functional assumptions. No neural network can obtained this information from data alone, unless it is fed additional, Rung-3 assumptions.

12.2.2021 6:17am - Any paper that uses the First Law of Causal Inference should be taken seriously by deep learning folks. However, going from population data to individual behavior requires bounding methods (Tian etal (2001), I don't see how the paper can do w/o them.

12.1.2021 5:27pm - What happens when 60 distinguished professors in some university say: Enough is Enough"?. It normally yields some response, right? Not at USC: 60 distinguished USC professors are asking a simple question: Are we welcome on this campus?

12.1.2021 10:00pm - (Replying to @terryfahn and @kennethehrlich) Here is their hash tag @USCAlumni , can you let them know what you think they should do? And please ask them to spread the word, because USC would rather hush-hush the case than tell their professors "You are welcome".

12.1.2021 9:43pm - Yes, I can recommend the literature linked to here: and here: It's fun too.

12.1.2021 4:00pm - Beautiful. Thanks for reposting. I did not feel comfortable after the interview, because I saw so many spots where I could have done better. But watching a year later, it ain't bad at all. Even highly recommended!

12.1.2021 3:21pm - (Replying to @intrinsic_motiv and @causaLens) Much sooner than later.

12.1.2021 3:07pm - (Replying to @stephensenn @mendel_random and 2 others) Agree that it is very dangerous but, in our case it is not merely "never been written" but also "impossible to have been written" because it requires causal assumptions which statisticians had no language to express until 1974 and, even then, could not bring themselves to use.

12.1.2021 10:29am - Glad you bring up Lord Paradox again. It is still my favorite. It should replace Simpson's Paradox in all stat. textbooks.

12.1.2021 10:25am - (Replying to @attilacsordas) I do not know how causal inference is related to Edie Falco, or to any of her statements. Can't read the whole interview.

11.30.2021 10:15pm - This happened to me once, upon arriving at Ben-Gurion Airport, Tel Aviv, though not on such a scale. The looks on the faces of these surprised passengers perfectly represent the spirit of Hanukah (3rd Candle tonight)

11.30.2021 11:59pm - (Replying to @offaltube) It's not impossible at all. Don't forget the Charedim are no longer in the coalition, and even when they were, they did not interfere with music and dancing at the airport.

11.29.2021 11:54pm - (Replying to @TheHofstatter and @SSI_Movement) There is not competition at all. I am receiving "Giving" requests from many organizations, some to fight antisemitism and some to fight Islamophobia. This year, for the first time, the number of the latter surpassed the former, and "protecting Jewish students" is the main theme.

11.30.2021 11:37pm - (Replying to @PavlosMsaouel @ESMDcan123 and @f2harrell) Oh, yes, it is listed indeed among other reviews of #Bookofwhy, see

11.30.2021 9:16pm - (Replying to @PavlosMsaouel @ESMDcan123 and @f2harrell) Do you have the notes to Drew Levy lecture? I would like to learn how to talk to regression analysts, in their vocabulary, about causal inference.

11.30.2021 8:07pm - Here is the "Giving Tuesday" appeal that I've recommended. Students who are proud to spell their noble cause, and are not shy to seek faculty advice on how to approach ill-informed university administrators or racist professors.

11.30.2021 7:54pm - (Replying to @ESMDcan123 and @PavlosMsaouel) An interesting paper followed by an interesting discussion. The title, however, throws me into a pause: Regression Modeling Strategies: Causal Inference... In my mind "Regression" connotes disregard for causal considerations. What did you mean to convey by "regression"?

11.30.2021 2:50pm - (1/3) The number of "Giving Tuesday" appeals I have been receiving from organizations pledging to "protect Jewish students on campus" has surpassed the number of appeals to combat Islamophobia -- that's an awful lot. This year, my preferred choice is @SSI_Movement . Here is why:
11.30.2021 4:50pm - (2/3) (1) SSI (Students Supporting Israel) spell "Israel" explicitly on their banner, and I wouldn't trust leaders who can't spell. (2) SSI has a Faculty Advisory Board of which I am a proud member. Most other organizations, including big ones, seek no faculty input,
11.30.2021 4:53pm - (3/3) which renders their efforts to "protect Jewish students" largely ineffective, especially in view of growing Zionophobia among university administrators. Thus, if you value my "Giving Tuesday" recommendation, my choice would be. @SSI_Movement

11.30.2021 10:02am - (1/2) Had fun last night on a panel celebrating Jewish Thanksgiving Day, and read from Einstein's Letter of June 13 1947 where, pleading for "justice and equity," he tried to convince Nehru to vote for the UN Partition plan My point was that Einstein,
11.30.2021 10:02am - (2/2) unlike many of my colleagues, did not mind criticism of the type: "Stick to Science and don't meddle with politics." He understood that there are moments in history when you must speak for "justice and equity", even though some label it "politics". Something to keep in mind.

11.29.2021 4:08pm - (Replying to @DKedmey and @BayoumiMoustafa) I agree. There should be a litmus test separating scholars from those who use their academic positions to propagate falsehoods, racism, if not genocidal agendas. The slogan "equally legitimate, equally indigenous" is such a litmus test. It has served me faithfully in the past.

11.29.2021 1:32pm - Another glance worth taking, what Mahmoud Abbas means by "We made a mistake" which he can correct today in one word: "Equally indigenous". Words he can't utter.

11.29.2021 1:27pm - Going to celebrate soon the Miracle of Nov. 29: For a glimpse at the content: But it's illuminating to take a good look at the map below, the map that Palestinians rejected, with a State double the area of West-Bank + Gazza.

11.29.2021 11:09am - (Replying to @CausalHuber @artistexyz and 2 others) Surely in some examples D||(Y(1),Y(0))|X holds and in others it doesn't. The question is whether the language of conditional ignorability is adequate for a researcher to tell, for arbitrary X, when it holds and when it doesn't.

11.29.2021 11:01am - (Replying to @ajaydiv) I'm hopeful, but not confident. Research papers from each bubble need an outlet in the very same bubble, reviewed and edited by light-avoiding gurus in that bubble. Galileo had easier time publishing "Discorsi" in Lyden than publishing a graph-based paper in Econometrica. (#=0).

11.29.2021 9:43am - Max Planck says: " a new generation grows up that is familiar with it." Unfortunately, the isolated bubbles currently reigning science education make it very hard for a new generation to be "familiar with it".

11.29.2021 9:37am - Woe to science education that commonsense needs stickers to reach students.

11.29.2021 9:07am - (Replying to @CausalHuber @artistexyz and 2 others) Stratification on pre-treatment X appears credible, until you get to M-bias and fail, as Rubin did: No revolt in PO land?

11.29.2021 9:01am - Keen observation: #badscience = #idontgetit. Another Nobel Laurate in Econ. just labeled do-calculus "based on ad-hoc assumptions". No revolts in Eco-Land? No pride?

11.29.2021 8:45am - Here you have it, all 3 bullets in one sentence, repeatedly, year after year. And what do PO gurus say? [In our applications (= everything on earth)] "we have not found this [graphical] approach to aid drawing of causal inferences" (Imbens & Rubin, p.22). No revolt in PO-land?

11.29.2021 8:22am - (1/2) One big difference:There is no human judgement in quantum mechanics, and lots of judgment in Causal Inference. Given that every modeling assumption must pass a plausibility judgment by a human scrutinizer, it ought to match the format in which humans think. SCM offers 1/2
11.29.2021 8:22am - (2/2) such a format, for it's based on relationships among measurable variables. Now, before you say that PO's are also meaningful, try judging whether {Y(0), Y(1)} || D | X holds in any problem of your choice. Take eg the napkin problem, or any from .

11.29.2021 2:07am - I have addressed the golden bullets persistently since I encountered PO. See post w/ links: The reason the PO literature refrains from addressing them is, I must theorize, an embarrassment of ineptness. You think I'm kidding? See

11.29.2021 1:34am - (Replying to @herdiants) For PN you imagine that X=1 and Y=1 are observed, and you ask what would Y be had X been 0 instead of 1.

11.29.2021 1:28am - (Replying to @CausalHuber @artistexyz and 2 others) This is an argument I have had with Imbens since day one: Is transparency of assumptions a matter of "getting used to" or an objective feature of the language in which the assumptions are expressed? Do you know anyone who has "gotten used to" judging {Y(0), Y(1)} || D| X ?

11.29.2021 1:13am - (Replying to @yudapearl @CausalHuber and 3 others) The reason I am asking is that, Imbens and Rubin definition of "unconfoundedness" (p.479) is so cryptic that any reader trying use this definition to judge whether a given assumption is plausible would rather give up than judge.

11.29.2021 12:58am - (Replying to @CausalHuber @artistexyz and 2 others) Correct me if I am wrong, but I think only economists use the term "identifying assumptions," tacitly implying that certain assumptions are made in order to identify things, not because they are actually believed. Do you think PO assumptions can ever be judged for believability?

11.29.2021 12:38am - (1/2) This is a great piece that I wrote for a Pakistani newspaper, thanks for reminding me. Note how drastically Palestinian strategy has changed between 2005 and 2021. In 2005 they were trying desperately to convince the world that they stand for a 2-state solution. Today, 1/2
11.29.2021 12:38am - (2/2) they no longer see any need to hide the truth, and they go "From the River to the Sea" in broad daylight, in English -- eliminationism normalized.

11.28.2021 11:38pm - (Replying to @Daniel_J_Im) I was young then.

11.28.2021 11:27pm - (Replying to @KyleCSN) If you have a counter example, please share. @zcccucla

11.28.2021 11:23pm - (Replying to @synapticity) It has undergone a revival in Israeli songs, especially in 1970-80's.

11.28.2021 11:19pm - (Replying to @bariweiss) Another take:

11.28.2021 6:51pm - (1/3) Just finished lighting the Hanukah Candle -First night. My grandson asked: "What does Hanukah mean to you, personally? Don't tell me you believe in that miracle with the oil vessel." I told him that for us, the generation that grew up in Israel, Hanukah is captured by a poem:
11.28.2021 6:51pm - (2/3) "A miracle did not happen to us; We have not found a vessel of oil; We carved the rock till we bled; And there was light! There was light!" It was sung by Zionist pioneers in the 1930's who literally built a nation by carving the rock till they bled. And they prevailed.
11.28.2021 6:51pm - (3/3) In 2007, I wrote an oped about them, seeing the torture that American Jews go through when interpreting the miracle of Hanukah. Here it is:

11.28.2021 5:01pm - (Replying to @artistexyz @CausalHuber and 2 others) Again, we need not seek graphs, we seek universal standards of commonsense. Would readers ever know what "confounding" means? Would they know when assumptions have testable implications? Most PO researchers don't.

11.28.2021 4:22pm - (Replying to @jazchaz) I smelled the Maccabee's accent, baked for 2,000 years.

11.28.2021 3:04pm - (1/2) In describing the Golden Bullets we should not invoke DAGs or paths, lest we get accused of circularity. Instead, Transparancy -- compatibility of assumptions with existing knowledge Identifiability -- estimability of research question from
11.28.2021 3:04pm - (2/2) assumptions and available data. Testability -- Compatibility of assumptions with available data. Phrased in this general vocabulary makes PO champions embarrassed of past negligence, yet not enough to break the silence, they simply do not have the tools.

11.28.2021 1:22pm - (Replying to @hmatejx and @yihanxu) Fully agree. You need to take a class in statistics to appreciate how revolutionary commonsense can be. This lecture of mine was given 25 years ago, and commonsense is still begging to enter the standard curriculum of statistical education.

11.28.2021 1:11pm - (1/2) Readers urge me to explicate what I mean by the three Golden Bullets: Transparency, Identifiability and Testability. I find them listed here: and here: and perhaps in dozens other places. Unfortunately, PO students have been 1/2
11.28.2021 1:11pm - (2/2) kept unaware of these bullets, tightly isolated in a bubble of silence (I suspect deliberately) for the past 20 years. I was hoping that books published in our decade would break through this bubble. Some do. More will.

11.28.2021 10:17am - (Replying to @CausalHuber) The issue is not DAGs vs. PO. The issue are the three golden bullets: Transparency, Identifiability and Testability, which are blatantly neglected in PO-based books (eg., Imbens and Rubins, 2015) and which I have hoped would be illuminated in yours. Have I missed any?

11.28.2021 8:27am - (Replying to @erikbiz) #BDS goal has never been to protect Palestinians, not even to boycott Israel, but to intimidate pro-Israel students into silence while staging kangaroo trials against Israel, so that gullible students may conclude: Israel must have done something wrong again. Ingenious.

11.28.2021 8:15am - (Replying to @synapticity) The dialogue is known as Ladino, the language spoken by Middle Eastern Jews who were expelled from Spain and Portugal in the 15th Century.

11.28.2021 8:09am - (Replying to @nathanb_science) I do not see how we can construct an SCM in the idioms of conservation laws. The arrows stand for relations among variables.

11.27.2021 4:36pm - I wish I could join your excitement, but a book that starts and ends with Potential Outcomes would most likely deprive readers of three key elements of Causal Inference: Transparency, Identifiability and Testability. A glimpse at Martin's book tells me I was right. Wasn't I?

11.27.2021 4:15pm - I am melting with this song, this voice, this innocence, this holiday -- I can't help but share it with all readers, and let the whole world melt with us. Ocho Kandelas para me! Happy Hanukah!

11.27.2021 12:58pm - (Replying to @tailcalled) There's no question conservation constraints are critically important, even indispensable. But, cognitively speaking, they do not possess the explanatory power of event-based explanations. Good example: Feynman's explanation of why gas heats up when compressed.

11.27.2021 12:06pm - In case you find the link broken, here is a better one for "Causation in Science", a book review:

11.27.2021 12:00pm - Econometric education is in such a dire need of Causal Inference understanding (here are two proofs:, that I feel obligated to spread the word on Mixtape Sessions as wide as I can.

11.27.2021 9:29am - (1/4) A new book "Causation in Science, by Yemima Ben-Menahem makes the point that, in ordinary scientific practice, conservation constraints often serve as explanations. For example: "Why did the roller coaster slow down"? "Because energy must be conserved"
11.27.2021 9:29am - (2/4) To include such constraints as "causal explanations" Ben-Menachem advocates abandoning the paradigm that causation is a relation between events, or variables. I hesitate! Considering the fact that conversational utterances are in themselves products of language constraints,
11.27.2021 9:29am - (3/4) they are hardly in a position to illuminate the nature of causation. To elaborate, all scientific languages, until 1920, were wedded to the symmetric algebraic equality "=", lacking notation for the assignment operator ":=", with which causal asymmetries can be expressed.
11.27.2021 9:29am - (4/4) We can't let this deficiency of scientific discourse dictate what is or is not a "causal explanation". Scientists used conservation laws as explanations to bypass this deficiency. The roller-coaster slows down because the gravitational force acts upon it, not to conserve energy.

11.27.2021 5:49am - (1/3) Here is another example how University administrators perpetuate, rather than curb campus hostilities. A Palestinian student at USC, elected "EDI-senator," made death threats against Zionist students and @USCViterbi came up with an empty condemnation:
11.27.2021 5:49am - (2/3) No mention of the offense, nor the offender, nor the offended; even the offensive statements are not condemned, they are merely "disturbing". Quoting: "we understand they are disturbing. USC rejects and condemns hatred in all its forms." Sorry, @USCViterbi , those who
11.27.2021 5:49am - (3/3) "condemn hatred in all its forms" condemn none. Those who can't name the offense or offender, meekly cater to the latter. Most importantly, those who refrain from supporting the offended [by NAME] condone the offense and invite many others.

11.26.2021 12:00pm - LA residents who wish to join me in celebrating Jewish Thanksgiving Day, will be able to do so Monday 7pm, when we will be commemorating The UN VOTE of Nov. 29, 1947, with a panel discussion and re-enactment of the vote. Free! Twitter invitation + flyer

11.26.2021 9:34am - (Replying to @anaterezaduarte) I hope you say nice things about #Bookofwhy. After all, never mind answering, how can scientists live w/o asking?

11.26.2021 6:30am - (1/2) If you are an alum of a US University you probably got a message reminding you that Thanksgiving marks the time of "giving back". I've sent them the Tweet below with a copy of Lerman article on HOW to give back I urge you to do the same, if you want (1/2)
11.26.2021 6:30am - (2/2) University officials to understand that alumni are watching them, and are expecting them to end the BDS circus by telling their campuses squarely what they know about the genocidal aims of BDS. Thanks.

11.26.2021 5:38am - Readers ask what NYU administration can do to stop the BDS circus and protect their good name. I wrote about it here: @erikbiz , @nyuniversity , @nyulaw , @nyulawreview , @PresHamiltonNYU

11.25.2021 9:54am - (Replying to @joshua_thorpej) Yet it was invented in all primitive tribes, no exception, so it must have served some computational function, possible w/ survival advantages, eg. communication, market for promises.

11.25.2021 9:39am - (Replying to @eslmaher) This is indeed what we do! In the Jewish traditions we lament the destruction of Jerusalem by narrating, again and again, how beautiful a city Jerusalem was before its destruction. It inspires hope and rallies people for action and change, b/c "you deserve better"

11.25.2021 7:22am - (Replying to @topkara) But don't forget that "understanding" is defined as a capability to answer questions from all three levels of the causal hierarchy.

11.25.2021 7:19am - (1/2) This leads to an interesting dilemma: Would/should robots invent God once they acquire AGI? Computationally, it would be beneficial to have one explanatory variable for all 'why' questions. On the other hand, lacking parenthood experience, that variable would fail to evoke
11.25.2021 7:19am - (2/2) the "our father in heaven" metaphor, so its explanatory power will not be as compelling as it is for humans. Tough question: Would/should pilgrim robots celebrate their first harvest in a Thanksgiving ceremony? A tough one to consider today, as we gaze at the poor turkey.

11.25.2021 6:22am - (Replying to @ColocateDisrupt) It's much easier and, indeed, this is what Sapiens did as soon as they acquired the cognitive ability to imagine counterfactuals. You could never convince a monkey to give you a banana by promising him limitless bananas in monkey afterlife [Harari, Sapiens]

11.24.2021 9:47pm - (1/ ) (Replying to @erikbiz) NYU administration should tell those "editors" that, using the name NYU in the title of their publication entails compliance with certain standards of civil discourse, among which is respect for other minorities on campus and their symbols of identity. Once they make it clear, 1/
11.24.2021 10:22pm - (2/2) (Replying to @yudapearl and @erikbiz) next step is to identify Jewish, Zionist and Israeli students as such minorities, Israel being their symbol of identity. More importantly, treat these "editors" as if they had published a racist, anti-Muslim, anti-women, white-supremacist editorial. 2/2

11.24.2021 8:54pm - (Replying to @Jay5w) You just used the word "better", which invokes a comparison, albeit implicitly.

11.24.2021 8:48pm - (Replying to @killerkristi) Please join me in listening to how the founders of BDS define their aims: and why even progressives like Chomsky declared it "hypocrisy rising to heaven"

11.24.2021 8:28pm - (1/ ) As we celebrate Thanksgiving, I am reminded that "thanking" is purely counterfactual concept; you don't thank anyone unless you can imagine an alternative world, significantly worse than ours. While such worlds are sometimes hard to imagine, the mere sound of the word "thanks" 1/
11.24.2021 8:28pm - (2/2) evokes the existence of those worlds and puts a smile on your face: Yes, things could be so much worse. Now I understand why we celebrate Thanksgiving Day -- we celebrate the capacity of our software to invoke counterfactual logic and smile in the face of adversity.

11.24.2021 7:05pm - NYU Denounces Student-Run Publication's Endorsement of BDS via @jewishjournal
11.24.2021 7:05pm - Here is NYU again, my alma mater, naively assuming that reaffirming its commitment to academic exchange is sufficient for stopping Zionophobic groups from using the good name of the university to promote hateful propaganda on campus.

11.24.2021 5:11pm - (1/2) @artistexyz , glad you decided to include Selection Bias (SB) in your book. SB is a funny topic. Statisticians believe they own it because of Abraham Wald. Economists, because of Heckman Selection. But none formulated the problem in its generality, not to mention solving it (1/2)
11.24.2021 5:25pm - (2/2) because both shun graphical models (eg., see why: However, I could not find where you present the solution by Bayesian Networks Is this still in the making?

11.24.2021 9:08am - Hard to believe! Chanukah will be with us in 4 days. So let's start with Ocho Kandelikos and then LA Shir Choir, under my humble baton, in Sevivon (The Dreidel), about 11 min into the video. Happy Chanukah.

11.24.2021 8:09am - Here is an illuminating panel of prominent policy makers, analysts, solution-proposers, and PR analysts on the Israeli-Palestinian conflict: Amazing to watch how consistently the latter continue to ignore the big elephant in the room: Palestinian agenda.

11.24.2021 6:33am - (Replying to @killerkristi) Please join me in listening to how the founders of BDS define their aims: and why even progressives like Chomsky declared it "hypocrisy rising to heaven"

11.24.2021 6:33am - (Replying to @bzaharatos and @lakens) Not sure what is meant by "precisely define", but @yudapearl defines all shades of "cause" (direct, indirect, necessary, sufficient, singular, type, etc) in terms of ONE primitive: "listen to", the latter being the least controversial and most intuitive primitive ever proposed.

11.24.2021 5:44am - Past behavior is a weak evidence for exposing Palestinian agenda. The present is much more compelling: "Dear Palestinian neighbor, forget the past, can you utter the words 'equally legitimate and equally indigenous'? Here and now, in Arabic, please?"

11.24.2021 5:23am - (Replying to @ABookmaker) Support for anti Zionism will continue as long as Democrats and social media types refer to it as "anti Zionism" instead of Zionophobia -- the former connotes a semblance of rationality & legitimacy.

11.24.2021 4:58am - There is one positive aspect to the walk out on this Israeli-Palestinian dialogue: TRUTH. One side simply has zero interest in dialogues, a fact that peace-negotiators, dialogue-makers and ordinary folks have tried so long to refuse to believe. Time for knowledge update.

11.23.2021 7:19pm - Jerusalem terrorist's last will and testimony revealed | World Israel News
11.23.2021 7:19pm - An important document for the archives of #Palestinian_Mentality_Week Nov. 21-28, 2021. A glimpse at the victims:

11.23.2021 6:27pm - (Replying to @artistexyz) Found it, thanks.

11.23.2021 5:35pm - (Replying to @rao2z) What I could not find was the use of Bayes Nets to solve Selection Bias problems, as is done here:

11.23.2021 5:11pm - (1/2) @artistexyz , glad you decided to include Selection Bias (SB) in your book. SB is a funny topic. Statisticians believe they own it because of Abraham Wald. Economists, because of Heckman Selection. But none formulated the problem in its generality, not to mention solving it (1/2)
11.23.2021 5:25pm - (2/2) because both shun graphical models (eg., see why: However, I could not find where you present the solution by Bayesian Networks Is this still in the making?

11.23.2021 4:45pm - (Replying to @SurviveThrive2 @ylecun and 11 others) Where/when?

11.23.2021 7:07am - (Replying to @LekhtNaya) Where have you noticed this heretical "trend"? I was under the impression that I am a lone soldier with that banner. Do I have comrades? I can use their help.

11.23.2021 7:00am - 11-year-old finds `Holy Jerusalem' silver coin likely minted in the Temple
11.23.2021 7:00am - Beautiful find! Yet Ancient coins, ancient synagogues, ancient scrolls, Arch of Titus, etc. will not convince Israel's neighbors that history calls for co-existence. So, I am using Hanukah, a vibrant, un-fakable Living-Trust-Deed to Jewish History, see

11.23.2021 2:45am - (Replying to @borgesvit_r @aihkas and @sharkdiveruk How do you know you're winning a debate? When your opponent says: "Every educated person knows...", meaning: "I can't substantiate my claims". Arafat became famous for using it on Clinton (2000) "everybody knows Jews never had a Temple in Jerusalem". #Palestinian_Mentality_Week

11.22.2021 8:13pm - Here is an interesting paper by Shingaki & Kuroki showing that probabilities of causation can be identified when certain conditions are satisfied in the model's graph.

11.22.2021 5:36pm - Lessons Learned When Planned University Giving Inadvertently Funds Antisemitism
11.22.2021 5:36pm - (1/2) This article touches on the future of Jewish education on US campuses. Please bring it to the attention of every "would be donor" that you may know. Universities keep such donors top secret; blinded to campus controversies, unknown to the public, and unknown to each other. (1/2)
11.22.2021 5:36pm - Help them open their eyes to what is going on in academia, how (some) universities misuse their donations, how their names can be used to intimidate and marginalize their grandchildren, and what they can do about it. It's urgent.

11.22.2021 4:39pm - (Replying to @mrcvhoy and @DavidDeutschOxf) Agree with @DavidDeutschOxf that "excess death" is a counterfactual quantity, but this does not make it "vague". We can define it formally and estimate (or bound) it precisely, modulo what we are willing to theorize about reality.

11.22.2021 2:43pm - (Replying to @TaraZehut1981 @aihkas and @borgesvit_r) Evidently, the barbaric murder of Eli Kay has unleashed an unending fountain of wishful visions from Israel's neighbors. It seems useful to extend "Palestinian Mentality Day" into "Palestinian Mentality Week". I am all for it: Nov. 21 to Nov. 27. Just in time for Hanukah.

11.22.2021 10:28am - (Replying to @JustCallMeOmer and @Bonerrchamp) It's Palestinian Mentality Day. Exhibit 2

11.22.2021 4:58am - "You have a partner", Western intellectuals tell Israel. "You have a partner", wishing diplomats repeat. "Listen to my partners!" Israel implores. "What partners?" Palestinians retort. November 21, "Palestinian Mentality Day" declared! Exhibit 1, below!

11.21.2021 4:48pm - My son

11.21.2021 4:48pm - (1/2) Eli, Eli, my newly fallen son. When I weep for you, I weep for my son Daniel - your brother in pain - two treasures crushed in the claws of Barbaric terrorism. Danny and Eli, my two fallen sons, it was not the barbarians alone who killed you; Western intellectuals were (1/2)
11.21.2021 4:48pm - (2/2) there all along, watching Barbarians sharpen their knives, explaining, justifying, rationalizing, legitimizing, everything but criminalizing. Let there be no silence on your graves, Eli and Danny, no rest, until those justifiers stand trial, in the high court of history. (2/2)

11.21.2021 3:11pm - (Replying to @yihanxu) You are absolutely right, in these two cases, and also in Joshua (7:16. the case of Achan), as well as ancient Egypt, casting lot meant asking God for a signal. I described the transition from signal to randomness in

11.21.2021 1:06pm - (Replying to @yihanxu) Jonah's lots are even more significant, they convey a divine message of guilt and innocent, not merely who will get what from the booty.

11.21.2021 12:39pm - (Replying to @desai_pratik) I believe these innate templates are indeed causal-graphs, or something close to it, perhaps an algorithm for constructing such graphs. What I cannot accept is the attitude that we do not need to worry about how that template is represented, b/c data will discover it.

11.21.2021 1:28am - (Replying to @DKedmey and @MEMRIReports) Great idea. The Zionist new/old slogan in every language! But branding and promoting take precious resources. All Jewish resources are currently in the hands of a fragmented and yesterday-looking leadership.

11.21.2021 1:18am - (Replying to @DKedmey @MEMRIReports and @EinatWilf) I would ask her, beg her, to lead the war-cabinet that we must urgently convene to replace our failed, fragmented leadership, who can't unite the community against the new enemy, Zionophobia, and can't bring themselves to modernize their rusty WWI pistols.

11.21.2021 12:47am - (Replying to @DKedmey and @NeilRogachevsky) @NeilRogachevsky raises good points. Investment in Jewish-Zionist education is the key, including Hebrew, Bible and History. To be chosen carefully, because much of Jewish education is already hijacked by TikunOlamniks, & I bet aberrations like Beinart do speak Hebrew.

11.20.2021 1:23am - It is sad though that, among the 23 discussants, Nathan Scharansky was the only one to tell American Jews (in essence): Stop torturing yourself with how you should deal with Israel and start asking how Israel can save you from committing suicide, as a people, and as a culture.

11.20.2021 12:43pm - A good photo can make you smell the place.

11.20.2021 10:16am - (Replying to @Eve_Barlow) It is happening because Palestinians, unable to promote their victimhood on its own merit, must latch on to other narratives for attention. Who ever heard about a society defining itself by the demise of its neighbor, yet proclaiming itself "victim".

11.20.2021 4:09am - Parents pride. I, too, would have travelled 10.000 km on such an occasion.

11.20.2021 1:10am - Seeing the term "hallucinated SCMs" gaining popularity among some readers, my tweet below comes to their defense by viewing SCMs as the engines that drive our scientific methodology. The basic roles these "Hallucinated" theories play in ML are discussed in

11.20.2021 12:32am - (Replying to @artistexyz @erichorvitz and 2 others) Neural Nets cannot look like Causal Bayesian Networks, no matter what names they take on, because the latter is about how the world behaves, the former is about what we do with the data received.

11.19.2021 10:43pm - Glad to see this paper discussed by researchers concerned with "trustworthiness", a notion bound to counterfactual reasoning, hence, the Ladder of Causation.

11.19.2021 10:13pm - (Replying to @austinvhuang @holdmytowel and 3 others) Replace "given the graph" with "given a world" and you are in standard scientific method, going from theories to evidence, then backward, to refuting theories when evidence so dictates. Take away the former, gone is the latter. Take away the graph, and you lost touch w/ reality.

11.19.2021 12:06pm - Fascinating conversation by Stephen Worfram on the history of science from Wiener to AI Not much on causality (yet) but the evolution of ideas is instructive and inspiring.

11.19.2021 11:34am - (Replying to @DJMinded) Suppose someone proves (factually and mathematically) that the ONLY way of saving the idea of Israel is to implement the policy of Israel. (Don't jump, just suppose, no one would call you right-wing for supposing). What then? Who is the oppressed? Does it resonate w/ scripture?

11.19.2021 6:54am - (Replying to @anaterezaduarte and @causalinf) Great! Please share.

11.19.2021 6:17am - (Replying to @KordingLab) Love it!

11.19.2021 6:13am - Readers complain that my vision of Jewish future is too gloomy. On the contrary, I predict a forward-looking, Israel-inspired future, free of its Zionophobic yesterday-Jews who are currently confusing our youngsters into a history-blind type of Jewishness they find irrelevant.

11.19.2021 5:36am - (Replying to @jep353) The next diving into some equations is done in Primer, which @ScotMueler has taught in high school. Its wrong to assume that rigor implies torture.

11.19.2021 5:24am - (1/3) @PengzhouW , this phenomena is of grave concern to me: ML ppl, new to CI, are drawn to PO (b/c once they assume ignorability, CI looks like regression -- easy), publish "review papers on CI" inviting their peers to join the causal-ML hype festivity 1/3
11.19.2021 5:24am - (2/3) (even citing "other approaches" like "DAG models" to sound inclusive) resulting in the misconception that CI is just ML with a fancy feather. To protect myself from such "review papers" I would use a few litmus tests. For example, go straight to the "assumptions" section,
11.19.2021 5:24am - (3/3) if it takes "ignorability" for granted, quit. Or, search the figures for DAGs and see if the text uses them to define interventions or counterfactuals. If not, quit. Such crude litmus tests have saved me hundreds of hours. Try it.

11.19.2021 3:32am - That's a great idea: #Bookofwhy for dummies. Searching for a co-author.

11.19.2021 3:12am - I wish I could play the Ukulele, so I could play in Israel and be threatened by Hamas -- what an honor; side by side with Justin Bieber.

11.18.2021 2:23pm - (Replying to @algentilini) Thank you. I should bug the publisher, we must have #Bookofwhy in the language of Galileo.

11.18.2021 12:06pm - What 23 hard-thinking Jews think about Israel and why I (one of them) say that Judaism is due for an irreparable yet healthy split: Israel-inspired forward-looking Jews vs. Israel-bashing yesterday-Jews. A "two fate solution" for what once was one people.

11.18.2021 11:05am - (Replying to @MichaelALewis10) I think you should read Woodward after Pearl, but only if needs arise; see why: He uses SCM (DAG) as a basis for defining "interventions", then posits "intervention" as a primitive, and never tells us how human knowledge is organized. I'd get confused.

11.18.2021 9:09am - Hanan Ashrawi has a point. US Amb Linda Thomas Greenfield made the mistake that most American diplomats make, that Palestinians are persuadable by reason. She should have used an ultimatum: You want statehood, freedom and dignity? Your children must know that you have neighbors.
11.18.2021 9:09am - (2/2) "We know how to educate our children" Ashrawi howls, thinking that the world is too dumb to see what they learn, or that people are too dumb to understand of that those who vow to eliminate their neighbors are hardly in a position to be given the opportunity. @USAmbUN @UNRWA 2/2
11.18.2021 9:09am - Ashrawi

11.18.2021 8:11am - (Replying to @Kalana_World) I appreciate this feedback, because I am trying to refrain from posting negative comments on published paper; knowing that I've saved people precious time pacifies my guilt somewhat. The deficiencies noted are typical to papers avoiding graphs & ignoring the Ladder of Causation.

11.18.2021 8:00am - (Replying to @artistexyz @NandoDF and @DeepMind) Love your Piscina Mirabilis vision of Sequential Backdoor, which is truly a powerful solution to the problem of sequential decisions: Each decision is contingent on an observation affected by past decisions, see However, did you discuss it on your blog?

11.18.2021 7:42am - (Replying to @ccaballeroh10) I may have commented on SEMNET listserve, I do not recall. Keith is so cryptic and de-grounded that it is extremely difficult to pin point his difficulties, especially concerning an equivalence that is so visibly demonstrable. Glad someone else did it.

11.18.2021 3:39am - No. I cannot recommend this survey of causal inference It assumes away the problems of identification, transparency and testability and, ignoring the guidance of the Ladder of Causation, confuses what is and is not inferable by black-box machine learning.

11.18.2021 2:45am - I am glad N. Weinberger took the time to refute Keith Markus (2021), who argued that "the causal modeling frameworks of Pearl and Rubin are not strongly equivalent" See also my response to Guido Imbens, who ignored the equivalence.

11.18.2021 1:15am - Confirming the great news! The couple landed in Israel, and are on their way from the Airport to their home in Modi'in. Everyone is praising the Turkish authorities, me too, but I ain't going to any conference in Turkey before Erdogan steps down.

11.17.2021 8:18pm - I hope they bring with them all the photos they took of Erdwan's dressing room.

11.17.2021 8:14pm - Unbelievable! I've just watched TV and all the "experts' predicted them moths, perhaps years in prison. Worth a global sigh of relief -- HaGoMel

11.17.2021 4:18pm - Another must read by @EinatWilf

11.17.2021 4:11pm - As a statistician in exile, what IS the WAY of inferring risk of bad outcomes from data? Are our fellow statisticians aware of this WAY?

11.17.2021 3:55pm - It was a good talk, I agree. Not sure I can replicate it today, given all the hype about "causal AI" and "Causal ML"; its hard to make people re-focus on the basics once they operationalize some aspects of the problem.

11.17.2021 3:13pm - Fascinating conversation by Stephen Worfram on the history of science from Wiener to AI Not much on causality (yet) but the evolution of ideas is instructive and inspiring.

11.17.2021 12:07pm - (Replying to @archstreetllc) Stories. I like it! And the stories, invariably, were about human-like actors, laden with beliefs, intentions and desires, a poetic description of what graphical models convey in the dryness of their arrows.

11.17.2021 11:42am - This thread received a "like" this morning, so Twitter brought it to my attention. Wow, it's so true! How else would you represent, qualitatively, what you know about the world? Any suggestion? Perhaps from those not using graphs? Say DL folks? Or economists?

11.17.2021 7:41am - You are right!!! #Chanukah will descend on us in 10 days (Sunday Nov. 28) and what's a better way to welcome the holiday spirit if not through this beautiful version of Ocho Kandelikas Happy Hannukah!!

11.17.2021 7:15am - Israeli Great-Grandmother of 21 Crowned Miss Holocaust Survivor 2021 via @TheMediaLine
11.17.2021 7:15am - Touching my heart!!!

11.16.2021 1:20pm - Congratulations go to Kevin Cummiskey and Jason Roy, recipients of the 2021 ASA Causality in Statistical Education Award: I hope this annual award will continue until causality becomes an organic part of statistics.

11.16.2021 7:06am - Did you ever wonder what P(y|do(x),z) stands for, when Z is affected by X? Or why the backdoor criterion insists on Z not being affected by X? A colleague asked me these questions and thanked me profusely for answering them clearly here: So, I'm sharing.

11.15.2021 3:33am - (Replying to @sanmayd and @rao2z) I am clear as to the kind of people who find causal reason counterintuitive. Where is the shock? What bad habit makes the most intuitive part of our thinking sound counterintuitive.

11.15.2021 5:58am - (Replying to @yudapearl @sanmayd and @rao2z) I meant "I am NOT clear as to the kind of people who find causal reasoning counterintuitive" Sorry for the typo

11.15.2021 12:45am - (1/2) Chapter 4 of Causality discusses the way actions are handled in traditional decision theory (even STRIP) and the way they are handled in modern causal inference. Bottom line: the difference lies in the input information and the output queries. Note,
11.15.2021 12:45am - (2/2) actions in CI are named by their direct effects (eg, make me laugh, or raise taxes) not by the changes they introduce in the state of the world (as in planning or RL). We must therefore use the do-operator, do(X=x).

11.14.2021 1:57pm - Kibbutz Maagan Michael, 35 km South of Haifa, is not an ordinary Kibbutz. Every year, around October, they dedicate a concert to Daniel Pearl World Music Day. Listen to the Shachar (Dawn) choir singing "I have seen a most beautiful bird":

11.14.2021 5:04am - After lengthy deliberations, in-depth discussions with readers and colleagues and, not less important, thousands of tweets and retweets, I am inclined to agree with @timothyvienne .

11.14.2021 4:13am - (1/2) (Replying to @havivrettiggur @TaymAlsalti and @DJMinded) Fully agree with (1), (2) and modified (3). Israel can assist Palestinian's ideological transformation by declaring the 2 state solution a Government Policy contingent on the transformation, with clear milestones. More importantly, the world can assist the transformation by
11.14.2021 4:25am - (Replying to @yudapearl @havivrettiggur and 2 others) making it absolutely clear to Palestinians that the transformation is key to any movement. Addendum: Ideological transformation can easily be monitored, once it becomes the center of attention, to replace secondary issues like borders, holy places, security arrangements etc.

11.13.2021 10:19pm - (1/2) (Replying to @TaymAlsalti @havivrettiggur and @DJMinded) The utility of establishing facts about Palestinians vows to Israel's destruction lies in sparing American diplomats time, effort, resources and disappointment of working on "peace plans" that ignore these vows, thus deepening the problem, by intensifying those vows,
11.13.2021 10:32pm - (2/3) (Replying to @yudapearl @TaymAlsalti and 2 others) raising hopes that Israel's destruction is near, and resulting in greater disappointments and hostilities when those "peace plans" fail. They are bound to fail because Israelis, well informed of the vows, cannot afford to experiment with ideas that diplomats, uninformed of
11.13.2021 10:46pm - (3/3) (Replying to @yudapearl @TaymAlsalti and 2 others) the vows, deem to be reasonable compromises for parties wishing to end a conflict and move forward to build a better future. What both Israelis and Palestinians understand, and those diplomats don't, is that "end of conflict" is not an option for one of the parties.

11.13.2021 2:54pm - Well put. I've always fround the "vicious cycle" metaphore to be misleading. There is no cycle when one side aims to maximize deaths and the other to minimize deaths, or when one says: "equally indiginous" and the other: "You're colonialist" or between "We We We" and "Me Me Me"

11.13.2021 10:34am - (Replying to @TaymAlsalti @DJMinded and @havivrettiggur) I would leave discussions about "fairness" to another occasions (we had many in the past). This thread deals with 2 issues: 1. Is it a fact that Palestinians see the elimination of Israel as the only acceptable, long term solution? 2. Why can't Westerners stomach this fact.

11.13.2021 9:23am - For my personal recollections of Nov. 29, 1947, and why I call it the "Jewish Thanksgiving Day" please watch this youtube video: See you Nov 29, when we give thanks to Lady History, and to 33 countries who voted their conscience:

11.13.2021 9:19am - Invitation: November blesses us with two miracles. 1. The Balfour Declaration (Nov. 2, 1917) 2. The UN VOTE of Nov. 29, 1947 You are invited to join us in the 2021 LA celebration of Nov. 29, 7:00 pm here:

11.13.2021 9:11am - (Replying to @havivrettiggur and @DJMinded) Thanks for chiming in. My strategy in proving Palestinians incurable eliminationism is not to cite rejections of past peace proposals, but to challenge their spokesperson directly: "Face this audience here and now and utter the words 'Equally Legitimate' in Arabic, can you?"

11.13.2021 8:58am - (Replying to @DJMinded and @havivrettiggur) I am not sure what you mean by "majority opinion", "their #colonialist opinion" and "#apartheid label". All labels Palestinians stick to Israel are logical corrolaries of their eliminationist agenda, hence, are as valid and morally justified as the latter.

11.13.2021 6:46am - I must retweet this thread because the question: "How do we know what the vast majority of Palestinians want?" is asked again and again, by well meaning readers, mostly Westerners, who find the evidence so hard to stomach. See also @EinatWilf book "The War of Return".

11.13.2021 6:27am - (Replying to @DJMinded and @havivrettiggur) I would bet 100:1 that you are NOT a Palestinian. How do I know? Statistically. Out of the thousands of Palestinians I met, not one of them would disagree with my statement. Only Westerners refuse to accept the hard fact that Palestinians see Israel as illegitimate & temporary.

11.13.2021 6:17am - (Replying to @DJMinded and @havivrettiggur) This is not "anecdotal evidence" but a most informative and scientific evidence one can obtain for the assertion "vast majority". When you position dozens of sensitive sensors seeking a signal and no signal is detected, it's not anecdotal to conclude: "no signal was transmitted."

11.13.2021 5:29am - (Replying to @PeterS80148175) Judging by his name, al-Kurd may indeed be one of the many thousands of Kurds who emigrated to Palestine in 1930's, following the economic prosperity created by the Zionists, and assumed Palestinian identity. But if every son of emigrants is sent back, Palestine will be empty.

11.13.2021 5:17am - I like the style of your lectures, and your examples, but I would sprinkle them with equations, for the sake of seeing the forest for the trees. I would also discuss how probabilities of causation are identified (bounded) using a both observational & experimental data.

11.13.2021 5:01am - (Replying to @DJMinded and @havivrettiggur) Fair question. Several NGO's in Israel are tuned to Palestinian schools, media & mosques for an INKLING of acceptance -- none thus far, just plans of elimination. From your question I surmise you can name ONE Palestinian who accepts (in Arabic) Israel's permanence; please do.

11.12.2021 5:58am - (Replying to @Mauriciogs99 and @DeepMind) As Imbens used to say: Better LATE than nothing. But to the credit of the @DeepMind team, they do not present it as a new discovery, but as a necessary means for rectifying certain problematic "delusions".

11.12.2021 5:38am - It is hard to believe how resistant American Jews are to the idea that El-Kurd's fantasies (about Israel's disappearance) are not a rare aberration but the ONLY worldview among Palestinians today. Americans refuse to buy it b/c it spoils all solutions they enjoy dreaming up.

11.12.2021 4:58am - (Replying to @Mauriciogs99 and @DeepMind) As Imbens used to say: Better LATE than nothing. But to the credit of the @DeepMind team, they do not present it as a new discovery, but as a necessary means for rectifying certain problematic "delusions".

11.12.2021 4:46am - My personal blessings to the 2021 "Causal Data Science" Meeting, coupled with a hope that in 2-3 years from now we would not need to add the prefix "causal" to "Data Science" meetings; it would be self evident that the "Science" in "Data Science" implies causal inference.

11.11.2021 8:37pm - Great thread indeed ( @havivrettiggur ) to which I add: We need to see more Palestinians speak their minds honestly.. The majority of Americans refuse to believe that El-Kurd's attitude toward Israel is shared by the vast majority of Palestinians; they must see with their own eyes.

11.11.2021 2:09pm - (Replying to @jacyanthis and @DeepMind) I haven't studied it carefully yet. I should.

11.11.2021 9:23am - (Replying to @skoosharama) "People of the Book" remain the most competitive claimants to ownership of ideas, patriarchs, and narratives. Accordingly, they must be visibly identified as less than equal. Do you want quotes? Or would practices suffice?

11.11.2021 8:37am - There is an important adage here. Druids could perhaps be viewed as equal, but Jews should know their place, as it is said in the scriptures.

11.11.2021 8:23am - The logic is simple. If the Balfour Declaration was about Druids, rather than Jews, the Mufti would not start any riot. Jews were a real threat to him, he knew they were genuine home-comers, with no intention of becoming a minority again.

11.11.2021 4:42am - Happy to see that Amit "will show that causal graphs are fundamental to understand the fairness of a ML system." However, #Bookofwhy warns readers against the use of black-box "matching". Finally, the title "CI for ML" sounds inverted to me -- ML is not a goal in itself.

11.11.2021 4:31am - (1/5) Finding a do-operator in a @DeepMind article is a tectonic progress that deserves welcoming blessing. The "delusions" treated in this article are endemic of "Evidential Decision Theory" which Causality (ch 4.1.1 summarizes in a mnemonic limerick:
11.11.2021 4:31am - (2/5) - Whatever evidence an act might provide - On what could have caused the act, - Should never be used to help one decide - On whether to choose that same act. Typical real life ramifications of these delusions are: (1) patients should avoid going to the doctor "to reduce the
11.11.2021 4:31am - (3/5) probability that one is seriously ill" (2) workers should never hurry to work, to reduce the probability of having overslept, and more. The deployment of the do-operator eliminates these "delusions" and has led to the "sequential backdoor criterion" of Sec. 4.4.3.
11.11.2021 4:31am - (4/5) Ortega etal are justified to title the paper "Shaking the Foundations", since even seasoned decision analysts often use the wrong probabilities on their decision trees (see Sec.11.6 not to mention ML researchers, many of whom still equate "knowledge"
11.11.2021 4:31am - (5/5) with "training". I haven't read the last part but would recommend that the authors take a good look at the "sequential backdoor criterion" to ensure that we do not introduce new "delusions" in trying to eliminate old ones. #DataScience #MachineLearning #DeepLearning #AI

11.11.2021 12:51am - (Replying to @stephensenn @EJWagenmakers and 3 others) Honored to see Causality in such stellar company.

11.10.2021 8:42am - (1/2) An interesting announcement has crossed my screen: A company named Causalens is applying Causal Inference in capital markets applications. I have predicted such development a year ago. If academia blinds itself from science, capital marketers will 1/2
11.10.2021 8:42am - (2/2) replace academia, and will teach the science (in their own way) to entrepreneurs and other stake holders who could not get it in college. The trouble will emerge in 3-5 years, when thousands of ML folks will seek retraining, and academia will find itself with helpless Cathedrals.

11.10.2021 7:56am - (Replying to @VC31415) What's the verdict? Is P(Y_x| covariates) identifiable under monotonicity, as it is under linearity? If not, do we have universal bounds?

11.10.2021 6:33am - (Replying to @stephensenn and @Genstat) My point is not that causal notation is useful but that it is necessary whenever "causal questions are asked". In other words, such questions cannot be asked formally in the language available to John Nelder in the 1980's. That's why my curiosity is heightened: How did he do it?

11.10.2021 6:22am - (Replying to @stephensenn and @Genstat) Causal analysis can only help researchers who can articulate what their research question is, and do it in a general language, say statistics, which is communicable across disciplines.

11.10.2021 6:11am - (Replying to @vdcalhoun and @mrsaladfingers2) try "mud does not cause rain".

11.10.2021 6:10am - In 2002, when Causality received the Lakatos award from LSE, London appeared much more civilized. What we have been witnessing in the past 20 years is a banalization of evil masquerading as noble passion. Historians tell us that Nazi youngsters on Kristallnacht felt same passion.

11.10.2021 5:53am - (1/ ) A great piece, indeed, with one sad observation: This young generation of "enough is enough" activists lacks a Herzl to coordinate its many activities and direct them systematically toward achievable strategic goals. All our institutional resources, eg, Hillel, ADL
11.10.2021 5:53am - (2/ ) AJC, Federations, etc etc. are in the hands of divided leadership, armed with rusty ideological weapons, who fails to respond to events on the ground. Absent a Herzl, how about a "war cabinet"! @EinatWilf , willing to assemble one?

11.10.2021 4:18am - (Replying to @learnfromerror @stephensenn and 2 others) Not so? Please write down formally "Mud does not cause rain". The nice thing about notation is that, when science is facing a profound conceptual revolution of thinking, it shows up in the notation - no escape.

11.10.2021 4:09am - (Replying to @stephensenn) "Effect" is a causal notion. Moreover, are those "auxiliary assumptions" articulable in statistical notation? Now you see why I insist on ONE SENTENCE rather than shelves full of statistics books. The paper cited (unopenable, probably by Greenland), uses counterfactual notation.

11.10.2021 3:53am - (1/ ) Thanks for reminding me of this excellent (ancient) paper I wish I could write that clearly today. But note what it says about why econometrics fell from grace after the founding fathers handed the field over to their (not so brilliant) disciples.
11.10.2021 3:53am - (2/ ) Unable to "do it in their head", and lacking notation to remind them what needs to be done, they let their field slide into the "confusion of the century". How did statistics cope with same "lack of notation"? Was it saved by Nelder? I am waiting to see what notation he used
11.10.2021 3:53am - (3/ ) for saying "mud does not cause rain". I hope @stephensenn unravels this notation for us, who are not versed in "block design" and yet are curious to see how a "formal analysis" of causal questiond could be conducted with no new notation.

11.10.2021 2:52am - (Replying to @stephensenn) Just to make sure that Yate's understanding of "confounding" coincides with mine (ours?), can you define it in one Tweet ? And, please, in the language of Stat 101. Do we need to know "factorials", "fractional" and other trade secrets before understanding what "confounding" is?

11.10.2021 2:22am - Here is were I get stuck: (1) He mentions "confounding" which we know is a causal, not statistical notion. Does he really mean he has a causal model in mind? (2) What is the aim of the "analysis"?

11.10.2021 2:17am - (1/ ) What I am missing from Nelder's story is: (1) What was his research dilemma? (2)"know that I am right" about what? (3) "found a method", starting with what assumed knowledge? Correlation matrix? Finite data only? Notice how DAG-based stories specify all these elements: 1/
11.10.2021 2:17am - (2/ ) (1) Research question (2) Assumptions, (3) type of data available. Notice also their absence in Nedler's story, which makes it hard for me to follow, to the point where I begin to believe that there is something else DAGs may offer researchers: a universal structure of narratives

11.10.2021 1:58am - (Replying to @stephensenn and @f2harrell) By saying "it is formal" you mean, I suppose, "it is not based on causal assumptions", only on statistical assumptions. Some people get awfully offended when someone insinuate that "DAG methods are not 'formal'".

11.9.2021 8:38pm - (Replying to @KMcIntyre94 and @PhDemetri) Strange! I think so too. But I was afraid to say it, lest someone accuses me of being biased or worse.

11.9.2021 6:50pm - Sharing a youtube panel on the Persecution of Minorities in the Middle East: It has been a privilege to partner with @ADL and the Committee to Protect Journalists @CPJMENA

11.9.2021 5:29pm - At the risk of downing my Twitter account I am tweeting a Twitter-critical interview with my friend Rabbi Abraham Cooper, who doesn't buy Ms. @FadahJassem apologies. If she regrets past stupidities, all she has to do is tell us today how Israel WAS born.

11.9.2021 5:00pm - This graphical criterion for selecting "best" covariates should be celebrated by my statistical mentors @stephensenn , @f2harrell ,., because, even if statisticians knew it all in 1925, they certainly couldn't do it in their head. Can it be done in Nedler's calculus of experiments?

11.9.2021 4:11pm - Alerting Jewish students among our followers: A serious mistake must have occurred on @ADL 's website. @JStreetU is NOT "pro-Israel activists" but one of Israel's misguided defamers, assisting Zionophobic hostilities on campus. I hope @ADL correct this oversight. @JGreenblattADL .

11.9.2021 1:24pm - (Replying to @Emily87460254) DL - Deep Learning ML= Machine Learning CI = Causal Inference SCM = Structural Causal Model

11.9.2021 1:22pm - This is a powerful generalization of (Henckel et al., 2019) results, from linear to nonparametric models. We should add it to our "crash course in good and bad controls" @AnalyseReal

11.9.2021 9:40am - (Replying to @mrsaladfingers2) I am not sure the "regularization" metaphor is valid, because a causal model does more than just "limit"; it takes correlations from different parts of the data and combines them in non-trivial ways. Moreover, unlike regulation, the combination varies with the world & w/ query,

11.9.2021 8:33am - (1/ ) Agree, but note two differences. (1) The word "identification" means nothing to DL folks, because it's model dependent so it ain't taught. (2) The word "parameter" evokes resistance everywhere, e.g., "who cares about model's parameters?" or "the answer we need may reside in 1/
11.9.2021 8:33am - (2/2) in not one but a combination of several "parameters", or " there are no parameters in nonparametric models". In comparison, "killing vs. savings" people drives home the point, even to DL folks. @ecotweets , @causalinf , @VC31415 , @PHuenermund 2/2

11.9.2021 8:02am - (1/ ) We have had a lively debate in 2019 regarding the notion of "blocking" in experimental design. My statistics mentors @stephensenn @RonKenett @learnfromerror , @f2harrell tried to convince me that it is not causal, but a statistical notion. I remained skeptical because I could
11.9.2021 8:02am - (2/ ) not understand from their description what makes one blocking good and another bad (my fault). This paper has enlightened me: perhaps because it uses causal graphs (what else?). Q. Has it captured the essence of "blocking"?

11.9.2021 6:17am - (1/ ) A reader asked what I meant by "undoable". Let me explain. By "undoable" I mean "It is mathematically provable that every answer based on data only may have a conflicting answer in the real world, equally compatible with the data". So, when a DL system says: "This vaccine 1/
11.9.2021 6:17am - (2/2) is killing people," it may actually save lives. This notion of "doable" requires a model of reality and an understanding how data are generated by it -- a two-body science. It cannot be conceptualized, and surely not "proven" in DL, where one looks at data only. 2/2

11.9.2021 5:47am - (Replying to @artistexyz and @stanislavfort) I tried to watch the video you linked to, but it was unavailable.

11.9.2021 4:18am - (Replying to @learnfromerror @stephensenn and 2 others) Not so? Please write down formally "Mud does not cause rain". The nice thing about notation is that, when science is facing a profound conceptual revolution of thinking, it shows up in the notation - no escape.

11.9.2021 12:19am - (Replying to @stanislavfort) When I say "undoable" I mean "It is possible to prove that every answer based on data only may have a conflicting answer in the real world, equally compatible with those data"

11.8.2021 10:11am - (Replying to @SurviveThrive2 and @stu_frost) Isn't that enough to consider causality when generating an explanation for a human user?

11.8.2021 5:35am - I love the title: "Students for Just Us in Palestine". Brilliant! I've long said, if we do not win this war in court or in academia, we can surely win it through poetry and humor. Our enemies are just too grotesque for us to miss a laugh.

11.8.2021 5:12am - (Replying to @BorromeanBholes) Gee! I love it! I can hear the music ringing in my ears already: Cathedrals for a handcuffed God, Priests to praise His bright shackles, Donors to erect more cathedrals, And Public to pay for it all.... Yes, Public to pay for it all.... Love it.

11.8.2021 4:41am - Someone just asked and deleted: "And what if we don't have a model?" I answered: Much better! The answer is trivial but extremely valuable: You can't do it, no matter how big your data. Now you can go back to your donor and say: "We do not need to build another cathedral"

11.8.2021 4:14am - (Replying to @db_rumble @gottfriedmath and 2 others) Causal models can be viewed as a game against nature; does any version of your theory capture this case?

11.8.2021 3:56am - (Replying to @JohnVial) This is the first time that I hear about such difficulty, and I am alarmed. Please tell us what your research question is, what assumptions you'r willing to make about the world, what data you have and, of course, where do you get stuck.

11.8.2021 3:35am - Sharing 3 videos from the Los Angeles Press Club Awards Ceremony, featuring acceptance and presentation speeches by: (1) Bari Weiss: (2) Sanjay Gupta: (3) Judea Pearl:

11.8.2021 12:47am - What the Ladder of Causation tells us is that "keeping throwing data and computation at a problem" does not get it solved. No matter how much computation you throw into measuring the base of a triangle, you get no information about its height.

11.8.2021 12:26am - (1/ ) Nancy Cartwright has a new paper promoting "evidential diversity": It echoes my usual problem with Catwright's writings: While convincingly arguing for supporting "many subsidiary claims that must hold if the overall one is to be true," she fails to
11.8.2021 12:26am - (2/ ) tell us how this kind of support differ from the way SCM establishes the truth of the overall claim, or the probability of that claim. See my open letter to Cartwright

11.7.2021 11:11pm - (Replying to @joedotfaith and @tdietterich) I'll go along with the "system architect" metaphor, though my old "project manager" was very good in dividing a task into subtasks, assigning subtasks to the right people, assessing what they accomplish and combining them coherently. That's what a causal model gives us.

11.7.2021 9:18pm - While it may be pointless to discuss what DL can or cannot do in conjunction with other sources of knowledge, it is in fact essential, to discuss what DL cannot do relying on data ONLY, with no knowledge of the data generating process. #DeepLearning is #MachineLearning #Bookofwhy

11.7.2021 8:57pm - (Replying to @OriKatz3) We need to start with a query: What are we trying to explain?

11.7.2021 8:48pm - (Replying to @tdietterich) All true from my perspective too, with one twist: What we currently call "Causal Inference" is already a hybrid architecture, with the causal model serving as a "project manager", identifying subtasks that could be accomplished by DL tools and combining the results coherently.

11.7.2021 6:46pm - My perspective: Most ML researchers are missing two basic points. (1) All the nice goodies (eg, explainability, transfer learning, fairness, data fusion, etc) are doable once you have a causal model. (2) None is doable w/o a causal model. DL = Cathedral for a handcuffed God.

11.7.2021 11:59am - (Replying to @DonskerClass and @eliasbareinboim) The canonical reference to SCM is chapter 7 of Causality The acronym NPSEM-IE appears only in the writings of those who criticize SCM for invoking SOME error-independence assumptions, as if it is avoidable.

11.7.2021 11:42am - (Replying to @franciscpitthan) SCM remains the universal causal framework, cyclic or a-cyclic, together with the 1st Law of causal inference. We simply need to acknowledge that some of the powerful methods provided by the 2nd Law and do-calculus are no longer applicable. Note, no other framework provides them.

11.7.2021 10:29am - (1/ ) Comments: (1) Any econ. class that starts with "causality" and "causal hierarchy" is bound to be on the right track. (2) The term NPSEM-IE is a misnomer, it should read SCM, or "Nonparametric Structural Equation Model with SOME Independent Errors", or Semi-Markovian SCM. (3)
11.7.2021 10:29am - (2/ ) (3) Curious: what are "implicit characterizations: e.g. optimality". (4) After "take assumptions on potential outcomes as primitives" add: "and lose comprehensibility and testability. (5) "What experiments don't identify", mention bounding of individual causal effects,
11.7.2021 10:29am - (3/ ) using fusion: (6) The "minimal assumptions" assumption in "Credibility Revolution" is a myth; structural analysis can commence with as many assumptions. (7) The "CI Road Map" is a gem; I hope every econ. student chants it as a daily mantra.

11.7.2021 9:04am - Apropos, here is the link to @BHL conversation at the Wilshire Temple, last Sunday:

11.7.2021 8:09am - (Replying to @trulyconservat3 @BHL and @FareedZakaria) Thanks for letting me know about @FareedZakaria show.

11.7.2021 8:05am - (Replying to @MarcosGallacher) Can you explicate what you mean be "challenging"? For me, stat. is a subset of econ., focusing on Rung-1 tasks.

11.7.2021 8:01am - (Replying to @charleskfisher) Beyond me. The Lord Paradox is still paradoxical under the Hall=Diet assumption, yet Senn insists on addressing the Hall =/=Diet complication. Why? What for? Can we settle the Hall=Diet case first? Beyond me.

11.7.2021 7:56am - Hats off to the Jewish Journal which, unlike most of the press, Jewish and non-Jewish, refused to let the month of November pass without celebrating the Balfour Declaration, and its profound conceptualization of the "equally indigeneous" peace formula.

11.7.2021 6:52am - (1/ ) Lord Paradox, the most vivid demonstration of the need to abandon statistical orthodoxy and resort to causal analysis (See & continues to elicit disagreements between statisticians and causal analysts:
11.7.2021 6:52am - (2/ ) The former insist on loading the problem with extraneous issues that often trouble experimental designers, but fail to address the original paradox, defined as a discrepancy between two ways of viewing "causal effect" -- a population property that is design-neutral.

11.7.2021 5:44am - I bring this up, not as indictment of ML leaders but as an enticement of their students. These "goodies" are essentially solved using causal modeling, a fact that should jolt every curious student into asking: "HOW?" and into exploring the Laws and tools of Causal Inference.

11.7.2021 4:27am - A free link to Ford's book "Architects of Intelligence" allowed me to search for "explainability", "transfer learning", "fairness" and other goodies ML folks confess are important for AGI. I'm struck by how few connect these goodies to causality.

11.7.2021 3:41am - (Replying to @hubert_misztela and @zeynep) If by "statistics and 'data-science" they mean the two disciplines as they are practiced and taught today, then hell no, "data science" isn't that kind of a thing: . But if they mean "data-enhanced-science", where calculus rules supreme, by all means.

11.4.2021 7:21am - (1/ ) It is refreshing to see epidemiologists articulating the Ten Commandments of Randomized Trials:, especially the counterfactual quote from Deuteronomy, Chapter 22 verse 8: "When you build a new house, you shall make a parapet for your roof, so that you
11.4.2021 7:21am - (2/ ) shall not cause blood to be spilled in your house, that the one who falls should fall from it." Interpreted as: "Otherwise, you will be guilty of bloodshed if anyone falls from it." A beautiful example of counterfactuals in the service of moral responsibility.

11.3.2021 10:53pm - (1/ ) There is some connection here. When people demand a "definition" of causality they really hope to find a reduction of causality in terms of other concepts, outside of causality, which are less problematic. We know that reduction to probability is impossible, so what remains?
11.3.2021 10:53pm - (2/ ) Hume suggested counterfactuals, and Lewis reduced counterfactuals to "closeness among worlds". I've found "closeness" to be ambiguous, hence problematic, and reduced counterfactuals to functional relations among variables, claiming that it is least problematic, because that is
11.3.2021 10:53pm - (3/ ) how we store knowledge about the world. So far I haven't found a reduction that is less problematic. The burden is now on those who continue to demand "define causality" to tell us what type of reduction would satisfy them.

11.3.2021 10:29pm - (Replying to @djinnome and @LeoUbb) These are beautiful slides. Thanks for reminding us of their shining humor.

11.3.2021 10:16pm - (Replying to @TacoCohen) My question: What makes "causal confusion" in imitation learning different from the "causal confusion" we always have in observational studies with unobserved confounding?

11.3.2021 12:02pm - (Replying to @JamesGrace111 @siminaboca and @cmirzayi) I do not understand who is hypnotized. Can you quote?

11.3.2021 12:00pm - (Replying to @siminaboca and @cmirzayi) I am talking about the ropes behind the data, not about tricks we use to unveil them. The ropes in themselves should have names.

11.3.2021 11:22am - (Replying to @siminaboca and @cmirzayi) Context which shows up in the data is still data-centric. The bulk of human "understanding" relies on context outside the data. Note, I haven't mentioned causality.

11.3.2021 10:30am - (Replying to @siminaboca and @cmirzayi) My theory is that most statisticians are philosophers at heart, who choose this field out of genuine interest in how humans learn from data. However, once drugged by the data-centric thinking of stat-101, they remain handcuffed for life. Few manage to escape, eg Dennis Lindley.

11.3.2021 3:37am - (Replying to @Ayaan) I am not sure my fellow Democrats get the message: Parents do care about education and, if our politruks do not lay off our schools, the 2022 election would be a Democratic party disaster.

11.3.2021 2:18am - (Replying to @qbadiamasoni) I've read Fodor decades ago, and I can't recall his theories but, based on your description, we differ. Semantically, I am not concerned about the representation of knowledge but about its SOURCE. NNs handcuffs come from their reliance on data as the sole source of knowledge.

11.3.2021 2:04am - My initial reaction to the election results in Virginia: "My Democratic Party should take a sober notice: Voters nationwide do care about education, and parents have had enough w/ attempts to push political agenda to K-12 under various disguise." Example:
11.3.2021 2:04am - We, responsible Democrats, must reign in our extremists, who may cause us a disaster in the 2022 election. I know first hand what they tried to do in Virginia's schools, and I know what they are trying to do to math education in California. Parents are no fools, they care.

11.3.2021 12:11am - A lively discussion about "Who is a Bayesian" can be found on Gelman's blog where the requirement of "subjective knowledge" is dismissed by some. This is tied, of course, to my question: "Why I am only half Bayesian":

11.2.2021 8:18am - (Replying to @TacoCohen) Thank you, this is much better. But I still believe the term "causal confusion" can be better explained in a couple of sentences. Crucially, what do we assume the expert to have that the imitator does not?

11.2.2021 6:30am - This paper has crossed my desk: which I've hoped would explain what the "causal confusion" is in "imitation learning," so it can be de-confused using DAGs. Unfortunately, the confusion is not articulated in a language I could understand. Anyone can help?

11.2.2021 4:04am - I like this metaphor: "Gothic Cathedrals for a handcuffed God". It describes so faithfully the way our @Datascience__ institutions are telling the public: "We are about to do something totally new, give us a license to ignore the handcuffs a bit longer" @JeffDean ,#AI #Bookofwhy

11.2.2021 3:01am - (Replying to @annareitman @MassadJoseph and @SandraZaloun) Fascinating findings, but genetic lineage is only one component of "indigeneity". An Irish native who serves in IDF, sings "Yerushalayim" and celebrates Hanuka in Modiin is no less indigenous to the land than @MassadJoseph for whom the pre-1900 history of the land means nothing.

11.2.2021 2:15am - This is one of those Ben-Gurionic cases where "Its not what the world thinks, but what we do that counts". If it weren't for the 400,000 naïve Shtetl Jews who took Balfour seriously (my grandpa being one of them), Israel would not have survived the 1948 war.

11.2.2021 1:54am - This is precisely what Abbas fears, and what Columbia professor @MassadJoseph cannot internalize.

11.1.2021 8:05pm - To my Zionist friends, @bariweiss , @DavidHarrisAJC , @ADL ,@Hillel_International, @AdamMilstein , @GalitPeleg , @EinatWilf , @Jerusalem_Post , We can't let Al-Jazeera re-write Jewish history. Please retweet this piece on the Balfour Declaration:

11.1.2021 6:58pm - (1/ ) Tomorrow, Nov. 2, marks 104 years to the Balfour Declaration. President Abbas already ordered flags flown to half-mast. Palestinian billionaire Munib Masri is suing the British government over its 1917 crime of issuing that Declaration What's all the commotion about?
11.1.2021 6:58pm - (2/ ) It's all about one words: "indigeneity", which we have discussed at length on this Twitter channel. Please read my commentary on how the Balfour Declaration redefined "Indigenous People", and why "equally indigenous" if my formula for peace:

11.1.2021 6:34pm - (Replying to @artistexyz @GiulioGrossi and 2 others) You use functional information to get unit-based counterfactuals.

11.1.2021 6:08pm - (Replying to @aleixrvr and @causalinf) This is not accurate, you need a collection of functions X_i = F_i(other X's) This collection defines a graph, and all counterfactuals.

11.1.2021 5:49pm - (Replying to @LekhtNaya and @GWtweets) It is meant to include Zionophobia, I presume.

11.1.2021 5:41pm - (Replying to @pfau and @gamlambard) I don't know what to make of all these new names. Hate to call them "perverse" but they are all skirting the issue of causation, hence are in effect giving up on "explainability", "trustability" "fairness," Transportability, and more. More Gothic Cathedrals to a handcuffed God.

11.1.2021 5:21pm - (Replying to @artistexyz @GiulioGrossi and 2 others) I don't believe it is possible, because the functional information is not there. See Causality p.35 Fig 1.6

11.1.2021 7:58am - Reminds me how we, as kids, were planting at least one tree per year in one or the forests of Keren Kayemet. Glad the tradition continues.

11.1.2021 7:44am - (Replying to @aleixrvr and @causalinf) We all know that you must start with something that is believed yet NOT observed (this is a theorem); shouldn't we then make that something intuitively meaningful as opposed to cryptic and incomprehensible?

11.1.2021 7:35am - (Replying to @GiulioGrossi @aleixrvr and @causalinf) What do you find to be less than trivial in the First Law:

11.1.2021 7:32am - (Replying to @SDullien) I'm familiar with this book #Bookofwhy and I join your recommendation, especially in light of our recent discussion of potential outcomes

11.1.2021 7:13am - (Replying to @IfJews) It's called the self-Goebbels effect. You repeat a lie n times so you begin to believe in it. This makes you sound much more convincing, even to yourself; now you truly believe in it, and you start chanting @IfNotNowOrg !! @IfNotNowOrg !! Totally forgetting where it started.

11.1.2021 3:05am - One important addendum to all readers of this education channel, especially newcomers. A searchable file is here available that should enable you to retrieve previous discussions on controversial issues such as, e.g., what's my take on Rubin. #EconTwitter

11.1.2021 2:38am - (1/ ) To appreciate what I mean by "assumptions whose plausibility you cannot judge" I often ask readers to examine how Imbens and Rubin (2015) define "unconfoundedness", the key concept needed for all causal inference. Quoting from their page 479, we find (fasten your seat belts):
11.1.2021 2:38am - (2/ ) First,"the conditional distribution of the outcome under the control treatment, Y i (0), given receipt of the active treatment and given covariates, is identical to its distribution conditional on receipt of the control treatment and conditional on covariates, and second,
11.1.2021 2:38am - (3/ ) that, analogously, the conditional distribution of the outcome under the active treatment, Y i (1), given receipt of the control treatment and conditional on covariates, is identical to its distribution given receipt of the active treatment and conditional on covariates." No!
11.1.2021 2:38am - (4/ ) I'm not making it up! It's the kind of judgements PO expects you to make whenever you commence a study. Readers who think I'm too harsh on PO, should also ask if it is reasonable to conceal from econ students an alternative framework that spares them this torture.#EconTwitter

11.1.2021 2:00am - (Replying to @yury_dvorkin) I'm willing to call him "Emperor of the Universe" if he would one say the words "equally indigenous".

11.1.2021 1:14am - These questions on Rubin vs. Pearl's frameworks are answered fully here:, and here: Briefly, having to rely on assumptions whose plausibility you cannot judge, or test, is more than "just notation." #EconTwitter

11.1.2021 12:36am - (Replying to @occamhanlon and @holdmytowel) The "focus on the brain (hardware)" is true in Neural Networks research, not in AI in general. My work, for example, focuses entirely on the "software", rarely asking myself what neural architecture computes one function or another.

10.31.2021 8:49am - This historical narrative didn't start from scratch; it's built on the Epilogue to Causality It is unique, in the sense that, unlike professional historians, I acted as a computer scientist and "followed the notation" in order to trace the flow of ideas.

10.31.2021 7:37am - (Replying to @NNrehman and @causalinf) Th #Bookofwhy, in addition to defining causality formally, using a logic of counterfactuals, also provides a conceptual history of how causal thinking evolved from antiquity to modern times. The latter supplements, not replaces, the former.

10.31.2021 5:27am - (Replying to @holdmytowel) It's not clear if your pessimistic characterization (reductive and harmful) applies to AI in general, or to neural network specifically. I'll disagree if the former is the case.

10.31.2021 5:18am - (1/ ) My co-author, Dana Mackenzie, gave a seminar at Stanford recently and was asked by econ. students: How come #Bookofwhy mentions LATE only once? My answer: Though useful in some applications, LATE is but one among hundreds of tools that emerge mathematically from
10.31.2021 5:18am - (2/ ) the two fundamental Laws of Causal Inference: I gave a similar assessment in Section 4 of my ET article:, and a more detailed assessment in my comments on Imbens, here: #EconTwitter, @causalinf ,

10.31.2021 4:23am - (Replying to @richard_landes) What do you mean "your/s"?

10.31.2021 4:20am - The tabula rasa paradigm of Data Science claims that, yes, we are just a point on the continuum from simple to complex neural nets, and all we need to do is wait for an Einstein to emerge from the latter: I often argue that counterfactuals are special.

10.31.2021 3:52am - (Replying to @pedrohcgs and @analisereal) I knew Seattle won't be the same once Carlos starts teaching. There should be a law against causal talking in a bar. #EconTwitter

10.30.2021 7:27pm - A good quote to retweet. It's not that people working on CI do not know it but, as strange as it sounds, those who need to work on CI do not know it. Some haven't heard it, others do not know what you we are talking about. I meet them whenever I speak to ML folks.

10.30.2021 11:28am - Looking forward to seeing you tomorrow, @BHL and, frankly, I must admit that this idea of "We did not fail in Afghanistan" has boosted my morale today, hugely, perhaps because I haven't heard it stated anywhere in the post-Afghani conversation. Thank you.

10.30.2021 9:14am - (Replying to @mikedenly @AnnualReviews and @mgfindley) Two comments. (1) It's not clear if you agree that "data fusion" unifies all variants of generalization in one framework. (2) What do you mean by: "lack of mundane realism is likely a prominent reason why political scientists have not adopted DAGs"? Do they have an option?

10.30.2021 7:27am - (Replying to @epigenci and @joshdubnau) The litmus test is simple: Does one use the First Law or not. Even Imbens, though a PO enthusiast, does not really object to this criterion. See

10.30.2021 6:46am - I can't explain the confusion in "Strategic Management". Here is a 2009 paper in the same journal which makes perfect sense, and should have introduced modern CI thinking to its readership:

10.30.2021 4:00am - The Renowned Bernard-Henri Levy in Rare Los Angeles Appearance:
10.30.2021 4:00am - Speaking of "active learning" and the idea that relevant data need be actively "sought after", BHL has a new book "The Will to See", and I will have the honor of introducing him on Sunday, 4 pm, at the Wilshire Boulevard Temple, in Los Angeles. BHL is my "Global Poet Laureate."

10.30.2021 3:25am - Predatory publishers' latest scam: bootlegged and rebranded papers
10.30.2021 3:25am - A phenomena to be watched, which also explains why we are getting so many offers from shady journals to re-publish old articles "at no cost."

10.30.2021 1:45am - (Replying to @Rodrigo00852376) On transportability: (1) (2) On cycles, Causality ch7 and ch11

10.29.2021 6:45am - (Replying to @AnothrAtheistJu) No idea. I was focusing on the normalization of the subnormal.

10.29.2021 6:41am - Some runners make my heart jump - must share.

10.29.2021 4:22am - (Replying to @IfJews) I dare consider @PeterBeinart an X-Jew. That would entitle him to repent and re-join the Jewish tent w/o a 2nd circumcision, perhaps even w/o a Rabbinical conversion, but would still mark him X whenever he says "As a Jew...", which has caused enough damage to our people already.

10.29.2021 3:40am - (Replying to @_goose_god and @ThManfredi) Not necessariamente, just for fun and out of curiosity I would recommend Primer

10.29.2021 12:37am - FYI, after studying the issue with colleagues at UCLA and USC, I have signed this open letter concerning California Math Education Curriculum: Please consider signing it; at stake is the future of science education in California

10.28.2021 11:05pm - (Replying to @greg_m_priest @epistwatchtower and 2 others) My tweet was a reaction to attempts to justify the confusion in Management Science by appealing to "pluralism". But it was baked by decades of debates with economists on the cognitive formidability of the PO framework. e.g.,

10.28.2021 10:41pm - (Replying to @xenophilictb @DKedmey and @EinatWilf) Zionophobia is a moral deformity which, unlike its sisters: Islamophobia, homophobia, antisemitism, etc., cannot be denied by the afflicted. And that's where its diagnostic power lies -- a perfect detector.

10.28.2021 10:20pm - (Replying to @greg_m_priest @epistwatchtower and 2 others) Good try, but note that, as a devout "monist", I fear not the possibility of pluralistic approaches to causation. On the contrary, I am inviting Nancy Cartwright to provide ONE example where it's needed: (ie, where SEM is insufficient). Can you?

10.28.2021 3:06pm - (Replying to @xenophilictb @DKedmey and @EinatWilf) Finally, a Zionophobic specimen, deciding conveniently and authoritatively who is indigenous and who is not, not noticing how racist it is. Zionophobes are experts in hiding their racism, but their refusal to say "equally legitimate" betrays their character.

10.28.2021 6:27am - (Replying to @PHuenermund and @eliasbareinboim) This is a good paper. Keep the rejecting reviewers comments -- priceless documents for future historians of 21st century science. I am keeping such a collection, but I have not found the historian yet.

10.28.2021 2:39am - (Replying to @jesper_wulff and @JohnAntonakis) Thanks for catching. It should read: Should not be a matter of opinion or choice.

10.28.2021 2:35am - People celebrate when celebrities' defy racist movements. I bemoan the need to celebrate such defiance.

10.28.2021 12:45am - (1/ ) I managed to jump over their paywall and find myself asking: Is this the state of the art in "Strategic Management"? I think you should submit a tutorial article saying: "Comrades, your 'heuristic' is now a science!". Interesting, they mention #Bookofwhy in the Conclusion,
10.28.2021 12:45am - (2/ ) which I theorize to be a lips service paid to a well-read reviewer who commented: "Comrades, your 'heuristic" is now a science!". Anyone among our readers in "Management Science"? Are things really that foggy there?

10.27.2021 11:27pm - (Replying to @xenophilictb @DKedmey and @EinatWilf) Racist is he who replaces "equally legitimate" with "not even the size of a postage stamp" -- an ideology preached, not by "some," but every teacher, cleric, author, or public figure, with no exception, backed by violent acts whenever possible. Recall, I've seen 1936, 1948, etc.

10.27.2021 8:49pm - (Replying to @xenophilictb @DKedmey and @EinatWilf) People have the right to resist things when they aim to gain life, liberty and dignity while respecting others' right to life, liberty and dignity; not if they vow to ultimately destroy others' life, liberty and dignity, and publicly state so.

10.27.2021 9:14am - Grandchildren of great leaders tend to dishonor their family's legacy by falling for BDS propaganda. Same happened to Arun Gandhi in 2008, who wrote that "Israel and the Jewish people create a Culture of Violence" that "is eventually going to destroy humanity." @washingtonpost

10.27.2021 6:47am - (Replying to @DJMinded @hillel and @ADL) Let me know when you are in the mood to condemn other violations of human rights, like a 73 yr denial of legitimacy coupled with 73 yrs of attempted annihilations. It's all a question of mood.

10.27.2021 5:32am - (Replying to @IzaTabaro) Agree with every word except ONE. Why are you calling them "antizionist ideologues" instead of Zionophobes, as if their ideology has no moral and mental deformity to it? They take the title "antizionist" as a sign of legitimization, if not a badge of honor.

10.27.2021 5:17am - Statisticians among our readers would recognize the name of Joyce Fienberg among the 11 victims, wife of professor Steve Fienberg, a leading statistician at CMU, (Co-author of "Discrete Multivariate Analysis, 1975, and "Statistics and the Law", 1980). YZB

10.27.2021 4:37am - (1/ ) I wish I could join the jubilee and endorse this new partnership. However, I must stick to principles and apply my unfailing litmus test: Does the word "Zionism" or "Zionophobia" appear in the public announcement? If not, the partnership will end up, like many before it,
10.27.2021 4:37am - (2/ ) in the dustbin of failed Jewish leadership. @Hillel International has announced several such partnerships (eg. with @ADL ) with predictable consequences: No Zionism = No impact. Students are longing for a forthright leadership. Would you consider revising the announcement?

10.26.2021 10:59pm - Well put, @LekhtNaya , our leaders have chosen to fight outdated religious wars, with outdated weapons, the weapons they were trained on in college. None has gather

10.26.2021 2:49am - Congratulations go to Professor Chad Hazlett of UCLA who got his tenure last month. Chad is teaching one of the two core classes in causal inference offered by UCLA and should be warmly congratulated as a key mover in the CI family.

10.26.2021 12:41am - I've always said: Gems can be faked, but not Gem Hunting.

10.25.2021 3:25pm - I'm glad I have the chance to RT @analisereal seminar on Sensitivity Analysis, and I wish I could attend the talk tomorrow.

10.25.2021 1:53am - (Replying to @amilkant) The relationships between causal inference and statistical decision theory are discussed in Chapter 4 of Causality

10.24.2021 5:00pm - A very valuable tool for both researchers and educators. My suggestions: (1) Start every task with a triplet: (a) model, (b) data available, (c) Answer needed. (2) Use conventional structural graph for every task (no shortcuts). This would make it useful for casual readers.

10.24.2021 8:54am - (Replying to @HumzahAlkindi) From what I was able to grasp thus far, our aims are different. She is trying to model a ill-understood reality and I am trying to model (and emulate) people's understanding of reality.

10.24.2021 8:49am - It is so true and, unlike T. Kuhn, today's historians of science can interview in real time those who are currently objecting to the #Bookofwhy, and dissect their objections piece by piece for a better understanding of those forces that hold back science from advancing faster.

10.24.2021 4:29am - (Replying to @attilacsordas) And this is the most optimistic thing I heard about causal inference; "to date" promises another breakthrough around the corner -- breathlessly awaiting.

10.24.2021 4:21am - (Replying to @daniellenewnham) I would be delighted, though I don't belong to the startup club.

10.24.2021 4:10am - One of the sweetest things I heard people say about the #Bookofwhy -- thanks. I wish more readers would view it as a reflection on scientific thinking in general, rather than a statement about artificial intelligence or machine learning.

10.23.2021 11:12am - I envy my sister, who can ride a bike to this spot.

10.23.2021 9:11am - Saddened to acknowledge (belatedly) the death of a friend, a colleague, a countryman and a teacher, Eran Zaidel, who passed away July 1st: Here is his latest video on consciousness: Born in Kibbutz Yagur, 1944, Yehi Zichro Baruch

10.23.2021 7:03am - (Replying to @CAUSALab and @_MiguelHernan) This sounds like a useful exercise in Data Fusion,, am I right? If so, why appeal to cryptic "exchangeability" conditions when we have meaningful conditions based on things that we know or believe?.

10.23.2021 6:53am - Father of man charged with Amess murder defended attacks on Israel via @jewishchron
10.23.2021 6:53am - As a direct victim of "grievance culture" I find it painful to comment on this discovery about Amess murder. What a monumental cesspool of hatred was brewing there, in Mogadishu, in the name of religion. For how long will it continue to contaminate our streets?

10.23.2021 5:36am - (Replying to @LentoBio) Viva Jujuy! Viva la Puna! Viva mi amada!

10.23.2021 1:43am - (Replying to @daniellenewnham and @JewishJournal) Oh, yes, this interview is mine. And the picture is of my grandfather and his family, about 10 years after arriving to Israel to re-build the Biblical town of Bnei-Brak. The little guy is my uncle, the first boy born in Bnei-Brak, who taught me English & about 200 songs.

10.23.2021 1:14am - Can't we be more honest and direct? Sunrise: YOUR RESPONSE DOES NOT say that singling out Zionist organizations for removal despite holding similar views is Zionophobic, hence unacceptable. WHY NOT? Your tweet skirts the issue:

10.23.2021 12:48am - This picture describes my mood upon arriving at California, 1966, happy to see everyone understanding the Spanish songs I knew, and acting like the Hippie I wanted to be.

10.22.2021 7:35am - (Replying to @ted43062045 @melbatutor and 3 others) No, algebra vs. graph is another issue. The problem with the PO strand that economists are using is that it forces them to posit assumptions that they themselves cannot comprehend or to judge for plausibility. See

10.22.2021 5:10am - (Replying to @manuelbaltieri) Never expected to find my name on a blanket that belonged to Markov, nor to find so many faces to one wrinkled blanket.

10.22.2021 3:36am - My heart goes out to them -- two yesterday's Jews lured by the bait of social acceptance. Please, if you have the contacts, allow me to meet them for 10 minutes; they deserve an understanding of when/how you earn genuine respect from peers that matter.

10.22.2021 1:40am - (Replying to @artistexyz @ted43062045 and 4 others) The strand of PO currently thriving in economics has two deficiencies: (1) No graphs to facilitate inferences (via d-separation) and (2) No structure to illuminate assumptions. It is surviving, like the monarchy, only because it is erroneously supposed to do no harm.

10.22.2021 1:40am - (Replying to @IfJews @IfNotNowOrg and 2 others) Notice how predictably repetitive are the priests of Apartheids when you call them antisemitic. Try "Zionophobes" and watch how creatively they react to a charge they can't deny.

10.21.2021 1:04pm - Interesting observation. In the slide at 1:31:25 into the video, I find Eq.(6), where Heckman comes close to the First Law (structural definition of counterfactual Y(x)) but, unfortunately, misses it, by using "statistical conditioning" bar "|" in a deterministic setting,

10.21.2021 12:31pm - Fascinating story!

10.21.2021 5:35am - (Replying to @EnLiEretzAheret @laltp and 2 others) But observe how careful these racists are to signal how they love Jews, and hate only Zionists. If we, Jews, do not make sure that this kind of hate in itself is as socially unaccepted as Islamophobia and homophobia, the will continue to play the word game of "We love Jews".

10.21.2021 5:21am - Congratulations, @david_lagnado on your new book, which I've endorsed thus: "a flowing, compelling and lucid account of how computational models of mental states are changing the face of cognitive psychology and decision-making sciences. A must for students of mind, (and more..)

10.21.2021 12:12am - I will support this campaign for "slander equality" -- every slander against Israel should be treated the same as slanders against Islam, Latinos or Asians, but the fight should move from the legal to the moral arena, using the causes of "Equity, Diversity and Inclusion".

10.20.2021 11:36pm - Acknowledging is not enough. Please write to @ADL and @AJCGlobal to invest 1% of their PR budgets into making Zionophobia the ugliest word in town. They can do it, and it's not too late! @DavidHarrisAJC

10.20.2021 11:27pm - It was a super pleasure for me to meet Branden and Valeria at the @LAPressClub Awards dinner and to learn the @ThePearlPost , of the Daniel Pearl Magnet High School was placed second in Best High School Newspaper category. You have to visit that high school to understand why.

10.20.2021 7:28am - (Replying to @lewbel) Why? Have you tried it? Or, better put, do you personally agree that the First Law is the way to start teaching economists the art of Causal Inference?

10.20.2021 6:02am - (Replying to @DonskerClass) No need to wait for the results. Once you have the First Law, the rest is sheer mathematics. Students can derive everything by themselves, including the Second Law. Theoretically, you can now can hand over the class to a TA. In theory, of course.

10.20.2021 5:47am - Explicit recruitments of this kind are sending signals to universities that the time may be right to convert some of the cathedrals they built for Data Science into modest Causal Science Centers.

10.19.2021 11:43pm - Thanks @bariweiss for elevating my remarks from the valley of the Biltmore Hotel to the height of

10.19.2021 5:17am - Any econ student who sees where POs come from will be immunized for life against the siren songs of model-free methodologies, and will be able to manage correctly even the assumptions that underlie "quasi experimental" studies.

10.19.2021 2:26pm - Here is my (4-min) presentation speech upon giving @bariweiss the Daniel Pearl Award:

10.19.2021 3:53am - (Replying to @gilled34) I have defined precisely what kind of 1-state Israelis are fiercely resisting. The current situation is not that kind, because defense and borders are still in the hands of those who believe in co-existence.

10.19.2021 3:27am - (Replying to @gilled34) I have defined precisely what kind of 1-state Israelis are fiercely resisting. The current situation is not that kind, because defense and borders are still in the hands of those who believe in co-existence.

10.19.2021 1:27am - (Replying to @mara74) My reference to "page 479" was to Imbens and Rubin book of 2015, the Blue pill.

10.19.2021 1:24am - (1/ ) I'm compelled to retweet this thread because so often I see well-intentioned people assuming that a "happy ever after" 1-state solution is inevitable because Palestinian's rejection of Jewish self determination (in ANY borders) is so total and deeply entrenched that any other
10.19.2021 1:24am - (2/ ) arrangement amounts to endless blood shed. As one born in Israel and tuned daily to the country's pulse, let me mention another factor which is often ignored in conversations about the 1-state fantasy. Israelis resistance to a 1-state solution is at least as total and deeply
10.19.2021 1:24am - (3/ ) entrenched as the Palestinians' rejection of Zionism. For Israelis, the idea of relinquishing Statehood (i.e., right to self defense and border control) towers as a collective suicide, against which people are willing to fight to the last soul (barring a few ultra orthodox
10.19.2021 1:24am - (4/ ) enclaves). It is often equated with a betrayal of 2 thousands years of yearning for normalcy and dignity and, therefore, no government would last a day even discussing such an option. This factor must be taken into account when considering the 1-state vs. 2-state options.
10.19.2021 1:24am - (5/ ) The former would not be as bloodless as some people fantasize; it will be genocidal. The current state of painful but manageable violence is probably the least genocidal, and will persist therefore until some "end of claims" ideology emerges from education for co-existence.

10.18.2021 11:54pm - (Replying to @hallam @titudeadjust and @MaxBoot) Two false premises do not make 1-state feasible. (1) Likud will accept 2-state if based on "end of all claims" (recall Olmert was Likud). (2) Israelis resistance of 1-state (left & right) is fiercer (to the death) than Hamas resisting 2-state. Conclusion: 2-state + End of Claims

10.18.2021 11:08pm - (1/ ) Disagree. The so called "informal" definition is not a "definition" of unconfoundedness, but a celebration of what we could do had we achieved it by some set of covariates. It's like saying: "Unconfoundedness is that which makes causal effects estimable by regression."
10.18.2021 11:08pm - (2/ ) Speaking 6 years past the publication of Imben & Rubin's book, the fact remains that there is no PO "definition" of unconfounding simpler than {Y(1),Y(0)}||X|Z which is as cryptic as what it's supposed to decrypt, because scientific knowledge is not stored in such a language.

10.18.2021 6:06pm - (Replying to @titudeadjust and @MaxBoot) Glad @MaxBoot alerts readers of the Post to BDS hypocrisy . My 2 comments: (1) We have a name to call such hypocrites -- Zionophobes -- if one hesitates to use "antisemitism" and (2) No need to signal virtues ("I too criticize Israel") before using it.

10.18.2021 4:39am - The Jewish Calendar marks 26 years to the day Yitahak Rabin was assassinated. His life story, below, is the story of Israel.

10.18.2021 12:53am - (Replying to @rickwsint and @thebadstats) BW question need not wait for a "complete model" like you described, which includes configuring and naming relevant variables. Their question can completely be answered using SCM.

10.17.2021 11:30pm - (Replying to @rickwsint and @thebadstats) Not so "speculative" because we have an existence proof named "humans". If I did not believe that a complete formalization is around the corner, primarily by virtue of what has been achieved already, I would go fly a kite instead of shouting: We are almost there!

10.17.2021 11:23pm - Your (and my) political family, the Zionist left, is in the process of splitting irreversibly into two non-communicating factions: The "pleasing" ones and the "self-respecting" ones. Join the latter, save/shun the former.

10.17.2021 11:12pm - (Replying to @bbohb2) Palestinians do not understand the notion of "historical connection" because they do not have it. (See books by Rashid Challidi). Is your professor Palestinian or British? Why not reveal her name?. Zionophobes need to be exposed.

10.17.2021 11:03pm - (Replying to @bbohb2) She underestimate the strength of the Jewish historical connection to Israel by comparing it to "all people". No other people has trust-deeds enshrined in collective memories, language, holidays, heroes, customs, poetry etc. connected to one landscape.

10.17.2021 10:51pm - (Replying to @rickwsint and @thebadstats) I do claim that all aspects of causal reasoning by humans are formalizable and will eventually be captured by AI programs.

10.17.2021 10:43pm - (Replying to @bbohb2) Guilt over colonialism is undoubtedly the main factor explaining UK obsession with Zionophobia. Guilt over UK's part in the Holocaust - actively preventing European Jewry from escaping to Palestine (1936-40) - is another. I'm sure professional psychiatrists will find more.

10.17.2021 10:05pm - Beautifully put!

10.17.2021 10:02pm - "Normal people" boycott a country that was denied "normalcy" from the day she was born and, yet, has contributed to the world more than all her "normal boycotters" put together.

10.17.2021 8:15pm - Prior to digital mob-rule, being normal was a virtue. Here is a proof. In 2007, I wrote "Why do homo sapiens recognize Israel's right to exist? " Nowadays, "normal" means "I have no opinion of my own" or "I'm scared to be bullied by the herd".

10.17.2021 6:57pm - (Replying to @Raamana_) Sarcastic? Why? I just quoted Imbens & Rubin definition of "unconfoundedness" verbatim. Recall, "unconfoundedness" is not something you get from data. You need to discern it from judgement, based on scientific knowledge. It's hilarious to see what I&R wish us to judge.

10.17.2021 6:28pm - It was a great honor for me, last night, to present the Daniel Pearl Award to Rabbi @bariweiss . Why "Rabbi"? Because "whoever learns anything from his fellow man, even an iota, calls him 'Rabbi'". I've learned a lot, partly about Hillel and Shammai, by mostly about "Courage".

10.17.2021 6:19pm - (Replying to @rickwsint and @thebadstats) These days, philosophers of science are listening to what AI researchers have to say about causation, because the latter are accounting for the problem of mental representation which philosophers, by enlarge, have neglected. See eg,

10.17.2021 1:36pm - (1/ ) Readers ask, justifiably, how can PO folks claim to be doing science when their assumptions are so convoluted? The "natural experiment" guys do have an escape route: "We limit ourselves to simple patterns, like IV, where assumptions can be comprehended by ordinary mortals."
10.17.21 1:36pm - (2/ ) But is this true? This post shows that, even in simple IV models, if you insist on PO notation and continue to ban graphs you won't go very far: Conclusion: Foggy eyeglasses may be enough for some folks, but science needs telescopes and microscopes.

10.17.2021 1:09pm - (1/ ) This clip sounds circular but, with some stretch, it can be repaired, to mean: {correlations + causal assumptions} ==> Causal-Conclusions. which is not circular. Moreover, the causal revolution adds something to this implication that did not exist before 1993: For ANY pair on
10.17.21 1:09pm - on the left, we now have a procedure that tells us which conclusions are implied and which are not implied. This makes the clip not only non-circular but also definitive. For fun, see " A Dialogue with a Hostile Examiner":

10.17.2021 8:14am - (Replying to @ajaydiv @KordoniN and @vardi) Can you remind us how Angrist define "confounding" or "ignorability"?

10.17.2021 7:15AM - (Replying to @PieterZijlstra2) You are not alone. I do not know a single person who can understand it. Yet I know many who swear by it. One of the authors even says he has no problem thinking that way, because thinking is subjective, each person with his/her own way of thinking.

10.17.2021 6:12am - Glad readers caught my mistake; the definition I quoted was from the "Blue Pill" (Imbens and Rubin 2015, p. 479. ). The Green Pill is known as the backdoor criterion, no swet.

10.17.2021 5:10am - (1/ ) I would check first how each defines "unconfoundedness", the key concept needed for causal inference. Here is the Green pill: " the conditional distribution of the outcome under the control treatment, Y i (0), given receipt of the active treatment and given covariates, is
10.17.2021 5:10am - (2/ ) identical to its distribution conditional on receipt of the control treatment and conditional on covariates, and second, that, analogously, the conditional distribution of the outcome under the active treatment, Y i (1), given receipt of the control treatment and conditional
10.17.2021 5:10am - (3/ ) on covariates, is identical to its distribution given receipt of the active treatment and conditional on covariates." I am not making it up, its on page 479.

10.16.2021 6:11am - Patient: I am afraid they are! But, no fear, they seem conditionally ignorable.

10.16.2021 3:53am - Search Engine Land: Peter Norvig is stepping back from Google after 20 years. What does it mean to Causal Inference? A lot. Because a "distinguished Education Fellow" to "explain key concepts of AI" can move HAI a long way toward the human side of AI.

10.16.2021 3:28am - (Replying to @janmendling) When the Jews migrated from Germany to Poland, 12th Century, they got to keep their German names. The latter were given by profession, so, I guess one of my ancestors, named Perl, was a dealer, or a polisher of precious stones. Same as "Perlstein", "Perlmutter" & "Goldstein".

10.16.2021 12:59am - The handsome fellow standing up is my father, getting prepared to go back to Poland to search for an eligible bride, my mother, who at that very juncture of history was looking for creative ways of escaping from inferno Europe. She made it, so did he, and so did I.

10.15.2021 10:48pm - (Replying to @lihua_lei_stat) I couldn't get the research question from your abstract. I'm missing the divine triangle: {Question + Assumptions + data} ---> Answer

10.15.2021 6:56am - For readers who asked me where I got the name "Judea". As you can see on these ancient coins, it was quite common name in 120 BC, during the Maccabees revolt and after; it wasn't my mother's imagination.

10.15.2021 6:21am - Why are we surprised? If Rutgers University is forced to apologize to Palestinian students for condemning anti-Semitism, then anti-Semitism becomes "controversial", and "opposing views" must be heard in class. It takes leadership to say: "Some coins ain't got two sides!"

10.14.2021 8:19am - (1/ ) The Forward had a story about UCLA graffiti: Apropos, I retweet my invitation to listen to powerful webinar by Dr. Einat Wilf, recorded a few days ago: Please retweet to your circle of friends because, as you can see, her
10.14.2021 8:19am - (2/ ) experience, insights and mode of presentation are not common among Jewish academics. None of that self doubt and rush to please. It's organically grown Zionism that knows and demonstrates its noble cause.

10.14.2021 6:59am - (Replying to @Raamana_ @artistexyz and 2 others) I have tons of discussions with Imbens on my blog but, as Thomas Kuhn said (paraphrased): Never blame scientists for not jettisoning the paradigm that has been the basis of their life's work; their firm belief that it can solve all their problems makes normal science possible.

10.14.2021 6:44am - (Replying to @RohanAlexander @aecoppock and 2 others) Glad someone is teaching "History of Statis". The available texts (eg Stigler, Salzburg) pay less than lips service to causality and then students complain: Where are our heroes?

10.14.2021 6:30am - (Replying to @artistexyz @Raamana_ and 2 others) Determinism was never strawman for banning the telescope, because the starting point of PO is also deterministic: the outcome Y for an individual u under treatment X=x has definite value. The strawman was the devil behind the telescope which makes the sky so much more visible.

10.14.2021 3:11am - (Replying to @DrMichaelOren) @DrMichaelOren , Twitter may not be the official way of endorsing your candidacy to head of the Jewish Agency but we, here, in the trenches of defending Israel and fighting BDS on campus do not always have time to abide by protocols. Please add my strong endorsement to your pile.

10.14.2021 12:36am - (Replying to @KordingLab @Raamana_ and @andpru) This is fine, "unconfoundedness" is a popular word and much of the book depends on it. But take a look at the cryptic "definition" they provide for "unconfoundedness" and tell me if the 2nd Law does not deserve the title "telescope" for sparing economists this acrobatics.

10.13.2021 11:49pm - It was great to read the list and find out that, despite the NYT, we have so many friends around the world. But my greatest thrill was to find a few familiar faces there, among them the courageous Los Angelino @Eve_Barlow . Congratulations!

10.13.2021 11:27pm - (Replying to @Eve_Barlow) Beg to differ. I may expect Sally Rooney to be antisemitic (eg, alcoholic parents, fanatic teachers, bad neighborhood etc.) but not Zionophobic; authors, writers, and other thinking organisms are expected to do some thinking.

10.13.2021 11:13pm - (Replying to @OChatain @v_m_bennett and @analisereal) Strange, I got into DAGs for a totally different reason: It was the only way to express what reality could be like. It's not a coincidence that our paths converged.

10.13.2021 10:10pm - (Replying to @KordingLab @Raamana_ and @andpru) Thanks for counting. Now we understand what world PO folks live it when they say: "[in the] examples arising in SOCIAL and MEDICAL science, we have not found this [graphical] approach to aid drawing of causal inferences" (Im&Ru, p.22). It is the world of LOTTERIES.

10.13.2021 7:03pm - (Replying to @KordingLab @Raamana_ and @andpru) If they were to live in this part of the world, they would never resort to the concept of "ignorability", but they do, repeatedly (I tried to count how many time it appears in Imben/Rubin book, but Kindel did not let me),

10.13.2021 6:23pm - (Replying to @KordingLab @Raamana_ and @andpru) If everything is connected to everything else you cannot get any causal conclusions. Even in elementary IV, you need exogeneity (of Z) and exclusion. Each implies some missing arrows. In RDD we have equality constraints. The need increases when we deviate from ideal models.

10.13.2021 6:02pm - (Replying to @v_m_bennett) The crash course you linked to is receiving universal welcome by many instructors. But I have not met anyone yet, be it reviewer or researcher, who could use counterfactuals to explain good control and bad controls. I hope my reviewers won't do it to me.

10.13.2021 5:54pm - (Replying to @KordingLab @Raamana_ and @andpru) The situation may be involve social and psychological forces which I have not listed, but is there anything in what I wrote that you find less than accurate or factual?

10.13.2021 1:41pm - (Replying to @guilhermejd1) Agree. Both "monotonicity" and "no defiers" are restrictions at the unit level, ie, they cannot be tested in experimental studies (eg, RCT). Therefore, LATE is a Rung 3 concept, while ACE is Rung 2.

10.13.2021 7:30am - (Replying to @causalinf) Don't retract yet. Your comments gives me an opportunity to show readers how this idea of "surrogate experiment" takes shape and is solved completely in the non-parametric setup. We called it zID:

10.13.2021 6:21am - My guardian angel whispered: Imagine that you have just one last tweet to post, what would it be? Normally, I would hesitate for a minute or two, not today. Please drop whatever you are doing and let Einat Wilf transform your mind, as she did to mine.

10.13.2021 6:01am - (Replying to @TheScottLove @ScottGottliebMD and 6 others) It would be a thrill to see such a center taking shape in the next 35 years of my life but, then, I'll have to quit Twitter, and who is going to philosophize with readers on whether data science is still a science w/o the science?

10.13.2021 2:15am - (Replying to @godmak) Authors who seek to comply with the guidelines of the BDS movement before checking the racist characters of that movements, not caring how they are being used to promote the racist mission of that movement, do not deserve readers' time.

10.13.2021 2:01am - (1/ ) Now, for the sake of readers who joined us only lately, let me reiterate what the 1st and 2nd Laws of Causal Inference are. The first is described and motivated in this blog Here is one of its flowers: The 2nd is d-separation.
10.13.2021 2:01am - (2/ ) Together, these two Laws imply everything that I have seen developed in causal inference, including the seven tools, and including of course everything developed by those who feared the two Laws, or banned them from classroom. That's the reason I regard them as a telescope.

10.12.2021 11:15pm - The "world of pharma"!! Sounds awful scary! But these are the people who really need to hear how observational and experimental data should be combined. These are also the people who should establish a Center for Causal Inference in some forward-minded university. Where are they?

10.12.2021 11:00pm - The telescope? Don't tell me you couldn't guess. The telescope is the 1st and 2nd Laws of Causal Inference, and all the tools that logically follow from them, of which IV is but one foggy eyeglass.

10.12.2021 10:38pm - How much insight into human nature can a book evoke, when written by an author who refuses to see the human face of 7 million Hebrew speaking human beings in Israel, their aspirations, and their longing for a day of normalcy and co-existence.

10.12.2021 10:19pm - (Replying to @IfJews) It is refreshing to hear someone finally call BDS "racist", rather than "antisemitic," which they love so much to hear, dismiss and prove innocent of. Same goes to @IfNotNowOrg and any other organization that feels it's cool & consequence-free to call Israel "apartheid". Finally!

10.12.2021 5:49pm - (Replying to @DJMinded) That is precisely my point. If we oppose #apartheid, we should not consume the BS that #BDS cronies feed us, we should call on them to use other populist slogans to drive their slander machine.

10.12.2021 5:08pm - As I've suspected, the 2021 Nobel Prize in economics will increase public interest in causal inference methodology. People already begin to ask: "If this is what you can see with eyeglasses, imagine what you can see with a telescope."

10.12.2021 4:01pm - Is anyone eager to read a book by an author who falls for BDS propaganda and rushes to parrot it? What kind of wisdom can such a book carry?

10.12.2021 3:07pm - The #Bookofwhy traces the IV idea to John Snow (1813-1858) and his investigation of the Cholera epidemic of 1853-54. This does not diminish the importance of Angrist-Imbens contributions, but highlights the universality of the idea, as seen through its graphical structure.

10.12.2021 8:21am - Op-Ed: Did the vandals who scrawled an anti-Israel slogan in UCLA's Bunche Hall know the history?

10.12.2021 1:34am - (Replying to @grbradsk) I don't believe representing "self" is feasible without the ability to represent one's environment. True, now that we have the computational facility to represent the environment (eg causal model) we would be able to representing "self", and achieve consciousness, free will,etc.

10.12.2021 1:17am - (Replying to @_charadams_) Why "tickled"? Did my comments contradict yours? Care to compare?

10.12.2021 1:08am - I'm glad you are asking this question and not a NYT reporter, which would have forced me to state in public why I value A&I contributions but oppose their methodology, as explained here; that would have spoiled the celebration they deserve.

10.12.2021 12:52am - (Replying to @matloff and @AaronMeininger) Oh, now I got it. By "our colleagues" I meant "researchers who share our interest in CI". I should apologize to Card, but I'm he has much on his mind to care about what I write.

10.11.2021 9:56pm - Viewed from different perspective, this prize brings a great honor to Israel, which has just increased the number of its Nobel Laurates in economics from two (Robert Aumann and Daniel Kahneman) to three (Angrist is Israeli-American). Not bad for a country under perpetual siege.

10.11.2021 7:20pm - As expected, yesterday's Nobel Prize announcement has evoked healthy curiosity among readers as to why I cannot endorse the methodology of Imbens and Angrist all the way. Part of the answer is in this blog The other part lies in the First Law

10.11.2021 7:04pm - (Replying to @AaronMeininger) I don't get it, sorry.

10.11.2021 6:55pm - (Replying to @KordingLab and @metrics52) The method is correct if the assumptions behind the IV formula are correct. The methodology, however, is wrong in that it excludes tools that would allow researchers to judge if those assumptions hold, if they have testable implications, and if they can be modified when needed.

10.11.2021 5:12pm - (Replying to @KordingLab) This is precisely the point. Once you have the Foundations right, the backdoor criteria and the IV formula are transparent special cases, but focusing on the later w/o the foundations leaves out much of the power.

10.11.2021 5:03pm - (Replying to @EdKwangl @constant_hevia and 4 others) Happy as described in my tweet This prize is likely to entice economists to study more general and more transparent ways of managing cause-effect relationships.

10.11.2021 9:14am - (1/ ) Congratulations are due to our colleagues Joshua Angrist and Guido Imbens on receiving the 2020 Nobel Prize in Economics, thus drawing the limelight to the science of causal inference and to the new methodology which they have helped develop. It is no secret that I have been
10.11.2021 9:14am - (2/ ) (and still am) a staunch opponent of Angrist and Imben's methodology, primarily for its overlooking the two Fundamental Laws of causal inference. Nevertheless, their method has caused a shakeup in economics and a greater appreciation for the general problem of drawing causal
10.11.2021 9:14am - (3/ ) conclusions from non-experimental data. This most recent award will increase researchers interest in the science of causal inference and, naturally, would entice economists to study more general and transparent ways of managing cause-effect relationships.

10.11.2021 8:52pm - (Replying to @gaudi_sachit) No problem. I am sure Angrist and Imbens will nominate me for next year award.

10.10.2021 8:12pm - You are right, Danny's jokes gives his birthdays the flavor they deserve. So, here is another one: Friends once asked him: Hey Danny, do you believe in the after life? "I'm not sure", he said, "I have more questions than answers,... but I sure hope Gabriel likes my music".

10.10.2021 7:52pm - (Replying to @desertcran @tadesouaiaia and 4 others) I do not know a single statistician who remains in disagreement with me, once we sit down and explicate the disagreements. That is why I beg my angry critiques to be specific, can we? Your comment on correlation and causation, for example, is specific, but in error.

10.10.2021 5:42pm - (Replying to @tadesouaiaia @desertcran and 4 others) No need to feel low, some of the best statisticians have converted. I always admire Lindley who had the guts to admit it (See Causality ch. 11) But I am curious, what specifically you disagree with, and what kind of "description" would satisfy: "we hardly describe what it is."

10.10.2021 5:15pm - Where is the British Emil Zola to write a 2021 "J'accuse!" - "I accuse the BBC, its editors, its writers, producers etc. of conspiracy to deceive the British public..." But who will publish it? The Guardian? The Independence?. This is not Paris of 1898, it's Corbyn's London.

10.10.2021 4:23pm - (Replying to @tadesouaiaia @desertcran and 4 others) No reader of #Bookofwhy would ever say: "causation is not defined". Nor will a reader say: "Some associations are meaningful wrt to manipulation, some are not." because the book tells us precisely which are and which are not.

10.10.2021 11:02am - (Replying to @FriedrichHayek) Before we explain Copernicus we need to understand how a 10 yr old explains why he/she was late to school. A theory that claims to do the former and can't do the latter may be right, like many other nice theories, but will not get my passion. Please forgive.

10.10.2021 10:51am - (Replying to @DebyNavarroR @droitstein and @Serherskovits) Nothing would please me more than knowing that the legacy of Danny inspires young Jewish students in far away Santiago, and empowers them with resilience and pride in who they are and what they have to contribute to this world.

10.10.2021 9:20am - Today, October 10, is Danny's birthday. His last words "I am Jewish", have become a banner of peoplehood and identity to many, and are enshrined in this Book: which I consider THE BEST for understanding how Jews define themselves in the 21st Century.

10.10.2021 9:00am - (1/ ) Germany is "committed to Israel's security", so we are assured, again and again. It's nice, but it doesn't occur to Germany (and EU) that Israelis take such assurances as a bleak prophecy to be needing this security for the rest of time. This, instead of concrete actions
10.10.2021 9:00am - (2/ ) to eliminate threats of annihilation from the region, invest in education for acceptance and co-existence, or, at the very least, defund current educational programs that promote annihilation. Germany and EU can do more for peace.

10.10.2021 5:36am - (Replying to @OriKatz3) Agree. And that is why we need a crisp criterion, taking into account all these considerations about proxies, correlations and prices (all are derivable from the model) to decide: When does vector C1 deserve the same price as vector C2?

10.10.2021 5:24am - (1/ ) Today, on his birthday, we celebrate Daniel Pearl's life. If he grew to become a legend, it is not through the heroism of his death but through the principles by which he lived his life, as they shine in his writings and thousands of testimonials, his honesty, care and humor,
10.10.2021 5:24am - (2/2) the respect that he earned on both sides of the East-West divide and, finally, through the natural, non-heroic yet unyielding pride in his Jewish identity. He really loved this planet.

10.10.2021 5:07am - (Replying to @DebyNavarroR @droitstein and @Serherskovits) Se necesita un educador ilustrado para apreciar su poder y relevancia.

10.10.2021 4:27am - (Replying to @OriKatz3) My question assumes that the regulator has a list of "protected variables" scattered around the other attributes of the individual and, still, we need to come up with a criterion: When do two individuals deserve the same price?

10.10.2021 3:42am - Thanks for the quotes.Lately, it has occurred to me that there are hundreds of words in common discourse which reflect comparison to alternate realities. Take eg the words "Thank you", or "Bless you" or "It's wonderful". Bless God for this ability to envision alternate realities.

10.10.2021 3:29am - BDS Movement Irate: PA Participating in Dubai Expo Alongside Israel
10.10.2021 3:29am - Interesting. One of my esteemed colleagues has argued that the BDS movement is NOT really anti-Israel, "all they want is to ENGAGE" he said. Today, BDS joins Hamas in scolding the Palestinian Authority for participating in Dubai Expo, alongside Israel.

10.10.2021 12:30am - (Replying to @FriedrichHayek) You are much more ambitious than I am. You aim to emulate Copernicus and Einstein while I am to understand humble me, how I make sense of the world around me, taking evidence and producing actions and explanations, input-output, input-output. Start modest, Einstein next.

10.10.2021 12:17am - (1/ ) Discrimination-free insurance is a fascinating challenge. On the one hand, insurers are allowed to charge by predicted risk, which is a purely statistical notion. On the other hand, "discrimination" is a counterfactual notion, incommensurable w/ statistics. Bringing the two
10.10.2021 12:17am - (2/ ) together must be a cultural shock to the industry, but not to CI researchers, versed in both languages. I was hoping to see a concise definition of "discrimination-free pricing" cast in counterfactual language but, perhaps my oversight, I couldn't find it in the paper. The
10.10.2021 12:17am - (3/ ) transition from Section 3 to Section 4 was too abrupt for me. Let two units be describe by two vectors of characteristics, C1 and C2. Given a causal model M, which predicts E(risk(C1)) and E(risk(C2)); when should price(C1)=price(C2) for the pricing to be considered "fair"?

10.10.2021 12:30am - (Replying to @FriedrichHayek) You are much more ambitious than I am. You aim to emulate Copernicus and Einstein while I am to understand humble me, how I make sense of the world around me, taking evidence and producing actions and explanations, input-output, input-output. Start modest, Einstein next.

10.9.2021 9:46pm - (Replying to @annakaharris @SamHarrisOrg and 5 others) I've never been able to understand what is so mystical about consciousness that demands a special kind of explanation, irreducible to a computational model.

10.9.2021 8:37pm - Malki Roth was murdered a few months before my son, Daniel, by the same hate-driven ideology. Their murderers have still not been brought to justice, and the world keeps looking at the mirror: "Is that me?"

10.9.2021 8:19pm - (Replying to @IfJews and @ifnotnow) Dear @IfJews , I love your opening paragraph: "One day we hope to claim...." Where is this quote taken from?

10.9.2021 7:53pm - This, IMO, is a very good paper, which you deem "a false model of causation". Instead, you propose: "an open-ended mechanisms: learning, rules, selection." I'd consider switching, if you tell me what kind of questions your "open-ended" alternative would answer: Input=? Output=?

10.9.2021 1:05pm - (Replying to @FriedrichHayek and @PhilHaile) From what I've sensed on our channel, things have improved significantly compared to the experience you are describing. I believe at least 50% of econ students would rebel today if their prof tries to avoid basic CI tools or principles, such as those in

10.9.2021 1:29am - This paper on "confounding equivalence" should be of interest to readers who wish to see how abstract ideas of "bias reducing potential" are incarnated in the context of logistic regression.

10.9.2021 12:14am - (Replying to @AlondraNelson46 @WHOSTP and @EricLander46) It is misleading to give the public the impression that "math tools exacerbate health inequalities." The new math of cause and effect does exactly the opposite, it gives scientists the tools to define what "inequalities" we wish to eliminate or sustain, and the means to do it.

10.8.2021 1:19pm - (Replying to @lewbel) The Zoo is well populated, what we need is a cage-manager - a warden.

10.8.2021 7:05am - @_MiguelHernan , what is the simplest graphical model I could use to show that combining experimental and observational studies gives us added benefits? I listened to the interview, but could not understand the principle, save for agreeing that adjustments emulate RCT. Help=?

10.8.2021 5:23am - As part of LA's contribution to the Daniel Pearl World Music Day, you are invited to listen to the LA Jewish Symphony's performance of the movement "Mourning" from Sharon Farber's "Only A Book,'' featuring a beautiful violin solo. October 17th, 7:30pm,

10.8.2021 1:49am - (Replying to @analisereal and @ben_golub) @analisereal , I was told all economists are familiar with the SEMs of the Cowles Commission era. SCMs are no different, except being explicit about interventions, counterfactuals, and other aspects of the environment modeled.

10.8.2021 1:21am - "Free Palestine," "From the River to the Sea" Graffiti Found at UCLA via @jewishjournal
10.8.2021 1:21am - UCLA has the best chalk removing crew in the Western Hemisphere, the best spokespersons to assure us that the graffiti has been removed, yet zero budget to clean up the poison that their faculty spew, which is still posted on UCLA websites, creating more chalk, more crews, etc. 10.8.2021 12:12am - I am still under the magic spell of this lunch with the amazing @EinatWilf , digesting every word she has taught me, and planning our next assault on ignorance, apathy and Zionophobia.

10.8.2021 12:05am - It's mind boggling, painful and alarming to realize how many intelligent, well-meaning and otherwise knowledgeable people refuse to accept the fact that Palestinians still reject the Jewish right to self determination in any part of the land and are totally open about it. 1/
10.8.2021 12:05am - It's so painful for most people to accept this fact that they refuse to listen to what Palestinian spokespersons, intellectuals & educators are saying, loud and clear, and fantasize instead that, under certain conditions, some would be willing to grant Jews that right. Boggling.

10.7.2021 12:56pm - How many book-clubs are there on the clouds? Any club planning to discuss #Bookofwhy will get an honorary mention on this educational channel.

10.7.2021 12:50pm - @alexpghayes , I was going to click the 'like' key, before realizing that it is an old Tweet of mine. I wish there was a key saying: "Well said, still concur!"

10.7.2021 10:52pm - (Replying to @AndoWoods) #Bookofwhy traces several fields through transitions from "data-centricity" to "model-centricity". I dont think there is any field today that is still stuck in the former. One exception: machine learning, where intoxication by low-hanging fruits makes the transition traumatic.

10.7.2021 1:16pm - (Replying to @PHuenermund @pablogerbas and @ben_golub) One twist: SCM accommodates cycles as well, as exemplified in Causality chapter 7

10.7.2021 1:09pm - This tweet calls for a major session of reflection and retrospection: Where have gone wrong? I have assumed that all readers are familiar with acronyms such as SCM (structural causal model), 1st-Law, d-separation etc. It calls for a Glossary of CI (causal inference) terminology.

10.7.2021 1:00pm - (Replying to @Raamana_ and @JDHaltigan) Spreading commonsense among students of SEM and encouraging them to rebel against their professors. This channel is trying to do both.

10.7.2021 8:07am - The positive side of the story is that @GilOfarim is shocked and outraged. Many of my Jewish American colleagues would just hide their Magen David, or refrain from wearing it in the first place. "Why offend the angry Gods?" they would say, crawling back to their cocoons.

10.7.2021 8:07am - The positive side of the story is that @GilOfarim is shocked and outraged. Many of my Jewish American colleagues would just hide their Magen David, or refrain from wearing it in the first place. "Why offend the angry Gods?" they would say, crawling back to their cocoons.

10.7.2021 6:59am - (Replying to @therubykid @FJnyc and 2 others) The suffix "phobia" precludes spaces critical of identity-based Zionism, which is different from the Zionism you possibly have in mind, defined as a people's right to normalcy, and normalcy entails sovereignty in one's historically baked homeland.

10.7.2021 5:44am - Einat Wilf: Western Anti-Zionism and Arab Zionism
10.7.2021 5:44am - How can I miss a Webinar with this title and by this speaker?

10.7.2021 5:31am - (Replying to @LuisaBorrell @Raamana_ and @MariaGlymour) Forgiving the manipulability problem, what can we do knowing that a group G1 of units is "exchangeable with group G2 of units that we couldn't do without knowing?

10.7.2021 5:23am - Jamie Robins lecture on "Causal Inference without Randomized Experiments: How do we know that we are right?" should be of interest to readers of this educational channel. Especially given that (last I checked) Jamie & @_MiguelHernan do not believe in the 1st Law. Worth attending.

10.6.2021 9:51pm - (Replying to @jedgarnaut and @VP) Thanks for correcting my woes. This is even more embarrassing that @VP Harris wooing a juvenile student.

10.6.2021 9:36pm - A thoughtful opinion about the @VP Harris embarrassment Woo to a nation whose leaders are afraid of juvenile students. Woo to higher education abandoned to the rein of juvenile students.

10.6.2021 8:23pm - (Replying to @Raamana_) I don't find the language of "exchangeability" to make life easier. I thought it was a useful language in the pre-causal era, then replaced by SCM, which enables you to determine if one individual is or isn't "exchangeable" with another, not merely intuite about it. @MariaGlymour

10.6.2021 6:14pm - (1/3) The UCLA administration acts like Mahmoud Abbas, condemning ethnic hostilities with one hand, and inviting them with the other. On May 21 this year, the Department of Asian-American-Studies posted (on UCLA Website) a racist statement that criminalizes Israel's very existence; 1/3
10.6.2021 6:14pm - (2/3) it was tacitly approved by an administration that, today, condemns the logical consequences of such criminalization. Jewish students and faculty expect UCLA to be clear on whether Zionist and Israeli students are welcome on campus, by condemning Zionophobia explicitly and 2/3
10.6.2021 6:14pm - (3/3) unequivocally. Thus far the word "Zionism" has not appeared in any of UCLA official statements, as if Israel, the culmination of Jewish history does not exist. 3/3

10.6.2021 8:14am - (Replying to @f2harrell @stephensenn and 7 others) They are red herrings in the context of Lord. e.g "clusters" & "assignments" do not exist b/c there is no analyst or experimenter to "cluster" or "assign". "variance" is shown in the distribution (the ellipses). "within Halls," no variation since Halls = Diet. More? Speak Lord.

10.6.2021 4:39am - (Replying to @JonNichEdwards) I haven't noticed it. Every instantiation of a variable, say X, to one of its values, say x, is an "event" and, similarly, the operator do(X=x) represent an "action". Where have you felt that a more explicit definition is needed?

10.5.2021 11:35pm - (1/2) (Replying to @stephensenn @PavlosMsaouel and 6 others) The power of data fusion to integrate observational and experimental data is well established, and doesn't need my defense. What needs my defense is Lord paradox which is being loaded with red herrings such as "clusters" "assignments" "variance" "within Halls," more?
10.5.2021 11:49pm - (2/2) (Replying to @yudapearl @stephensenn and 7 others) I will be back in 2-3 weeks to see if the herrings are still hanging there, and whether we can get back to enjoy the music of the original Lord Paradox, and the clear, non-numeric way it displays the need for causal assumptions to resolve lingering statistical dilemmas. Chow.

10.5.2021 11:35pm - (1/2) (Replying to @stephensenn @PavlosMsaouel and 6 others) The power of data fusion to integrate observational and experimental data is well established, and doesn't need my defense. What needs my defense is Lord paradox which is being loaded with red herrings such as "clusters" "assignments" "variance" "within Halls," more? 1/2
10.5.2021 11:49pm - (2/2) (Replying to @yudapearl @stephensenn and 7 others) I will be back in 2-3 weeks to see if the herrings are still hanging there, and whether we can get back to enjoy the music of the original Lord Paradox, and the clear, non-numeric way it displays the need for causal assumptions to resolve lingering statistical dilemmas. Chow.

10.5.2021 6:54pm - (Replying to @Raamana_0 I am talking about "ranking", and ranking is done by surveying academics like you and me. What do you think mob's rule does to the perception of MIT in academia?

10.5.2021 6:14pm - Toilets are the greatest mystery of ancient times. Why? Because historians, writers and prophets did not take the time to describe them. An unforgivable case of selection bias.

10.5.2021 6:07pm - Speaking of MIT, where does it leave John Hopkins' reputation? This Washington Post piece rattles me to ask our Hopkins' readers: Isn't it reflecting on you, and your silence?

10.5.2021 5:30pm - Princeton rises to where MIT once was. I hope our MIT-affiliated readers write and explain to their President what this does to MIT's reputation.

10.5.2021 5:20pm - (Replying to @stephensenn @PavlosMsaouel and 6 others) In my understanding, Lord's story does NOT have any design in it (especially in the Wainer and Brown version). The treatment (diet) varies between halls in the same way that the barometer reading varies with the atmospheric pressure. No "choice" involved.

10.5.2021 4:57pm - (Replying to @MarkAnnuncio @stephensenn and 7 others) I am familiar with many things that analysts found useful in various studies but, given our problem, all technical terms must be translated to behavior of students in the story environment. "Clustering" resides in the analyst's lens, not in students' choice of diet.

10.5.2021 4:29pm - (Replying to @MarkAnnuncio @stephensenn and 7 others) I explained why I must excuse myself 1. It's not a "study" it's a story. 2. Yes! As was clearly stated. 3. "Clustering" is an "experimental design" notion. Forbidden in our discussions, unless translated to behavior of students in their environment.

10.5.2021 4:17pm - (Replying to @MarkAnnuncio @stephensenn and 7 others) Sorry, diet and halls are perfectly synonymous in the story that I read. I must excuse myself from this discussion. Willing to resume it with anyone who agrees to refrain from using experimental design vocabulary which, IMO, confuses rather than illuminates the problem at hand.

10.5.2021 4:05pm - (Replying to @MarkAnnuncio @stephensenn and 7 others) Lord's story has no "treatment" nor "assignment" nor "hierarchy". All these terms are borrowed from a different planet. If we stick with "students" "Halls" (="Diets") weights, and who chooses what, clarity will shine. (I disagree with Hopkins on the definition of "unit")

10.5.2021 3:01pm - (1/2) (Replying to @yudapearl @PavlosMsaouel and 6 others) 1/2) Lord's story does have a design (at least in the Wainer and Brown version) The treatment is varied between halls. It could have been varied within. That's a choice & the choice has consequences.
10.5.2021 3:04pm - (2/2) (Replying to @stephensenn @yudapearl and 7 others) The theory behind this is nearly a century old. It's pretty basic and, without taking anything away from other theories, experimental design is a rather important theory of statistics about estimating causes.

10.5.2021 2:45pm - (Replying to @artistexyz) I don't believe I ever said " PO does not include the 3rd rung." The expression Y_x(u) is by nature 3rd rung. What I might have said is that PO (as defined by Rubin) is narrowly defined on manipulable variables X. (eg "no causation w/o manipulation")

10.5.2021 2:34pm - (Replying to @stephensenn @PavlosMsaouel and 6 others) In Lord's story, I see only one way to adjust for initial weight, and I don't see "hierarchical dataset". Let's stick to the story and see all hierarchies collapse and clarity shinew.

10.5.2021 2:18pm - (Replying to @PavlosMsaouel @stephensenn and 6 others) It is not a matter of choosing focus. Lord defined his dilemma clearly and we have to help him resolve it, even if our favorite occupation is to estimate variances and design blocks and clusters. Lord's story has no "design" and the dilemma concerns bias, not variance.

10.5.2021 7:23am - If MIT caves to the mob, who will stand up for science? A depressing story, that shouldn't end here!

10.5.2021 5:34am - Congratulations to Carlos Cinelli @analisereal for joining the Editorial Board of the Journal of Causal Inference (JCI) Carlos arrival strengthens JCI ability to offer friendly & thoughtful reviews of new ideas in all aspects of CI.

10.5.2021 12:28am - (1/3) @pablogerbas , this is indeed an important observation. The entire theory of do-calculus, hence of interventions & causal effects, can be conducted at Rung-2, with no reference to unit-level counterfactuals. This is done (neatly) on p.23-4 of Causality
10.5.2021 12:28am - (2/3) showing how the vast space of interventional probabilities can be encoded economically, in the form of a single "causal Bayesian network" G, with no reference to counterfactuals. Neyman and Rubin could have started here, and get the blessings of Fisher and others (eg Dawid)
10.5.2021 12:28am - (3/3) who deemed counterfactuals unscientific. But Rubin opted for Y_x(u) and the siren song of "missing data" from which generations of statisticians haven't recovered yet. Yes, there is more to the Ladder than meets the eye; some called it the "double helix" of CI.

10.4.2021 10:25pm - (1/3) You are right. Every do-expression (ie, Rung 2) follows from the 1st Law, which is Rung 3. The separation between the two was made to distinguish what can and cannot be estimated from RCT. Hume, we should remember, lived before RCT's were invented and, more importantly,
10.4.2021 10:25pm - (2/3) before the notion of "causal model" came into being (ie, a mathematical object that contains all the causal information you would ever need), so his notion of "definition" was different than ours, and he couldn't have annunciated the 1st Law. That's too bad; he could've saved
10.4.2021 10:25pm - (3/3) philosophers and economists much headache. Economists could have adopted the 1st Law in the heyday of the Cowles Commission, but...I'd leave it to historians to tell us what happened, and why some still resist it. Here is a glimpse,

10.4.2021 9:01pm - (1/ ) Your question deserves more than a twitter blip but, for now, I would say YES, P(Y=y|do(X=x)) is a "measure" of "manipulative causation", with two qualifications: (1) It applies to non-manipulable X as well, and (2) It's a population-level measure as opposed to unit-level,
10.4.2021 9:01pm - (2/ ) which is defined by the 1st Law: Y_x(u)=Y_{M_x)(u). Note: In contrast to PO, the 1st Law is non-manipulative; e.g., "If it weren't for this earthquake, my house would still be standing" is perfectly well defined. Note: In contrast to PO, the 1st Law is model-based.

10.4.2021 5:42am - (Replying to @stephensenn @ESteyerberg and 4 others) #Bookofwhy, p.216 says: "the students eat in one of two dining halls with different diets." I interpret it to mean: "each student eats in one of the two dining halls, and these two dining halls serve different diets." How did you interpret it?

10.4.2021 12:49pm - What do you know? I'm going to have a new Alum Colleague, Angela Merkel, who will be awarded a doctorate degree from my alma mater, The Technion. Honored to have you join our family, Angela.

10.4.2021 12:37pm - (Replying to @lewbel) The Zoo was not wrong. But speaking of parameter theta when we have in mind a complex Question like Q=Conditional ACE may be confusing to students who would be looking for a parameter named theta, instead of a PROCEDURE for computing Q.

10.4.2021 8:12am - (Replying to @stephensenn and @NP_Jewell) If you see difficulties with the experimental version of Lord Paradox, and if these difficulties persist in the original observational story, please express them in the vocabulary of the latter - I get lost hearing about "block design" in the context of students choosing Diets.

10.4.2021 8:02am - (Replying to @stephensenn @camjpatrick and @NP_Jewell) I thought we agreed Hall = Diet Namely each Hall has one and only one Diet, printed on the menu and mailed to the student's home to choose. Once chosen, the student eats at that Hall for the entire year. No design, just a story.

10.4.2021 7:56am - (Replying to @stephensenn and @NP_Jewell) Disagree. The Book of Why analysis of design 1 is right. There is no "block structure" in the story, and no one is "designing" any experiment here. It is a natural process involving 3-variables that IMO need not be complicated with alien notions taken from design of experiments.

10.4.2021 7:37am - (Replying to @stephensenn @ESteyerberg and @NP_Jewell) Please, do not hold us in suspense. Tell us what that "untestable assumption" is. I am all ears!!!

10.4.2021 7:31am - (Replying to @stephensenn and @NP_Jewell) I beg NOT to consider design 2, and you reply, again and again, that I have a problem recognizing the difference. The graphs for the two designs are of course different, but I DO NOT WANT to consider design-2. Please, Please, let's stick with design-1.

10.4.2021 7:21am - (Replying to @ESteyerberg @stephensenn and @NP_Jewell) Causal experts simply answer the question asked: What is the causal effect of diet on weight gain (is it positive as statistician-2 claims, or zero, as stat-1 claims). They would stay away from all unnecessary complications. Have you read or #Bookofwhy

10.4.2021 6:45am - (Replying to @yakir_bella) Dying to see it again.

10.4.2021 6:44am - (Replying to @cubic_logic) Where did we get 1941. Netanya was founded 1928, and I thought the top picture was taken not much later.

10.4.2021 6:41am - Important observation. Causality is needed because the change from one domain to another acts like an intervention.

10.4.2021 6:37am - @NP_Jewell and other statisticians of good standing. Can anyone see a problem with: "each student eats in one of the two dining halls, and each of these two dining halls serves one diet, different from the other." I am told statisticians would see here complications -- I don't.

10.4.2021 6:19am - Another breath-taking change, from 1928 to 2021. Nostalgic Netanya. I knew Netanya from day trips our school took in 1944-47 to her beautiful beaches, overlooked by steep cliffs. I also had a summer job there during college, conducting a youth choir. Nostalgic and breath-taking.

10.4.2021 6:07am - (Replying to @DumisaniTemsgen and @Eve_Barlow) Nor have we heard much from Jewish leaders. Had he denied the holocaust or talked "Benjamins" we would hear them screaming Haven and Earth.

10.4.2021 5:45am - (Replying to @camjpatrick @stephensenn and 5 others) #Bookofwhy, p.216 says: "the students eat in one of two dining halls with different diets." I interpret it to mean: "each student eats in one of the two dining halls, and these two dining halls serve different diets." How did you interpret it?

10.4.2021 5:42am - (Replying to @stephensenn @ESteyerberg and 4 others) #Bookofwhy, p.216 says: "the students eat in one of two dining halls with different diets." I interpret it to mean: "each student eats in one of the two dining halls, and these two dining halls serve different diets." How did you interpret it?

10.4.2021 5:12am - (Replying to @stephensenn @ESteyerberg and 4 others) Corrections: 1) You have a group of students sampled from some population; Heavy students tend to prefer one diet over another, (2) We are given the entire distribution of the final weights, initial weights and diet. (3) Wanted: Effect of diet on weight gain. See #Bookofwhy p.216

10.4.2021 4:49am - (Replying to @stephensenn @ESteyerberg and 4 others) I don't recall introducing complications. But if you recall, please remind us what they are. Three Halls? Each Halls having 3 different diets? Please tell us, and why it is so important to introduce complications to a simple story whose purpose is to demonstrate the power of CI.

10.4.2021 4:04am - (Replying to @ccaballeroh10 and @rpatrik96) No major changes if you have the errata sheet.

10.4.2021 4:02am - (Replying to @stephensenn @ESteyerberg and 4 others) I don't understand this universe. We have only two Halls, each has one and only one diet. Treatment =diet. Why complicate things? I'm not dismissing Nelder's calculus. I want to focus first on a simple version of Lord's Paradox. Can we?

10.4.2021 3:56am - (Replying to @ccaballeroh10 and @rpatrik96) I havn't noticed. second edition (2009) is the latest. We are working on 3rd, but Cambridge is dragging its feet,... dragging....dragging...

10.4.2021 1:29am - (Replying to @mendel_random and @BristolJsoc) Illuminating! My heart jumps from joy every time I see an anti-Zionist professor treated the same as Islamophobic professors. My heart sink seeing the former proud of their affliction, while the latter deny it.

10.3.2021 11:01pm - To readers who ask what is "a good Jew"' Ans. A "Good Jew" is one who condemns Israel whenever you do, even knowing that she can and ought to be defended. It is a formal definition that fits about 50% of American Jews.

10.3.2021 10:22pm - (Replying to @HannesMalmberg1 @pablogerbas and 2 others) But what do we do with the "exclusion restriction"? Leave it unspecified? Further, if I go to a PO-statement, I lose all the beauty of do-calculus, with the help of which we get generalizations. We further loose testability. Can you tell if a set of PO statements is testable?

10.3.2021 8:41pm - (Replying to @HannesMalmberg1 @pablogerbas and 2 others) So, can I conclude that what makes "design-based" different from "model-based" is the reliance on some variable(s) Z known to be physically randomized. We have been doing it, routinely, in what we labeled "Data Fusion" Have we been crossing into "design" territory? I hope not.

10.3.2021 8:25pm - That's why the world can't forgive Israel. It's utterly unforgivable to let sheer incompetency spoil what everyone is expected of her. That's probably what made @GovHowardDean of Vermont mad this morning; he had to do something to save face -- Israel hasn't delivered.

10.3.2021 8:00pm - Not at all. Sarah still can't believe the angelic Squad would betray her so painfully. How could they do it to her, the "good Jew", who always condemned Israel when the Squad did, and things were so ecumenical and cozy. How could they?

10.3.2021 6:16pm - (Replying to @pablogerbas @instrumenthull and @lewbel) This is an interesting new twist. So, the word "design" comes from the assumed randomization of some instrument. What about exclusion restriction, is it "design-based" or "model-based"? What about "fusion-models" which allow us to combine experimental and observational data.

10.3.2021 5:55pm - (Replying to @instrumenthull and @lewbel) Please, let's continue with the differences. Do you mean, when strengthening the model with "function shaping" restrictions, we do the shaping on some of the functions, but not all? Which?

10.3.2021 3:27pm - (Replying to @FriedrichHayek and @causalinf) The causal explanations used in #Bookofwhy are not based on laws.

10.3.2021 3:05pm - Thanks for posting. Happy Kincaid sees CI as a "revolution" in economics but, where on earth does he get the idea that "moderating causes" and "necessary causes" are "not easily handled" in SCM? The whole theory of attribution is built on them. See

10.3.2021 2:32pm - (Replying to @DKedmey) They can borrow some from the US Congress.

10.3.2021 2:26pm - I refuse to believe that @SarahKSilverman is so naïve as to be surprised by the Squad, whom she courted as a "good Jew", show no interest in protecting Sarah's relatives in Israel from Hamas rockets. Zionophobes, unlike antisemites, respect you only when you respect yourself, and
10.3.2021 2:26pm - (2/ ) they will never, but never, show any empathy for Israelis. Why? Because this may remind the world that Israel has a human face, perhaps even a reason to exist, which goes against the Squad's mother milk and against what their voters elected them to deny. Her heart is broken!! 10.3.2021 3:29am - (1/ ) If I may, I think the whole idea of a "causal parameter" (which I call "causal query") can stand refinement, using the 1st Law. The Zoo paper gives the impression that identification deals with parameters in the model; no so, it deals with counterfactual quantities that need
10.3.2021 3:29am - (2/ ) be computed from the model using Y_x=Y_{M_x}. Example 3 does not tell readers how Y(0) and Y(1) are defined/computed. I've found it to be a major stumbling block in communicating with econs, see; it's hindered by lacking a formal definition of Y(x).
10.3.2021 4:21am - (3/ ) If I still may, a concise general definition of "identification" would be helpful: A causal query Q(M) is identifiable, given a set of assumptions A, if for any two (fully specified) models, M1 and M2, that satisfy A, we have: P (M1) = P (M2) ==> Q(M1) = Q(M2)

10.3.2021 4:02am - (Replying to @AndoWoods) I wish I could suggest an alternative to Causality, but I honestly can't.

10.3.2021 3:57am - (Replying to @avicenna @LuisaBorrell and 2 others) If it is "very much the same", why even mention RCT? If it is "first better than the second", show one "design" decision that is facilitated by the first. I'm not objecting to the latter, I'm just asking whether "many people know how" comes natural, or from programmed training?

10.2.2021 4:46pm - (Replying to @ModelYendofICE) Interesting line of logic, typical of Zionophobic minds: No one should dictate to others what to do except WE -- the ultimate authority to dictate who should or shouldn't dictate ... A crown trophy for the "Museum of Zionophobic Thought." More contributions invited.

10.2.2021 8:47am - (1/ ) I don't feel comfortable when the fitness of a Zionophobic professor to teach at a university is judged by the degree to which his words qualify as "antisemitic". Zionophobic utterances should be judged on their own racist content, and fitness to teach should be assessed
10.2.2021 8:47am - (2/ ) the same way fitness of Islamophobic professors are assessed. We have Bristol University affiliates among our readers, and they should be able to tell us if an Islamophobic professor would be allowed to teach on their campus. @BristolJSoc @mendel_random

10.2.2021 8:15am - Bristol University professor fired for antisemitic comments

10.2.2021 7:42am - (1/ ) Anyone who mentions Imbens 2020 JEL paper without my response is endangering the well-being of many econ students. As to the pitfalls of propensity score, they can clearly be understood by students of CI, eg. . This chapter
10.2.2021 7:42am - (2/ ) contains an important result that is not seen in the graph, and is under-appreciated: If you run propensity score matching with a bad W, the asymptotic bias is not reduced, but remains the same as the one resulting from adjustment for W. @causalinf @VC31415 @lewbel @PhilHaile

10.2.2021 6:47am - Unforgettable courage. A badge of honor to our Danish readers; noting that this story is told in every school in Israel and, consequently, saying "I am from Denmark" is a ticket to a warm welcome from anyone you wish to befriend.

10.2.2021 4:22am - All this "savvy", hush-hush work "behind the scene" has resulted in abandoning the "scene" to the control of anti-Israel mafia, and has given young Americans the impression that standing publicly with Israel is less than "cool", super-just and utterly noble. Change this game @VP

10.2.2021 12:07am - I've learned to mistrust people who tell me "we are working with 'X' behind the scene". Invariably, what they mean is: "We give X a cover for inaction and X give's us an appearance of action." @VP Harris should state PUBLICLY what Israel means to her. @MarkMellman @DavidHarrisAJC

10.1.2021 9:51pm - Harris' office does damage control over student's Israel 'ethnic genocide' comment via @politico
10.1.2021 9:51pm - "Damage control behind the scene" is what @VP Harris should NOT do. She owes her voters a public apology for falling into the trap laid by that student, endorsing a "fabricated lie" as "your truth," and a frank reflection how she plans to prevent such blunders from reoccurring.

10.1.2021 9:14pm - (Replying to @fishmanjulie @AJCGlobal and @MarkMellman) The cultural damage of the Squad far exceeds it's immediate political influence and will eventually evolve into political influence, if we continue to ignore it as a passing phenomenon. It allows professional Israel bashers to spread lies with impunity, as @VP has experienced.

10.1.2021 2:29am - (Replying to @isVy08) Elias Bareinboim has a couple of papers in this direction. But the basic bridge from CI to NN is defined already in CI - take the resulting estimand and fit every conditional probability term using a NN functional approximation method.

10.1.2021 2:13 am - (Replying to @BrendanNyhan) Do you see any push in soc sci research to elevate the "credibility revolution" into a "modernity revolution" based on structural causal models?

10.1.2021 2:00 am - How is it possible then that all the so called "Peace makers" (Gamal Helal, Martin Indyk, Daniel Kurtzer, Rob Malley, Aaron David Miller and Dennis Ross) spent years chasing pebbles, blinded to a ten-mile tall barrier to peace, which every Israeli child can see.

10.1.2021 1:43am - (Replying to @AJCGlobal @DemMaj4Israel and @MarkMellman) We are still waiting to learn what Nancy Pelosi has heard from both @ADCGlobal and @DemMaj4Israel and @MarkMellman regarding this issue, whether she responded, and whether her response is deemed satisfactory.

10.1.2021 1:24am - We were too eager to rejoice. Back to the dark ages.

9.30.2021 7:58pm - This telling story happened on my lawn, here, at UCLA, and I didn't know. I am retweeting to alert colleagues, here and elsewhere, to check their lawns. Every innocent looking piece of grass may harbor another such story, in which innocent you is the main character.

9.30.2021 5:27am - (1/ ) As I listen again to this panel discussion and ask myself: "what is it that hinders communication between CI and ML folks?" I conclude that it stems from differences in appreciating the heights of the barriers between the Rungs on the Ladder of Causation. ML folks
9.30.2021 5:27am - (2/ ) view the integration of causal inference in ML as just another task, like recognizing giraffes instead of elephants. For CI folks, a transition between Rungs represents a violation of basic conservation law, analogous to the conservation of energy or momentum. For example, the
9.30.2021 5:27am - (3/ ) fact that we cannot climb from Rung-2 to Rung-3 even if we perform all possible interventions, on all variables and combinations of variables, is not usually appreciated in ML circles. This, in turns, affects the appreciation of extra-data information in the form of models.

9.30.2021 4:24am - (Replying to @dianaleonie1) Well put. And this sort of spineless leadership erodes American confidence in Gov't. I used the holocaust example precisely to demonstrate that there are limits even to extreme spinelessness.

9.30.2021 2:51am - This is a powerful new paper on partial identification, adding a new layer to the symbolic and Gibbs sampling methods that have been used thus far. The spectrum of examples is also impressive. Highly recommended.

9.30.2021 2:22am - (Replying to @kahntra @statwonk and @joshrfeldman) I'm saying @VP Harris came prepared to embrace and forgive whatever those students heard on Al- Jazeera, or in their dorms. She was not prepared to listen carefully to what they said and to stand, as a leader, for her values and truth.

9.30.2021 2:06am - (Replying to @PHuenermund) Agree. I did not see @mikejrob 's thread. Now that I read it, I am wondering how may of these examples: an econ student will get correctly, if guided by the "New School" criterion of "as if randomized." Also, would @mikejrob switch to the "Modern School?"

9.30.2021 2:06am - (Replying to @CindyDarrison) Her followers are learning quickly that, when it comes to Israel bashing, any lie can be spread with impunity. Just observe her followers among UCLA faculty who felt it's perfectly safe to post those lies on Department's website. The cultural damage of the Squad exceeds it's #.

9.29.2021 9:24pm - (Replying to @joshrfeldman and @statwonk) Quite evident, aggree.

9.29.2021 9:23pm - (Replying to @Surly_Frog) Let's agree that we uphold different sets of values. I do not think it's funny for a member of Congress to campaign for removing protection to hundreds of thousands of children, while admitting her pitiful state of ignorance.

9.29.2021 9:11pm - (Replying to @JonathanCorbin2 @joshrfeldman and @statwonk) Blood libels are not funny either, they actually kill. I am surprised that people upholding decency and sensitivity would tolerate words such as "Israel committing ethnic genocide" and, numbed by repetition, would not jump to the ceiling with condemnations and disgust.

9.29.2021 8:53pm - (Replying to @learnfromerror) No, the holocaust example was my way of ridiculing the conspiracy theory fancied by the student. I should have made it clear. I would apologize to VP Harris, if she apologizes to my friends in Israel for not stopping that student with: "Enough! this is blood libel - words kill!"

9.29.2021 8:29pm - (Replying to @FaithQuintero7 and @HillelNeuer) I know of at least one such movement: Zionism. It goes even further by advocating national rights as well, if only reciprocated to their neighbors.

9.29.2021 7:07pm (1/2)
* Student: It breaks my heart to see Israel committing ethnic genocide...
* Harris: Your voice, your perspective, your experience, your truth cannot be suppressed.
* Student: And it breaks my heart to visit fancy Museums to a Holocaust that never happened.
* Harris: Your
9.29.2021 7:07pm (2/2) (Replying to @yudapearl)
[Harris: Your] voice, your perspective, your experience, your truth cannot be suppressed.
* Student: And it truly breaks my heart to hear how the 2020 Presidential election was rigged.
* Harris: ...What?

9.29.2021 1:47pm - (Replying to @PhilHaile and @PHuenermund) You are absolutely right. They tried and succeeded to show equivalence. It has some merits, but you need to construct a new "moral graph" for each conditioning set.

9.29.2021 7:13am - I must watch it for the 7th time.

9.29.2021 7:11am - @RepAOC represents the knowledge and mentality of most people I met who use the word "occupation". What they're trying to say is: "I must say something nasty about Israel, I know that "apartheid" and "ethnic cleansing" are idiotic allegations; "occupation" sounds factual & ugly".

9.29.2021 5:17am - (Replying to @rina31420183) Enjoy the beach and the people, and may an "iron dome" protect them both. Don't forget to go to the synagogue and dance -- today is Simchat Torah, when Jews rejoice the reading of Genesis again.

9.29.2021 5:02am - (Replying to @EyalShay2) Amazing! Are these camels crossing the river?

9.29.2021 4:54am - Jeremy Corbyn sincerely believes he has "fought racism all his life", for he has never considered Zionophobia a form of racism, nor does he view the destruction of a people's homeland to reflect a sign of moral deformity. @AdamBienkov

9.29.2021 4:27am - She looks really good in that dress, but I would be more accurate (she really loves some Jews, eg. Peter Beinart) and paint it: "Kill only Israeli babies", an undeniable consequence of her votes and rhetoric.

9.29.2021 4:19am - The Ambassador's last statement is significant, dispelling rumors that the Biden administration will try to distance itself from anything connected with Trump's legacy.

9.29.2021 2:55am - Greatly astonished, I see with amazement that the paper I've recommended:, has been cited ZERO times. I assume it's the title that turned people away "Graphoids over Counterfactuals". Today I would label it "Inference with Potential Outcomes". A beauty!

9.29.2021 2:20am - I consider myself fortunate to have witnessed the miracle of Tel Aviv growing before my eyes. My grandfather landed on these dunes in 1924, and I was born, educated, and grew up with the bustling city on the right, its beaches, theatres, markets and bars. May she be protected.

9.29.2021 1:10am - (1/ ) We have discussed at length the pitfalls of postulating "conditional ignorability" assumptions as if their truth is cognitively discernable. Some researchers continue this practice, as seen here:, in which I've noticed another pitfall: non-testability.
9.29.2021 1:10am - (2/ ) It is misleading to state that ignorability assumptions, even in their clumsy PO outfit are not-testable. A highly recommended paper provides systematic methods of unveiling the testable implications of any set of PO assumptions. Moreover, when it comes
9.29.2021 1:10am - (3/ ) to missing data, where testability is truly tricky, Mohan & P.(2021) provide three simple conditions of testability: I hope PO-minded researchers take notice: your allegiance to an outdated notation does not exclude modern ways of testing assumptions.

9.28.2021 6:06am - (Replying to @Claudia_Sahm) I meant to say: Trying to mechanize is the best teacher. Success and failure are secondary surface phenomena.

9.28.2021 5:08am - (Replying to @hangingnoodles @Claudia_Sahm My jerk reaction upon hearing "I have a set of skills" is to ask: "Can we mechanize them?"

9.28.2021 3:06am - It sounds like Isaiah's Lamb and Wolf incarnated. True, this conference takes place in Erbil, not Baghdad. Still, note the size of the audience, and her unwavering understanding that all roads to peace and modernity must go through Jerusalem.

9.28.2021 2:48am - (1/2) As some of you know, October is when we celebrate -a global concert dedicated to "Harmony for Humanity". As usual, Taipei is the first to announce its participation by continuing Taiwan's longest running outdoor concert:
9.28.2021 2:48am - (2/2) Enjoy your favorite bands if you happened to pass by downtown Taipei on October 23rd, 2pm-8pm. (We trust the Chinese Gov't does not take it as a provocation.)

9.28.2021 1:44am - (1/ ) The reason Bennett does not accept it today is because "mutual acceptance" sounds untestable and unenforceable. But working on testability and enforceability conditions is much more productive than leaving things as they are, especially when the inevitable rejection by Abbas 1/
9.28.2021 1:44am - (2/2) will turn you into a saint, and will justify why the status quo must be maintained, however painful, until the Palestinian leadership undertakes some meaningful action. If they do, peace is 10 times closer, if they don't, you are still a "peace-seeker", not "peace-avoider". 2/2

9.28.2021 12:57am - 2-state with mutual acceptance is something that the vast majority of Israelis aspire to achieve. The alternatives, 2-state with no mutual acceptance, or 1-state, everyone knows will NOT work. Isn't it better to propose something that at least one side believes in?

9.28.2021 12:36am - (Replying to @yakir_bella) Given the qualifiers "fair to say", "many", "may be" and "most", I would say: yes, it's fair to say. Though many social science academics are acquiring competency in CI.

9.27.2021 9:34pm - (Replying to @southremembers2 and @DJMinded) It does work with social media. Not only are they 'considered' lambs after repeating the calling hundreds of times, but they themselves begin to believe they are lambs. I bet @DJMinded does.

9.27.2021 8:50am - Here is Bennett's speech at the UN which I've found to be honest and inspiring. IMO, he should have reiterated Israel's commitment to a two-state solution, "Equally legitimate and equally indigenous." This, of course, no Palestinian can accept.

9.27.2021 8:11am - (Replying to @DJMinded) Apartheid means "separation". And most lambs prefer to be separated from their predators for obvious reasons. No one would call that preference "apartheid" since it is imposed by the howling wolf who, btw, does not even try to hide its intentions while howling: "Apartheid!!"

9.27.2021 7:33am - (Replying to @DJMinded) I would rather be an ethnic-nationalist lamb, behind a defensible fence, than a predator wolf pretending to be a globalist.

9.27.2021 7:18am - (Replying to @DJMinded) The first reaction of the wolf upon seeing the fence was a long howl: "War crime!!!" In time, the wolf has learned to moderate its howl into: "Apartheid!!". It has become the national anthem of all the lamb's predators. They can't stomach the fence.

9.27.2021 4:44am - (Replying to @PHuenermund) By saying "state your assumptions clearly" one normally means: "state your assumptions using DAGs", because we do not know an equally clear language for stating assumptions. I assume this is what Jason meant, though I would hate to put words in his mouth.

9.27.2021 2:11am - (Replying to @AlanDersh) The problem with Levin's bill is that it scolds the lamb for building fences prior to asking the wolf whether he ever read Isaiah.

9.27.2021 1:50am - What is it about SEM education that I find to be most unacceptable? Ans. Standard SEM texts lack the concept of a "Query". For example, they identify a model, rather than a specific combination of parameters that answers the research question. They estimate a "model fit," rather
9.27.2021 1:50am - than fitness of that portion of the model which answers the query. This avoidance reflects the traditional discomfort with the causal interpretation of structural parameters. How can you pose a query if, lacking d-separation and do-operators, you don't know where/how to find it?
9.27.2021 1:59am - Put another way, how can you pose a causal query if you believe that the model parameters have no causal interpretation, and that SEM is just a parsimonious encoding of covariance matrices.

9.27.2021 1:26am - (1/2) So, what's wrong with Congressman Andy Levin's new bill, contrived to encourage a two-state solution to the Arab-Israeli conflict? Indeed, why not study seriously Isaiah prophecy: "The wolf shall dwell with the lamb (Isaiah 11:6); a powerful reminder
9.27.2021 1:26am - (2/2) that all conflicts may come to an end? The trouble with Levin's bill is that it scolds the lamb for building fences prior to asking the wolf if he ever read Isaiah.

9.26.2021 4:54am - Four years ago, the ACM organized a panel discussion on "Advances in Deep Neural Networks". It is illuminating to listen to the discussion today: and wonder how little has changed and how hard it still is for CI people to communicate with DL folks.

9.26.2021 10:42pm - (Replying to @hardsci) Glad you are using Kline's book, it's the only SEM text that I recommend. For reading material, don't miss which demonstrates working with counterfactuals, and, from Handbook of SEM, which connects SEM to modern causal analysis.

9.26.2021 9:34pm - Reading old papers is like drinking old wine. Glad you re-posted it. I was especially mused by all the bad things Rubin said about graphs which, since he stopped saying them, are likely to be rediscovered by his disciples (e.g., Imbens or be forgotten.

9.26.2021 1:57pm - Israel sending finalist to entrepreneur contest in Dubai via @JNS_org
9.26.2021 1:57pm - This is what "patient-engagement platform combining artificial intelligence and machine learning, behavioral science and psychology" can do for you. Not to mention the human touch patients are getting.

9.26.2021 1:52pm - That's the kind of graduate students I'm hoping to attract.

9.26.2021 12:41am - AOC Proposes Amendment Blocking U.S. Weapon Sale to Israel via @jewishjournal
9.26.2021 12:41am - It seems @AOC could not stand her own stench. She forgot the 1st rule of Zionophobia: "Do not touch Rabbis and you would be allowed to spread lies with impunity, especially Zionophobic lies. Rabbis are too toxic."

9.25.2021 8:18pm - (Replying to @JohanDH2O @edehtiar and 2 others) I've asked for ONE, not all.

9.25.2021 6:39pm - (1/ ) A 82-page study on the current state of AI It should be exciting to read, but I find myself yawning, why? Is it because I've heard these claims and challenges so many times before? Or because the advances that turned me optimistic about the feasibility 1/
9.25.2021 6:39pm - of AGI do not seem to have excited the authors of this report. I am referring of course to the development of computational models of "deep understanding". It seems the authors do not fully appreciate what I once described as "sitting on a gold mine."

9.25.2021 5:53pm - The Squad never promised anyone moral integrity or protection of Palestinian lives. It promised voters and donors one thing: To malign Israel whenever given a stage. They deliver on this promise admirably, exceeding all expectations. Why are people disappointed? @OAC

9.25.2021 4:08pm - Who says @AOC has no principles, and the @NYT no ethic? Here it is: "Every human life is sacred and every child should be protected, except those born in sin, and those whose being is a 'war crime'".

9.25.2021 2:12pm - (Replying to @stephensenn and @MarkAnnuncio) I also like the "cleaner" Wainer-Brown version. And I challenge all those who see "many variances" there to tell me which sentence in my treatment of the paradox causes them a pause. Here it is:

9.25.2021 2:01pm - We all know that politicians make difficult choices; no matter how you vote someone gets disappointed. But this is 1st time in history that a politician "CRIES" for a choice he/she made. @AOC 's tears reflect her honest assessment of the moral character of her voters. I'd cry too.

9.25.2021 1:27pm - (Replying to @PWGTennant @AgEconomist and 2 others) It is no longer under my control, since most people know me as Judea - pronounced: Jew-de-Ah -- I can no longer revert to the way my mother called me: Ye - Hu - Da, or the way my classmates called me: Yu-Dah. So, let's use Jew-de-Ah, which rymes with David's kingdom of Judea.

9.25.2021 3:32am - Sweden decided to boycott Durban? Amazing! And I placed all my hopes on Norway! Just because the Norwegians I met impressed me as descent. A classical example of selection bias: I shamefully forgot the descent Swedes I met in Uppsala 1991. Memory is the key to a biased judgment.

9.25.2021 12:06am - (Replying to @yudapearl and @stephensenn) distributions, rendering it independent on experimental design (eg Nelder) or estimation methods. We can then ask: Which of the two statisticians is right assuming infinite samples and an agreed-upon subpopulation of students? The 2 arguments are compelling & the answer elusive.

9.24.2021 11:56pm - (Replying to @stephensenn) Agree that "terminology is unimportant" but if you agree that what you call "different levels" can be captures by "conditioned on different factors" then the entire literature on partial correlations can be brought to bear, and Lord's dilemma can be cast in the language of joint

9.24.2021 10:46pm - Much of what he says about CERN is true about science in general, and about AI and Data Science in particular, that I think we should listen to him (Eliezer Rabinovic) carefully.

9.24.2021 10:26pm - (Replying to @JaimeFeRo and @Nichogachet) "sin retorno" is exactly what happened to me.

9.24.2021 8:05pm - (1/ ) Glad we have an example to guide us but, for the life of me, I see partial correlations, not "multi-levels". Suppose Z stands for the set of factors that determine the overall calories consumption of an individual (say, body-weight or income) then, we have beta_{xy.z}>0 and 1/
9.24.2021 8:05pm - (2/2) beta_{xy}<0 which is not surprising, as shown here Lord's paradox shows up when we consider a fixed Z, and do not jump from one Z to another. So, which statistician is right when we fix Z? 2/2

9.24.2021 6:17pm - (Replying to @bariweiss) The NYT can't tell the difference between a Rabbi and the bearded Taliban delegate to the U. N. So why did he convinced the poor bird to vote against her "principles" (Squad w principle?). Look at all the sympathy @AOC is getting and all the bad publicity Rabbis are getting.

9.24.2021 3:31pm - Thanks, @d_malinsky . This is indeed that paper I had in mind which lists the inaccuracies in Platt & Schlaifer (1988). It is dated 1992, namely, after d-separation but before the do-calculus. Thanks for bringing this paper to the surface.

9.24.2021 5:23am - (Replying to @stephensenn @mendel_random and 2 others) I am trying to understand where the multidimensionality of correlation comes from, but can't find a compelling example, void of links to other papers. Can you help? If it is an intrinsic feature of correlation than it be conveyed w/o reference to methods of estimating correlation

9.24.2021 4:15am - (Replying to @TFLangkaas @tkaiser_science and 2 others) I would add a 3rd takeaway: 3.Triangulation of diverse knowledge sources requires a formal language for fusing them together. See my response Deaton and Cartwright who hail triangulation but lack the language.

9.24.2021 2:23am - (Replying to @stephensenn @offaltube and @_MiguelHernan) I tried to read the paper you suggested has a connection to "transportability". But I could not find a connection. (Such screening, BTW, can be done quickly using syntactic analysis). Another BTW, where is the attached quoted from?

9.24.2021 2:09am - (Replying to @stephensenn @mendel_random and 2 others) Lord Paradox is one of my favorite, topping even Simpson's. See No statistical argument has yet been offered to decide which of the two statisticians is correct. Sheer beauty!

9.24.2021 2:02am - This must be a Polish translation of #Bookofwhy; I haven't seen it before, thanks for posting. I wish my mother was alive. She would really get a kick showing it to her Polish speaking neighbors and convincing them that Data Science is no science w/o science. Great!

9.24.2021 1:49am - (1/ ) Pratt and Schlaifer (1988) was one of the earliest econ. papers rejoicing the discovery of Potential Outcome notation. However, lacking graphs, they made a few basic errors, among them, that treatment-dependent variables can be ignored. (Causality, page 78-9) There is also 1/
9.24.2021 1:49am - (2/ ) a paper by Spirtes etal listing ALL the errors in Pratt and Schlaifer (Not hard to detect once you keep d-separation in mind.) Does anyone have a link to that paper? It will surface in due time.

9.23.2021 8:16pm - (Replying to @stephensenn @_MiguelHernan and @statsepi) This paper is no more about "transportability analysis" than correlation is about causation. But if you are interested in what "transportability" is about, I am sure dozens of readers on this educational channel will be happy to rush to your aid and explain it.

9.23.2021 6:34pm - (Replying to @anceldelambert @deborahlipstadt and 2 others) Sorry for overlooking your question at the time. Someone 'liked' it today, so I read it and noticed that it is still relevant, @RashidaTlaib continues her tantrum attacks on Israel, and her blind followers continue to harass Jewish students on campus. Solid logic.

9.23.2021 7:05am - (Replying to @DeadHansCastorp) @AlJazeera 's parrots know at least seven more atrocity squeaks, just watch @mehdirhasan in action. He too thinks annihilation is just hysteria, and "From the river to the sea" is not an act of "dispossession", just an ancient chant for spiritual catharsis.

9.23.2021 6:04am - (Replying to @InTheLoopBTC) no sorry. But I believe the link was corrected.

9.23.2021 5:46am - (Replying to @DrDavidMathers @RashidaTlaib and @JGreenblattADL) She didn't really have to do that. @RashidaTlaib never promised her voters, nor her donors, to watch for PA violations of human rights. She promised them something totally different, to malign Israel in every possible way, incessantly, and she lives up to her promises admirably.

9.23.2021 3:40am - Thanks for correcting the broken link to the Bayesian Netweork conference

9.23.2021 3:31am - Poor @RashidaTlaib ! Why is everyone jumping on her for being inconsistent on "human rights". Anyone who's obsessed with "Israel Apartheid" has a problem with Israel's birth, not with "rights." @JGreenblattADL ,

9.23.2021 3:06am - Speaking of Bayesian reasoning, here is a virtual event that promotes the use of Bayesian networks for research, analytics, and reasoning.

9.23.2021 2:47am - (Replying to @Perperuna2) Great Bayesian laughs!! It reminds me of my proverb: "Statistical Meta Analysts average apples and oranges to learn properties of bananas"

9.23.2021 12:53am - (Replying to @RashidaTlaib) Question to quantum computing experts: How can you tell when you live in an alternative universe? Ans. When a champion of inhumanity talks "human rights" and a Zionophobic racist talks "apartheid". Can U.S. Congress call for dis-entanglement?

9.22.2021 9:39pm - (1/2) It's now 34-countries boycotting the Durban Conference, but my eyes are still on Norway. Oh Norway, Norway! How could your Gv't face it's people: Sorry, we fell asleep, and found ourselves swimming in the cesspool of civilization. To honor readers whose Governments remained 1/2
9.22.2021 9:39pm - (2/ ) morally sober, I am listing the 34 countries that decided to boycott the Zionophobic Durban Conference: Albania, Australia, Austria, Bulgaria, Canada, Colombia, Croatia, Cyprus, Czech Republic, Dominican Republic, Estonia, France, Georgia, Germany, Greece, Honduras, Hungary, 2/
9.22.2021 9:39pm - (3/3) ,Israel, Italy, North Macedonia, Montenegro, Moldova, Netherlands, New Zealand, Poland, Romania, Serbia, Slovakia, Slovenia, Ukraine, UK, US and Uruguay. 3/3

9.22.2021 6:53pm - Educators, instructors and ordinary Data Science folks. Here is the paper we were all waiting for: How to teach Causal Inference in AI w/o blushing @smueller and A. Forney share their experience teaching Causal AI for undergraduate classes.

9.22.2021 4:28am - Illuminating!! Thanks Paul for posting this rare pre-revolution relic of causal analysis in economics. I wish all our readers take a look at these articles, one by one, and marvel at the progress made in the past 3 decades. I love the phrase: "causality has been hard to define"

9.21.2021 11:48pm - Removing the $1 billion Israel's Iron Dome from the spending bill is a wake-up call to pro-Israel Democrats, many of whom told me that Pelosi's Squad is just a paper tiger, not worth making waves about. They kept silent, facing the consequences today.

9.21.2021 8:27pm @noatishby is half right; this was done for show. But for whom? Not for pleasing the Squad's voters-base, but for blackmailing the American people: We got you by the neck; you want to protect some civilians from rockets? Fine, but give us first what we want.

9.21.2021 7:27pm - (1/ ) I find the statistics and quotations included in @rafalab article on "Academia in Data-Science" to be very valuable. However, the criterion of "extracting value from data" is too vague, eg. Isn't estimating the mean & variance of X valuable? The critical dividing line between 1/
9.21.2021 7:27pm - (2/ ) "modeling data" and "modeling reality" is not given due consideration and, I am afraid, will encourage universities to initiate grandiose "data science" programs void of training in "data interpretation" technology. More on data-interpretation here:

9.21.2021 6:13pm - Only in a morally inverted universe would Pelosi's Squad be considered "pushback from progressives"

9.21.2021 6:00pm - History is still waiting for Norway!

9.21.2021 7:51am - (Replying to @RaniaKassi @gilled34 and 2 others) I'll end this semi-civilized conversation by pointing out what it shows. Experts tell us the ME conflict is "highly complex". It is so very simple: One side says "we we we" the other says "me me me!". One says: "Equally indigenous", the other says: "Savage, colonialist, racist".

9.21.2021 7:31am - (Replying to @yudapearl @RaniaKassi and 3 others) "Dehumanizing the other", a despised and publicly condemned activity by an extreme minority on one side, has become an official cultural and educational policy on the other. And it started much before 1967, even before 1947; recall, I was there and I remember quotes.

9.21.2021 7:16am - (Replying to @RaniaKassi @gilled34 and 2 others) Last time I looked at the mirror I saw a modern & enlightened person, highly knowledgeable in how Israel was formed (I was there) and how it is sustained (miraculously) against all attempts to annihilate it. I've also noticed a glaring asymmetry in the art of dehumanizing others.

9.21.2021 6:58am - Gee, there is so much noise about Al-Aqsa Mosque and the Temple Mt. that I have forgotten completely about St. Mary of Justinian Church. I am surprised the Christian world is not more active in claiming ownership. Perhaps its better that way.

9.21.2021 6:44am - (Replying to @tiagoandresvaz @ch402 and @rkuchen) I feel likewise; never tired of thanking them for what they have given me.

9.21.2021 6:32am - (Replying to @RaniaKassi @gilled34 and 2 others) Your definition of Zionism comes from the Mufti of Jerusalem, Haj Amin al Husseini, who equated Palestinians' aspirations with crushing those of their neighbors. The rest is a century of Palestinian tragedy.

9.21.2021 6:12am - (Replying to @gilled34 @RaniaKassi and 2 others) True, but Edward Said taught us that it is an ugly colonial habit to allow someone else define who you are and what you are, let alone someone who wishes you dead, let alone someone who has no qualm making his intentions public. So whose definition would you buy?

9.21.2021 2:21am - Proud to be on the advisory board of New Zionist Congress @newzionists , and doubly proud to read these two winning essays, shining in wisdom, confidence and hope, exposing Zionophobia in its full immoral ugliness.

9.21.2021 1:42am - (Replying to @RaniaKassi @EinatWilf and @newzionists) I assume you are referring to the ethnic cleansing embodied in the slogan "From the river to the sea"? It's a big one, almost 7 million human beings, mostly refugees or children of refugees. Agree, the slogan is racist to the 1st degree.

9.21.2021 1:18am - (Replying to @TwomeyMaeve) I've felt cold air coming from Irish followers, but I didn't realize things are as bad as you describe. De Valera is probably turning in his grave to see his country turning Talinani.

9.21.2021 1:03am - (Replying to @ConceptualBIG and @brandeiscenter) I am not recommending "safetyism"; on the contrary, I say explicitly "dont wait for safe space, create one!"

9.20.2021 11:28pm - (1/ ) Summarizing our discussion of "demand" via "ceteris paribus" (CP), we've seen that, once formalized, CP amounts to comparing Y under two settings of X, say X=x and X=x', while leaving other variables in the structural equation for Y unchanged. The beauty of formal definitions
9.20.2021 11:28pm - (2/ ) is that they hold for all models and are independent on the meanings of X, Y,Z, etc, or the procedure by which we estimate things. Leveraging these beauties, we come to realize that the resultant CP definition of "demand" is none other but the counterfactual definition of
9.20.2021 11:28pm - (3/ ) Causal Effect, namely {Y(x, Z),Y(x', Z)}, where Z is the set of other variables in the eq. of Y, both observed and unobserved. Thus, the analysis of "demand" can benefit directly from the literature on causal effects, presenting no peculiarities that demand special treatments.
9.20.2021 11:28pm - (4/ ) Of course, the cyclicity of the equations prevent us from leveraging d-separation and do-calculus. But the logic of counterfactuals can still be summoned to advantage, as is done here From this point on, as they say in economics: Ceteris Paribus.

9.20.2021 10:04pm - I wish I could reach those students sampled in the @brandeiscenter study and: (1) assure them that the vast majority of faculty is behind them, (2) explain to them what makes me so proud to be Jewish and continue the tradition of upstream swimmers.

9.20.2021 9:11pm - (Replying to @tiagoandresvaz and @ch402) Best compliment I can imagine.

9.20.2021 3:02pm - This may be relevant to our discussion of "ceteris paribus". Yu Xie just sent me a video of a symposium on CI, with Jim Heckman and myself debating the meaning of Structural Equations and whether economists can learn anything useful from ordinary mortals.

9.20.2021 1:33pm - As we can see from wikipedia, the concept of "ceteris paribus" is slippery, depending on its usage and on how formal you want to be. But eon are lucky, because they start with SEM, which is a mathematical object, in terms of which other concepts can be defined, including CP.

9.20.2021 12:47pm - (Replying to @PhilHaile and @steventberry) The relationship "fully determines" is a better primitive for defining SE (hence "demand) than "all else", but it is still ambiguous, in case we have both X,Z and X,W fully determining Y. I believe you are hinting at "minimal set fully determining Y". Agree?

9.20.2021 7:45am - (Replying to @PhilHaile and @steventberry) Beautiful! Now we see that it is not really "all else held fixed" but only "all other arguments in a certain equation are held fixed". Thus, things boil down to the question what qualifies some variables as arguments in some equations to the exclusion of others. Any criterion?

9.20.2021 7:31am - (Replying to @PhilHaile0 I'd take the risk of conjecturing that the concept of "ceteris paribus", hence "demand", though taught in econ-101 and spoken in Marshal, begs clarification for economists, not only for non-economists. Once clarified, I conjecture, some difficulties will disappear from yr paper.

9.20.2021 7:17am - (Replying to @steventberry and @PhilHaile) You are way ahead of me. I am still trying to translate "ceteris paribus" to some mathematical language. Can we?

9.20.2021 7:11am - (Replying to @PhilHaile) Once we define "demand" formally (you owe me a definition) we can likewise define "the economic questions of interest" formally and, then, there would be no confusion with "average demands" or "unit-based" demands or any other detractors. This is the beauty of "defining".

9.20.2021 7:05am - (Replying to @PhilHaile) Why is it a "good" question and not an affirmative "yes"? If "demand" is definable in structural equations (you still owe me a definition) and SEs are but a collection of PO's, demand must be definable in PO!! ie, no ifs, no buts, but YES.

9.20.2021 6:55am - (Replying to @PhilHaile) Before we get to fancier questions, let's deal with Ceteris paribus. Certainly "all else held fixed" needs some qualification for, if we vary "price", all effects of price cannot remain fixed. Moreover, what kind of "fixing" are we to allow? by intervention or by conditioning?

9.20.2021 6:15am - Fair question. I've compared the "causal revolution" and "credibility revolution" in a blog entry here: Also, Section 4 of, compares the two under the rubric "structuralists" vs "experimentalists".

9.20.2021 4:30am - (1/ ) To all decent students, at UCLA and elsewhere, Jews and non-Jews, who are dreading the hostile campus climate awaiting them with the start of the Fall quarter, September 26, let me assure you that disgraceful statements like the one posted by Zionophobic activists at the UCLA
9.20.2021 4:30am - (2/ ) Asian-American-Studies Department will not remain unexposed. They will continue to be rebuked by concerned members of the faculty, and their authors held accountable for putting our university to shame. To further assist you in your upcoming struggle
9.20.2021 4:30am - (3/ ) (Replying to @yudapearl) for equity and inclusiveness, I am linking to my piece "Two New Weapons" which recommends two steps you can take to restore sensible discourse to campus life. Use them, and sanity will prevail. Apropos, Hillel-AEN statement:

9.20.2021 4:18am - (Replying to @LuisaBorrell) Causal inference provides the tools with which social scientists can formulate and algorithmitize ideas and desiderata. In addition, CI provides the tools for "social intelligence", ie, robots understanding of how we feel and what motivates us.

9.20.2021 3:52am - (Replying to @instrumenthull @analisereal and @PHuenermund) Vice Versa. I would genuinely like to hear takes from the "ceteris paribus crowd" on my innocent questions: (I was going to say "Yale crowd" but, upon recalling my honorary degree from Yale, I decided to be respectful.)

9.20.2021 3:17am - (Replying to @KaiGehring1) I believe the link you need is to this paper, titled:"Trygve Haavelmo and the emergence of Causal Calculus." It's the one that Jim Heckman labored so hard to emulate using "statistical methods". See

9.20.2021 3:01am - (Replying to @Perperuna2) We think about much more than Simpson; like what decision is appropriate in any given problem. But when Simpson strikes, we naturally begin to doubt the data we got or the methodology we use, and we do not know what to blame. That's why Simpson continues to provoke discussions.

9.20.2021 2:55am - (Replying to @sbuhai) I can't decipher what you are getting at. Please excuse my inability to continue.

9.20.2021 1:17am - (Replying to @steventberry and @PhilHaile) You are way ahead of me. I am still trying to translate "ceteris paribus" to some mathematical language. Can we?

9.19.2021 11:52pm - (Replying to @subniten) I believe if Norway joins the civilized world Sweden would not wish to stay behind, with Iran and the Taliban.

9.19.2021 11:43pm - If I were an econ PhD student I would ask: 1. The paper defines "demand" verbally as: "responses of quantities demanded to ceteris paribus changes in prices or other factors." But I do not see a formal definition of "ceteris paribus". Is there one? 2. Can such a definition 2/
9.19.2021 11:43pm - (2/ ) be cast in the language of counterfactuals? (or potential outcomes); I don't see one. 3. If we had such a definition, should'nt we be able to leverage standard methods of counterfactual identification/estimation for estimating "demand"? 4. IOW, a definition would help

9.19.2021 10:10pm - Interesting. Those who have historical/cultural identity do not seek one in their DNA. Can your double helix make up for stories you can't tell, poems you don't chant, holidays to celebrate and heroes to emulate?.

9.19.2021 6:02pm - Romania Makes 20 States Skipping 20th Anniversary of `Antisemitic Hatefest' Durban Conference
9.19.2021 6:02pm - We still need Norway; here is why:

9.19.2021 5:47pm - A reader sent me another article on Simpson's paradox in Covid-19 data: properly attributing the reversal to improper causal interpretation, as in Well done @jsm2334

9.19.2021 4:09pm - This grain of salt inspires two of my own. 1. Recall: CI enables you to wring a solution to the way your mind see the problem, putting an end to "wringing from your mind a problem to fit a solution." 2. It's easy to mystify how the mind works, CI helps demystify it.

9.19.2021 6:35am - Agree. My recommendation applies even more so to PhD students starting #machinelearing. Compared to econ, ML research is held back by two additional handcuffs: (1) Insistence on Tabula Rasa and (2) Insistence on mimicking neural circuitry.

9.19.2021 4:18am - (Replying to @LuisaBorrell) Causal inference provides the tools with which social scientists can formulate and algorithmitize ideas and desiderata. In addition, CI provides the tools for "social intelligence", ie, robots understanding of how we feel and what motivates us.

9.18.2021 9:54pm - Unbelievable News!!! No blood shed, No resistance, No martyrs, None of @4noura 's glorification of convicted killers, just simple, persistent and consistent policy: "No one sought remains uncaught." Hats off to all Israeli personnel who worked 24/7 through the high holidays.

9.18.2021 8:36pm - My recommendation for a PhD in epidemiology are here: It's important to emphasize that, unlike many economic professors, most Epi professors know and use the Two Fundamental Laws of causal inference. Big difference.

9.18.2021 7:39pm - (Replying to @LuisaBorrell) Yes, same recommendation. Except that the confusions in Epi education are of different nature, and less severe. Some Epis are hung up on "no causation w/o manipulation" and some resist the distinction between Rung 2 and 3 of the Ladder. But they know the 2 Fundamental Laws of CI.

9.18.2021 3:46pm - (Replying to @stephensenn) If every statistician was so irritated by asymptotic analysis, the symbol P(*) (or pr(*) , or f(*) for density) would be eliminated from stat books and papers. But last I checked, they were still there, guiding thinking and facilitating analyses. Should we purge them out?

9.18.2021 1:04pm - Asymptotic analysis is merely a first step in causal analysis, necessary to layout the research question you have, express your assumptions and derive the estimand -- the thing you want estimated. Causal analysts are eager to learn ways of skipping this stage if possible. Is it?

9.18.2021 6:01am - Most statisticians I know are philosophers of science who want to ground ideas about knowledge, beliefs and evidence in real life problems. Trouble is, the language of stat prevents them from handling all but shadows of knowledge. Some snap out of this confinement, most don't.

9.18.2021 5:43am - For econ PhD's I recommend: #Bookofwhy, then Primer, then *Do not touch any econ text before you read and solve the toy problems presented there* . You are now ready to do meaningful PhD research in the age of causation.

9.18.2021 3:32am - This is how Data Science students are going to feel in five years, despite all the key-note speakers and panelists today who agree on how important causal reasoning is in Data Science. Educators must start by placing the role of DL in the Ladder of Causation.

9.18.2021 1:58am - Eventually, it might be safe to teach stat before causality, w/o danger of confusion. But given today's confusion, especially among stat instructors, it is safer to start Data Science education with causality, as in Primer, then treat stat as a specialty.

9.17.2021 7:02pm - (Replying to @ErnestSDavis) Most American kids have only one glimpse at Jews - their annual school visit to a Museum of Holocaust. The amount of resources invested in those museums exceeds many times any other channel of Jewish-gentile communication. It should not be confined to crucifixion alone.

9.17.2021 6:50pm - (Replying to @DKedmey) I am still at that stage, aren't we all? And we will never feel satisfied until, perhaps, the day when we can program a computer to give us the answers.

9.17.2021 5:26am - (Replying to @jaredlander and @RabbiCosgrove) The pinnacle of Jewish renewal before the Holocaust took place in mandatory Palestine, 1917-47, a heroic period which is almost untouched on in American Jewish education, and which incubated the creation of Israel in 1948.

9.17.2021 4:00am - (Replying to @yudapearl) I think that's @RabbiCosgrove 's point as well. That we should also learn about life, both before and after, the Holocaust, in addition to all the death.

9.17.2021 3:45am - (Replying to @ilyaeck) One difference, the secular elements in the consensus happens to be in clear majority in Israel. I've just listened to a panel discussion on Chanel 13; not one of the panelists (all secular) has expressed any reservation to this unspoken consensus.

9.17.2021 2:58am - A bird's eye view of Israel's deserted streets on Yom Kippur
9.17.2021 2:58am - Amazing phenomena! This is the only 25 hour highway shutdown that is not imposed by any authority or civil ordinance, but by consensual agreement alone.

9.17.2021 12:39am - Christianity would never have become a world religion had the Gospel narrative stopped at the crucifixion rather than the resurrection. Holocaust museums, through which many Americans get their only glimpse of Jews, allocate zero floor-space to Israel - the Jewish resurrection.

9.16.2021 10:48pm - Great quotes from: "People Love Dead Jews". Which also explain why so many people, even intellectuals like Noam Chomsky, can't stomach the sight of a successful, normal and permanent Israel -- it spoils their cosmic script of where Jews belong, especially the "normal" part.

9.16.2021 10:14pm - Our sages forbade us from engaging in any pleasure producing activity on Yom Kippur, but I am sure even the holiest of priests wouldn't resist this pleasure - riding a bike on the main highway of Tel Aviv on Yom Kippur, your kids beside you. Hamavdil Bein Kodesh L'chol!!!

9.16.2021 7:19pm - Yes, I am scheduled to speak tomorrow morning, but I believe it is 10:30am PT, not ET! Night owls do not wake up that early.

9.15.2021 1:19pm - (1/ ) As I am preparing for Yom Kippur (Day of Attonment), tonight, I would like to share this prayer with you, written about one thousand years ago and sung by members of Kibbutz Geva, in memory of the Israeli soldiers fallen in the Yom Kippur war (1973)
9.15.2021 1:19pm - (2/ ) I lost two good friends in that terrible war - YZB. In the spirit of the Day, I am asking for forgiveness. In case I have offended any of you, be it by sloppy mention, or by careless omission, let me know (by email) so that I can reconfigure my software not to repeat. Thanks.

9.15.2021 1:57am - I will never get tired waiting for Norway. This is your chance Norway to choose between Knut Hamsun the humanist writer (whose books I read in Hebrew while growing up) and Knut Hamsun the blind and gullible Nazi sympathizer of the 1940's. Choose, Norway, Choose!

9.15.2021 12:47am - Only 4 days left for submitting your new insights to the WHY-21 workshop on CI & ML, see announcement at

9.14.2021 10:33pm - (1/ ) Good news from Burlington: The BDS Resolution was Defeated. But I hope the good people of Burlington understand that passing that resolution was never the intent of BDS activists. Their victory lies in using your City Hall as a platform from which to accuse Israel of 1/
9.14.2021 10:33pm - (2/ ) some crime (was it child pornography?) and having your neighbors see Israel on the accused bench. That is all your neighbors will remember from the incident and that is all they wanted and succeeded. To prevent a repeat, you must expose their tactics.

9.14.2021 7:30pm - "Causal inference happening *inside* the NN" should be a great incentive for DL folks to study causal inference, abandon the holy grail of Tabula Rasa, accept extra-data causal knowledge, and start reaping the benefits of explainability, domain shift, mediation analysis, & more.

9.14.2021 6:53pm - Palestinian Human Rights Watch: We should hold Zionism accountable for this child's shirt. @4noura

9.14.2021 6:34pm - We are waiting for Norway; so fair, so rational, so peace-loving, almost like the music of Greig and the Oslo Accord. This is your litmus test, Norway!!!

9.14.2021 6:25pm - Don't forget the other side of the coin, Zionism also means sanity.

9.14.2021 6:17pm - (Replying to @EinatWilf and @Georgetown) I have suffered this age anxiety for the past 52 years and, believe me, it gets worse with age.

9.14.2021 2:13am - Hats off to Mayor @MiroBTV for issuing a morally clear statement, not so much for opposing a BDS resolution (most Sapiens do) but for telling Burlington residents Why (most Mayors don't).

9.14.2021 11:17am - (Replying to @BaccarelliAA @MariaGlymour and 3 others) I believe NSF has the same ceiling, so the place to change it is perhaps pressure on US Congress by the National Academy of Sciences.

9.14.2021 2:13am - Hats off to Mayor @MiroBTV for issuing a morally clear statement, not so much for opposing a BDS resolution (most Sapiens do) but for telling Burlington residents Why (most Mayors don't).

9.14.2021 1:49am - (Replying to @GnomeSortingHat) Your question should be addressed to Stanford's people who coined the name and proclaimed it "a sweeping paradigm shift in AI". But I am intrigued by the notion of "general knowledge priors" What would be a classical example of such "priors"?

9.14.2021 12:14am - While most observers of Stanford's "Foundation Model" (FM) are puzzled by the title 'Foundation', I'd like to focus on the word 'model' and ask: "In what way it acts as a 'model'", namely, does FM tell us things about the world that we would otherwise be wondering about?

9.13.2021 12:58pm - Glad to see our Tweet quoted in this article on the shaky foundations of Stanford's "Foundation Model" which, for me, is still a "what is it?" puzzle, not whether it's successful or not. The article also discusses what deserves the title "foundation," a topic I'd like to pursue.

9.13.2021 4:10am - Indeed, Noura Erakat @4noura was a panelist in that shameful Rutger+SFSU+UCLA event that dishonored the memory of 9/11 victims,

9.13.2021 1:55am - @4noura 's fabricated victimhood and Orwellian rhetoric have metastasized in an academic bubble of Zionophobic racists, precisely because no one has ever alerted her to the racist ugliness of Zionophobia. I hereby do!

9.13.2021 1:22am - (Replying to @BryceBesler and @halflings) Save for size, I do not see a fundamental difference between the regression methods of 1920 and deep nets of 2020. Pierson's methods were both admired and revolutionary in the 1920.

9.13.2021 12:28am - (1/ ) I could see my birthday balloons still floating in mid air when Twitter handed me another gift: The number of followers on this educational channel has reached 50,000. I am grateful to you for trusting the way I am laboring to make the causal revolution normative, and the
9.13.2021 12:28am - (2/ ) way I am desperately trying to make sense of a world gone mad. Thank you for keeping me young, honest and alert through your challenging questions, as well as through your reassuring silence of approval. BTW, @claudiofantacci is our 50,000th reader. Welcome, Claudio!

9.12.2021 11:38pm - (1/ ) Entitled "Top-N Recommendation with Counterfactual User Preference Simulation", this paper caught my attention thinking it provides a generalization of the Unit Selection problem: It deals indeed with the counterfactual
9.12.2021 11:38pm - (2/ ) question: "what would be the user's decision if the recommended items had been different?". It seems to me, however, that the paper does not incorporate observational studies and, thus, generalizes the A/B method.

9.12.2021 2:44pm - (Replying to @eliasbareinboim) Thanks for explaining the gains. In particular, I am intrigued by the facility to easily generate compatible scenarios.

9.12.2021 1:31pm - (Replying to @eliasbareinboim and @kerstingAIML) You almost answered by my question. Alas, after the sentence "In summary ...." I have expected a "summary", and can't find one. Do you gain efficiency? Transparency? Explainability? Any other feature of merit?

9.12.2021 6:51am - (Replying to @DavidHarrisAJC) 20 years later - what has changed?

9.12.2021 6:44am - (Replying to @EinatWilf) 20 years later. What has changed?

9.12.2021 6:37am - (Replying to @AJCGlobal and @shawfrank) 20 years later - what has changed?

9.12.2021 6:33am - 20 years later, what has changed? In the months following 9/11/01, Jewish leaders championed efforts to combat Islamophobia. 20 years later, pro-Palestinian professors rebuke UCLA Chancellor for condemning antisemitism w/o taking their side on Gaza:

9.12.2021 3:20am - (1/ ) This paper deserves attention for deriving new theoretical connections between GNNs and SCMs. Moreover, it allows causal information to be provided the natural way, i.e., via DAGs, then translated to GNN. What is not clear to this reader, though it may be implicit in the text,
9.12.2021 3:20am - (2/ ) is: "What is gained by this translation?" Namely, why not conduct the analysis in the SCM framework and let GNN take over the estimation part only, as is done here An explicit characterization of the gain would be helpful.

9.12.2021 12:41am - (Replying to @ChrSzegedy and @artistexyz) In what way is "training on a large amount of data..." a "phonomenon" ?. In other words, why would a sequence of operations be called a "phenomenon"?

9.11.2021 11:10pm - (Replying to @HjalmarLucius) National election? While the entire world is eyeing Norway and expecting her to exercise moral leadership by walking out of the Durban conference? Its no time for national election!!

9.11.2021 10:41pm - Exhausted, out of food: How 4 fugitives were caught, with help of Arab Israelis via @timesofisrael
9.11.2021 10:41pm - "Caught with the help of Arab Israelis" - a devastating blow to Hamas' plan and an affirmation of the axiom: "No one sought remains uncaught". Some suggest @RashidaTlaib should declare a "day of rage" in Detroit, Michigan.

9.11.2021 10:13pm - (Replying to @artistexyz and @ChrSzegedy) You are talking about Stanford's Foundation Model (FM) as if it was a "model". I am still trying to understand in what way it differ from: "Do whatever works and call it FM".

9.11.2021 2:27am - Human Rights Legal Group Alleges That 9/11 Webinar Could Put SFSU, Rutgers in Violation of Anti-Terror Laws via @jewishjournal
9.11.2021 2:27am - I do not know how readers from Rutgers and SFSU feel, but I am ashamed that UCLA is involved in moderating this desecration of the memory of 9/11 victims. I wish they would use another date to demonize America and Israel.

9.11.2021 1:04am - (1/ ) This paper tells me that I was wrong in dismissing twin-networks as no-longer useful once you establish conditional ignorability. It shows them to be computational instruments, especially useful in Bayesian analysis. The paper will also be useful for readers working on
9.11.2021 1:04am - (2/ ) 1/ This paper tells me that I was wrong in dismissing twin-networks as no-longer useful once you establish conditional ignorability. It shows them to be computational instruments, especially useful in Bayesian analysis. The paper will also be useful for readers working on

9.10.2021 10:56pm - (Replying to @EvelynKleid) You are right!!! It's reported on Al Jazeera -- the last outlet to report on any setback to terrorist organizations. This is an extremely important development toward sustaining Israel's deterrence axiom: No one sought remains uncaught, even if it takes years. Awesome!

9.10.2021 10:14am - (1/ ) (Replying to @irinarish and @desai_pratik) Causality is certainly not the only way to describe reality, and science up to 1920 is a proof that non-causal descriptions exist and can be quite successful, up to a point. The causal revolution has marked that point and has enabled us to answer questions about interventions
9.10.2021 10:21am - (2/ ) (Replying to @yudapearl @irinarish and @desai_pratik) and counterfactuals (eg explanations). Any other formalism that can answer such questions should be considered an alternative; but I do not know any. The proof is in the pudding - answering certain questions - not in how you call it.

9.10.2021 4:40am - (1/) If you think Israel scapegoating was invented by the Squad, read what Robert Fisk wrote 9/12/01, a day after the 9/11 attack: "The attacks on Tuesday come at the end of a long history of betrayal of the Islamic and Arab peoples: the collapse of the Ottoman Empire, the
9.10.2021 4:40am - (2/) foundation of the state of Israel, four Arab-Israeli wars and 34 years of Israel's brutal occupation of an Arab nation". Robert Fisk was a chief ME journalist for The Independent (UK), an extremely influential Zionophobe. Even the collapse of the Ottoman Empire was US fault.

9.10.2021 4:35am - Police capture two of the six escaped prisoners in northern town of Nazereth via @timesofisrael
9.10.2021 4:35am - I'm so glad to have good news with me this Friday night to discuss with my family, ending a week that was especially gloomy and depressing. I hope you share my joy and hopes that the other 4 criminals will soon be apprehended with no loss of life.

9.10.2021 4:24am - Any Norwegians among our readers? I can't get my mind off Norway, so rational, so peace-loving, so neutral. I am hoping -- Norway!!!

9.10.2021 3:15am - (Replying to @4noura @nookyelur and 3 others) This framing errs by labeling Zionism a "solution for antisemitism" instead of what it in fact is: an "organic yearning of a people for normalcy". This error has costed the Palestinian people its normalcy. Hoping it's not too late.

9.10.2021 12:13am - A glimpse of love, ending a week of Talibans.

9.10.2021 12:06am - (Replying to @alwynkotzee @CMc_PhD and @HumaneTech_) Better conjectures is one way. Another is better methodology of leveraging existing conjectures towards learning and experimentation.

9.9.2021 11:53pm - (Replying to @EinatWilf) I had a similar question on the perennial "Day of Rage". How can you have a day of "RAGE" in the middle of 73 years of "WAR"? Is it a day of "Shouting" instead of "Shooting"?

9.9.2021 5:43pm - (Replying to @SlezakPeter) I was hoping to hear: "..and other crimes fabricated by BDS cronies." Or, at the very least: "... and as long as these non-silent Jews stop demanding "normalcy" "self-determination" and other disrespectful ideas unbefitting a good Jew.

9.9.2021 5:03pm - (Replying to @holdmytowel) I have this dream that: (1) DL folks are super-smart fellows, so (2) as soon as they see a causal model at work, they will revolt en masse and abandon curve-fitting forever. Perhaps that's why their professors try so hard to hide causal modeling away from them. Just a dream.

9.9.2021 4:23pm - I heard a fitting corollary to this Algerian judoka story. Kindergarten teacher: "Listen kids! Whoever cries for no reason, gets a reason to cry!"

9.9.2021 3:19pm - Oh, speaking of the Spanish Edition, don't miss the preface I wrote especially for this edition: where I dared express my firm belief that the current data-fitting direction taken by "Data Science" is temporary (read my lips!).

9.9.2021 2:04pm - (Replying to @annareitman) Thanks for renewing my respect for Uganda.

9.9.2021 2:02pm - I got a copy too. Great book!!

9.9.2021 2:00pm - (Replying to @brunofmr and @eliasbareinboim) Epicycle models are beyond reach of DL theology, for they are based on a man-made assumption of objects going around each other, as opposed to holes poked in the sky shell.

9.9.2021 1:52pm - (Replying to @DKedmey and @DavidDeutschOxf) Go tell a Taliban leader "you have a Pattern!" -- he would look at you with blank eyes: "What?". But try telling him "you are a Zionophobe" -- he would jump to the ceiling with pride of accomplishment and joy of "born again".

9.9.2021 1:41pm - (Replying to @desai_pratik and @irinarish) It seems that the impossibility of having both "no assumptions" and " perceiving the world as it "really is" (eg. a society of cause-effect relationships) has not occurred to Buddhist Scholars.

9.9.2021 7:07am - Joining @Eve_Barlow in this New Year resolution. Partly because of what it says, and partly because I love resolutions. Like counterfactuals, resolutions are unique to humans, revealing an awareness of redeemable weaknesses in one's software.

9.9.2021 6:51am - I am still waiting for Norway to make this move. Don't know why I am fixed on Norway, perhaps because last time I visited Oslo we discussed the Middle East and the economists there impressed me as decent people. Norway! I want Norway!!!

9.9.2021 6:42am - I disagree. The Pattern expressed by Taliban is not 'antisemitism' but Zionophobia. The same Pattern unveiled by UCLA department of Asian-American-Studies, whose members (my own colleagues) probably love Jews, as long as they are stateless, powerless, and at the mercy of others.

9.9.2021 6:21am - I am told Uganda never forgave Israel for that savage violation of Ugandan pride.

9.9.2021 6:14am - (Replying to @VC31415) It ain't new. As the #Bookofwhy describes, Smallpox vaccination was blamed for killing more people than smallpox.

9.9.2021 5:46am - (Replying to @ACT_Perspective) Emotions may be in the same category as intelligence: Only those who have it can fake it.

9.9.2021 5:00am - (1/ ) "increase in funding" does not begin to describe the stupendous amounts of resources society has and is investing in model-blind prediction. Some DL researchers have not seen a "model" in their lifetime and one wonders whether the recent interest some data-scientists show in
9.9.2021 5:00am - (2/ ) (Replying to @yudapearl) causal-reasoning will ever bear fruits, given the exclusively model-blind education that most universities offer under the rubric "data science". If I were a company like Google, Microsoft or Amazon, I would worry.

9.9.2021 2:43am - (Replying to @Abel_TorresM) Understanding the survival value of these facilities would lead to understanding their computational value in robotic societies, including the social interaction between the robot and its users.

9.9.2021 12:41am - (Replying to @gottfriedmath) I bought it on the basis of your description, though the fact that the words "artificial intelligence" appear only once in the text, which warns me that the authors have not had experience in dissecting a real life "reasoner".

9.8.2021 11:37am - Given that our mind shows an excellent symbiosis between two urges: (1) to explain and (2) to predict, the questions arise how we transfer this symbiosis to a robot, and what methodology we should use to manage conversations between the two. Surprise suggestion: Causal Inference.
9.8.2021 11:37am - (2/ ) (Replying to @yudapearl) Indeed, take any exercise in causal inference (eg and you will find an analytical effort to decompose a given task into two types of subtasks: (1) Those based on explanatory models (eg DAGs), (2) Those based on model-free predictions (eg, DL).

9.8.2021 3:10pm - (Replying to @rfarmer27 and @amandavande1) Following habit, I've scanned through the articles recommended in hope of finding nuggets of wisdom me or my students can use. Disappointedly, they seem to have been written prior to the domestication of causal thinking. These problems would be posed & treated differently today.

9.8.2021 10:48am - (Replying to @irinarish) The innate urge to explain accounts for children's curiosity, attention and learning speed, as well as scientists drive to come up with "pivotal" new experiments. This drive should not, in any way, compromise our ability to manage unexplained phenomenon, as the Babylonians did.

9.8.2021 10:29am - (Replying to @zacharylipton @tdietterich and @danieldennett) I dont know where you find things like "dump on" "idiot" "write off" "look bad" etc. ? Is @danieldennett guilty, or a victim of these bad manners? Am I? Are you? Beyond me.

9.8.2021 10:08am - (Replying to @si_tulli and @Looking_glass_u) Nice, even I could understand it!

9.8.2021 9:34am - (Replying to @SilvioZaina) There is a difference between hypothesis being "data-compatible" vs. "data-based". The germ theory of disease is "data-compatible" yet not "data-inspired". Ignaz Semmelweis (1818-1865) had the data yet could not come up with a plausible hypothesis to convince his contemporaries.

9.8.2021 1:42am - Read my lips. The "winners" will be machines that learn about the world around us, not about the data they happened to receive. The formers will not need to rewrite history but to learn from it, as I predicted in "Science vs. Data"

9.8.2021 12:53am - I am retweeting because, again, the last line sounds punchy: "Science w/o conjectures would not be science." It aims to refute claims that model-free (ie, data-centric) scientists are somehow less biased, hence more successful than the model-guided scientists we all know.

9.8.2021 12:33am - (Replying to @irinarish) I prefer robots that communicate with us at the level of Newtonian physics to quantum robots who can't explain things: why pumps get hot when compressed. As to curve-fitting, see Greek-Babylonian rivalry here: Science w/o conjectures would not be science.

9.8.2021 12:03am - (Replying to @zaffama and @irinarish) Agree with the statement "uniform priors" = strong form of bias, especially given that the former is "language sensitive". But the equation "model-free" = "seeing the world as it is" has another objection: "model-free" means "not seeing the world" but seeing its warped shadow.

9.7.2021 8:05pm - Retweeting, because the last line came out cute and profoundly true.

9.7.2021 8:01pm - (Replying to @irinarish) First, Bayes metaphors do not work for Causality for reasons shown here: Second, a theory of causal modeling is needed for 3 reasons explained here: Finally, if Aristotle were Buddhist scientists would still be fitting curves.

9.7.2021 7:23pm - (Replying to @AllesistKode and @KordingLab) Domain shift without distributional shift is demonstration here: My blog is here:

9.7.2021 7:17p - (Replying to @tlbtlbtlb @JohanDH2O and 2 others) This may be true but, now that you have both distribution shift and counterfactual shift, repairing the former without the latter lures you away from repairing the root cause of shift.

9.7.2021 7:11p - (Replying to @gottfriedmath @desai_pratik and @bjorn_hogberg) I can't resonate with these statements. Information theory is probabilistic hence orthogonal to causality.

9.7.2021 2:08am - (Replying to @holdmytowel) The good people of ML do not deny your claim. What they do deny, and understandably so, is that to incorporate causal thinking in machine learning the field needs to paradigm-shift itself.

9.7.2021 1:48am - Re-reading Gopnik etal's classic paper: A Theory of Causal Learning in Children I am finding several compelling arguments against the Tabula Rasa theory that has mesmerized the DL enterprise. This was still prior to the domestication of counterfactuals.

9.7.2021 1:15am - Psychologist of reasoning have long adapted graphical models (CBN) as an arena for testing theories of learning and argumentation. The paper: surveys these efforts. Strange that counterfactuals have not had a greater impact on this branch of psychology.

9.6.2021 2:48pm - Tonight we celebrate Rosh Hashana - the Jewish New Year. To all readers who celebrate, or join in celebration, here is my favorite Rosh Hashana song: Over the honey and the bee sting Please guard us my Good Lord For the lyrics see:

9.6.2021 1:20pm - My point is that I do not take it as personal, but as part of a deeply entrenched ideology that says: We don't need to see the needle, we can discover it tabula rasa. Moreover, the less we see of it, the less biased would the one we discover be. See :

9.6.2021 12:59pm - (Replying to @tdietterich @amanda_gentzel and 2 others) When you are conducting a conversation with a colleague and need to quickly decide what to assume about his/her background and understanding, statistical correlation is a very good predictor, it was actually developed for such predictions.

9.6.2021 12:47pm - I did not take it as "unfair" but, rather, as missed opportunity to tell DL readers what needle they are searching for in the haystack of representations. I therefore suggested the paper be read together with where the needle is vividly described.

9.6.2021 12:26pm - (Replying to @amanda_gentzel @Sergei_Imaging and 2 others) Terminology unveils the speaker's culture and, very often, the speaker's understanding as well.

9.6.2021 6:57am - (Replying to @JohanDH2O @zacharylipton and @tdietterich) True. But since "counterfactual shifts" may also occur without "distribution shift" the use of the latter may reveal lack of appreciation for the pivotal importance of the former.

9.6.2021 6:46am - (Replying to @zacharylipton and @tdietterich) When we move from the population of LA to that of NYC we change not merely the distribution but also all the counterfactuals that characterize the units in those two populations. The expression "distribution-shift" warns me that the speaker is likely unaware of the difference.

9.6.2021 3:47am - There is nothing wrong with probabilistic thinking. The fallacy is in the assumption that probabilistic reasoning is sufficient for acting and understanding. Few today advocate sufficiency explicitly, but many act as though sufficiency is axiomatic.

9.6.2021 2:35am - (Replying to @desai_pratik and @bjorn_hogberg) Simplicity is a language-dependent concept, so it plays only a minor role in explainability, see Intuition is a more powerful force, and it is driven by analogies with the physical world in which our early experience is shaped, encoded and compiled.

9.6.2021 2:05am - There is no such a thing as "system-1" or "system-2" tasks, as implied by BLH here: A task solved using a system-2 process by one reasoner may be solved using a system-1 process by another. This is what knowledge compilation is about.

9.6.2021 12:58am - (Replying to @zacharylipton and @tdietterich) The litmus test separating the serious people from the rest is that they do not use the term "distribution shift" but surrogates such as "environment shift" or "population shift"; they know that the latter shifts can occur without the former.

9.6.2021 12:41am - The quest for explanation goes back to the Greek astronomers and distinguished them from their Babylonian rivals. See for how the Greeks sacrificed accuracy for explainability. Oddly, science sided with them.

9.5.2021 9:52pm - (Replying to @jwbelmon @charleswangb and @tdietterich) Causality, like reality, has been there all along, it does not "come in".

9.5.2021 9:41pm - (Replying to @tdietterich) In medicine, even when we can only treat the symptom, we know that it's only a symptom and we have words such as "disease" to remind us that something deeper might be going on. I am searching for a DL word that is deeper than "distribution".

9.5.2021 7:06pm - (Replying to @tdietterich) They also say that headache is "the first clue" for Covid-19, but they change the name after a few tests, to make sure other treatments are used beside aspirin.

9.5.2021 2:52pm - The idea that "knowledge" is more than "distribution" has not caught on yet in DL circles and, consequently, we continue to see statistical vocabulary dominate the writings of DL authors. Expressions such as "distribution shift" and "out of distribution" will soon be replaced.

9.5.2021 2:46pm - This paragraph labels "theoretical pluralism" *inevitable* yet my problem is to explain scientific consensus, not pluralism. Why is it that, at any given time, only 2-3 theories are competing for attention out of the Zillions that are theoretically compatible with empirical data?

9.5.2021 2:13pm - Thanks for posting this CACM paper which, IMO, should be read together with another CACM paper:, because some of the "can't yet do" in the former are actually done and operationalized in the latter.

9.5.2021 12:31pm - (Replying to @bjorn_hogberg) And this brings us to the question of consensus of explanation: Why is it that, at any given time in history, the concept of "good explanation" garners general agreement.

9.5.2021 6:19am - (Replying to @rlmcelreath) How does one view the slides?

9.5.2021 5:35am - Wishing you a fun voyage; let us know about your safe arrival. Oh, don't forget to invite enlightened colleagues to attend, stat instructors, ML enthusiasts, etc. They will be forever grateful to you.

9.5.2021 5:19am - (1/ ) (Replying to @lewbel @JaapAbbring and 5 others) I am claiming that (1) our ability to distinguish the plausible from the implausible is heightened when assumptions are qualitative, rather than "shape restricting" and (2) many of the so called "identifying assumptions" found in the literature are made, indeed, to facilitate
9.5.2021 5:27am - (2/ ) (Replying to @yudapearl @lewbel and 6 others) identification (as the title implies), not because they are deemed to hold true in the world or to be scientifically plausible. Agree: we need theorems that extend SCM identification theory to incorporate plausible shape restrictions. The linear case is well covered, next =?Replying to @yudapearl @lewbel and 6 others) identification (as the title implies), not because they are deemed to hold true in the world or to be scientifically plausible. Agree: we need theorems that extend SCM identification theory to incorporate plausible shape restrictions. The linear case is well covered, next =?

9.5.2021 1:11am - (Replying to @lewbel @JaapAbbring and 5 others) SCM methodology is of help to those who posit assumptions because they are deemed scientifically plausible, not because they facilitate identification.

9.4.2021 1:18pm - (Replying to @Zehut1981) Anachnu Po

9.4.2021 12:34pm - (Replying to @Gemma_DLC) The definition I was hoping for would be crisp enough to allow one robot to tell another "You seem not to KNOW XYZ, are you aware of ZYX?"

9.4.2021 12:24pm - (Replying to @LiamMFox1 @Gemma_DLC and @DavidDeutschOxf) "Information with causal power" is a definition that I am prepared to accept, though I prefer to be more specific about what the "power" is, to screen away imposters and to operationalize the definition.

9.4.2021 3:40am - (Replying to @Gemma_DLC) What is the definition of knowledge?

9.4.2021 2:30am - (1/ ) The Balloons on my screen remind me that, yes, I'm 85 yrs young today and, yes, I'm fortunate to be able to continue defending commonsense reasoning through AI, continue defending the miracle called Israel, and continue growing with my children and grandchildren despite the
9.4.2021 2:30am - (2/ ) (Replying to @yudapearl) painful hurdles of modern life. I thank all my Twitter readers (almost 50K by now) for keeping me young, honest and alert through their challenging questions, their occasional "likes" and "dislikes" and, mostly, through the sounding trumpets of the causal revolution.

9.3.2021 9:17pm - Glorious hymns to counterfactuals were sung both before and after their demystification by structural models. This paper provides an encyclopedic mixture of both: But it leaves readers with the difficult task of separating mystifiers from demystifiers.

9.3.2021 2:35pm - (Replying to @metamnisia) Not necessarily. When we do induction, we are given the outcome y and we fit the input x as a function x=f(y), representing E(X|y). In Bayesian networks it is called "the most likely explanation" though there is no causation involved, just probabilities.

9.3.2021 12:55pm - (Replying to @DrBobGoldberg and @PHuenermund) Especially on Twitter, where things said yesterday are "yesterday news."

9.3.2021 12:53pm - (Replying to @RonKenett @Lester_Domes and 4 others) For me, the Ladder of Causation serves as a measuring tape for the "boundary of meaning". Are you proposing a more refined tape?

9.3.2021 1:50am - If "Data Science" was truly data science (as defined eg here, you wouldn't need to add "Causal" ahead of the title. But, given what it is today, this is an effective way of telling students: "This is not another function-fitting class."

9.2.2021 8:20pm - She does not tell us what this "teacher" tried to undo to her daughter, but I truly feel for her. I dread the day when my grandchild would come home to parrot BDS slogans. California education is sliding that slope, see what the LA Teacher Union is up to:

9.2.2021 4:35pm - Remember the words Bennett was saying to Biden My daughter just sent me a recording of my chanting these Biblical words in ancient Hebrew, exactly as I did in my Bar Mitzva, in 1949: "Arise, shine, for your light has come..."

9.2.2021 3:00pm - (Replying to @YusufSohoye and @PhilMickelson) The ice-cream is from the #Bookofwhy but the idea of confounding is from the Bible (e.g. Book of Daniel) and, not playing golf, so I could not follow the rest of the argument. Glad causality hit the golf course.

9.2.2021 6:24am - (Replying to @joe_sill) Agree. We are facing a 2-dimentional no-free-lunch theorem.

9.2.2021 12:22am - (Replying to @DKedmey) Can't stop imagining our Chancellor sending a message: `a person who has an issue with an Israeli flag being raised for one week here, after a UCLA department criminalized Zionist students of "colonialism" should look for a new job.' And with that, the BDS circus at UCLA ended.

9.2.2021 12:05am - (Replying to @dn1164) Good point! To remind us that it's dreams that make history, not facts. Still, I remember, Kafka was there too.

9.1.2021 7:46pm - (VIDEO 2m30s) NYTimes - Inside Look: Kabul's Airport After Taliban Takeover
9.1.2021 7:46pm - (Replying to @yudapearl) It reminds me how I woke up in Rome, the morning after Attila's takeover, yes, Attila the Hun. Kafka wrote a story about that morning, he was there too, but I forgot the title.

9.1.2021 7:05pm - (Replying to @Grady_Booch and @SFMagus) I han't seen it yet and, in general, I evaluate principles, not codes. For this I need to go to the citation provided which, on the surface, appears good, though I can't find the First & Second Laws of causal inference, nor the Ladder, so I'm uneasy.

9.1.2021 5:01pm - What I said about "ML = Tabula Rasa" applies more strongly to unsupervised learning, where the representation itself is sought to be learned "from the data alone". #ML.

9.1.2021 5:01pm - (Replying to @joe_sill) What I said about "ML = Tabula Rasa" applies more strongly to unsupervised learning, where the representation itself is sought to be learned "from the data alone". #ML.

9.1.2021 4:07pm - Tabula Rasa is the soul of ML, and what makes it so exciting to so many. 99.99% papers in the ML literature are Tabula Rasa, and success means: "Look! we can compute it from the data." If we define ML as "any study that processes data using a machine", then who is not in ML?

9.1.2021 2:30pm - If "machine learning" (ML) is defined as "tabula rasa", then it cannot "integrate" man-made causal information, or it would cease to be ML. If we seek "integration", we have it already, it's called "Causal Inference", which has always deployed ML in its estimation phase.

9.1.2021 12:01pm - So, what made my day? What offset the Taliban takeover of Kabul? None comes close to my reading about editors like Axel Springer CEO Mathias Döpfner, who told his employees: "Newspapers, too, have principles". See

9.1.2021 8:00am - (Replying to @seanjtaylor) I would like to emulate your strategy, and am wondering if you recommend that I'll spend more time estimating something before making sure it's what we want estimated? Or spend more time on "design" = a set of decisions driven by an experimenter's intuition. Seriously, should I?

9.1.2021 7:38am - Fall semester looms with BDS a key campus issue. Ben & Jerry's boycott faces pushback with BDS movement silent on Taliban takeover.
9.1.2021 7:38am - (Replying to @yudapearl) This report, by professor Alex Jaffe of Scholars for Peace in the Middle East (especially the links provided), gives a panoramic view of how higher education struggles against BDS racism -- not always successfully. My admiration to the brave students and faculty at the front line

9.1.2021 5:44am - I view it slightly differently. Causal Inference need to evolve to integrate and leverage the capabilities of machine learning. Why? Because CI provides us with the road map of the inference engine and can tell us which capabilities are needed in different stages of the process.

9.1.2021 5:14am - This article is written as if countefactuals were never algorithmicized and the #Bookofwhy never published. Try to convince the writer to listen to this educational channel.

9.1.2021 3:01am - Would love to hear how you "focus on finding/implementing randomizations", namely, how you "find randomizations" w/o a causal model in your head. Also, how/why you find reasoning with the model in your head to be harder than with something which is not in your head, but "clean".

9.1.2021 2:31am - (Replying to @quantumciaran @gileadamit and @TheEconomist) Glad @gileadamit understood that causality blind program has no chance of fooling the Nobel Prize Committee, not even in 2036.

9.1.2021 2:20am - Who doesn't remember his/her first day in school? I surely do, and these kids in Israel make my memory so much more vivid, perhaps because I know exactly what the teacher is about to tell them, the tone of her/his voice, and the colorful stories they are about to hear. Proud of U

9.1.2021 12:36am - Congratulations, @GalitPeleg , on your new and super important appointment to head @MASHAVisrael . In the name of all my comrades here, in the trenches, I wish to pledge our unreserved commitment to serve your cause, our cause. Yishar Kochech!

9.1.2021 12:21am - Good to know that the RCT estimand plays a special role in estimation as well. It remains to find out whether the "lack of guarantees" for other estimands reflects rare pathological cases, or prevailing obstacles. Non-differentiable functions do not make calculus useless.

9.1.2021 12:03am - Congratulations to Juan D. Correa (@ Bareinboim's group, Columbia U) for his outstanding PhD defense earlier today. Awaiting his final dissertation to learn more about soft interventions, "transfer learning" and selection bias:,

8.31.2021 7:11pm - (Replying to @tribelaw) Well put: "safety of potential future victims aren't theirs to give away." Echoing my sentiments here: concerning the millions of would-be killers emboldened by this Parole.

8.31.2021 5:42am - (Replying to @AvrilKD and @machine_ml) Why? Tell us more.

8.31.2021 3:36am - It's a Historical Day, Mr. Ambassador. When you touch the ground of Israel, you touch the heart of every peace loving person on this planet.

8.31.2021 2:43am - (1/ ) Extremely valuable study, thanks; a must read for every discipline on this educational channel. Just wish to note that published articles, whose authors fear peers and reviewers, may not be representative of how researchers truly think. Cafeteria talk may be more revealing
8.31.2021 2:43am - (2/ ) (Replying to @yudapearl) as I tried to exemplify in the Appendix of this article, how causal utterances are attributed to non-manipulative causes. @NoahHaber , @Lester_Domes , @_MiguelHernan , @MariaGlymour .

8.31.2021 12:29am - As soon as I cited this paper and started re-reading the introduction, a question bugled my mind: How come @machine_ml folks dealing with "transfer learning" "robustness" "invariance" etc. are not addressing these issues as directly as this paper does?

8.31.2021 12:04am - The simplest problem I've seen which deals with combining two data sources is a chain X->Y->Z: How best to combine samples from P(y|x) and P(z|y) to estimate P(z|x)? It was fun to analyze asymptotically Can your results handle finite samples? @shantanug7

8.30.2021 10:45pm - Honored to introduce to @ThePearlPost , the students newspaper of the Daniel Pearl Magnet High School, from whom I hope to receive insightful stories written from students perspectives. Adriana @adrianachavira , the journalism teacher, deserves much credit for the Post excellence.

8.30.2021 8:00pm - Rashida Tlaib Calls for Israel to Release Bodies of Slain Palestinians, Doesn't Mention Reported Terror Attempts via @jewishjournal
8.30.2021 8:00pm - (Replying to @yudapearl) Poor @Rashida , everyone is picking on her as if she had a normal education and is responsible for what she is saying. Of course she wouldn't mention the motives of a terrorist, why the puzzlement? Zionophobes are programmed for deceit, not for facts.

8.30.2021 7:12pm - Made my day: Ancient coin to take to the stars with Israeli astronaut via @JNS_org

8.30.2021 6:58pm - (Replying to @itamarcaspi) broken link. It seems that Angrist-style has done more damage than I thought.

8.30.2021 6:26pm - (Replying to @sbuhai) Is this what we should take to be the "indispensable" principle of economics: "No universal recipe will pick a model for you" ? That's all? I'm sure the new generation of economists can come out with more instructional principles.

8.30.2021 5:42pm - `All Israel Prayed For Him': Soldier Shot at Gaza Border Dies From Injuries After Nine-Day Battle
8.30.2021 5:42pm - (Replying to @yudapearl) I prayed for him for 11 days. For me he represented a country caught between the need to protect its life, and obeying "no live bullets" orders.

8.30.2021 3:18pm - (Replying to @artistexyz @MatheusFacure and 2 others) In general, "causing something is not a binary property", agree. This is not the case, however, for potential outcomes, {Y(0),Y(1)}, which are DEFINED as "all variations in Y occurring while we holding D fixed." Such variations, the diagram shows, cannot be caused by D.

8.30.2021 6:34am - (Replying to @zacharylipton) I assume the estimands you are dealing with are quite different than those used in the standard on-line estimation literature.

8.30.2021 4:49am - (Replying to @VC31415 @artistexyz and @causalinf) I have an even more minor point: Let's not call ignorability or unconfoundedness "CIA" = "conditional independence assumption", a name reserved for CI among ordinary variables, not counterfactuals - it's a totally different beast.

8.30.2021 4:41am - (Replying to @NazalKaradan) Still, understanding it does not make it less funny, less hypocritical or less racist.

8.30.2021 4:35am - (Replying to @MatheusFacure @artistexyz and 2 others) You are right, (Y(1), Y(0)) stands for all exogenous variables that affect Y through paths avoiding D. See Causality page 342.

8.30.2021 4:17am - Funny thing to watch: Women who find the veil "liberating" and "empowering" (eg, Linda Sarsour) also try to exclude Israeli feminists from the Women March. Some call it "ultra-liberation"

8.30.2021 3:46am - Trying to provide a better link: or

8.30.2021 3:31am - For readers interested in "explainable AI", especially those ready to share my voyage through the high waters of "artificial understanding," I recommend this paper: "How Understanding People Differs from Understanding the Natural World",

8.30.2021 1:49am - (Replying to @sbuhai) What principles or tools do you consider "indispensable" in your work?

8.30.2021 1:17am - (Replying to @sbuhai and @yudapearl) That, again, does not mean that it isn't useful--or that for specific cases, even ideal--but its yet to be shown (burden of proof falling on developper(s)) that it's pragmatic--before even *indispensable*!--in realistic app where existing tools in structural econ proved their worth

8.30.2021 12:58am - (Replying to @sbuhai) Let's not go into the zillions upon zillions of past mis-Tweets. If you want to discuss the "indispensability" of SCM, I am ready, because I know many economists who are still saying: "useful, yes, indispensable, no"

8.29.2021 11:25pm - (1/ ) The title of this paper caught my attention "Efficient Online Estimation of Causal Effects by Deciding What to Observe" because the idea that we should decide which data source to sample next has not occurred to me while working on "data fusion". On
8.29.2021 11:25pm - (2/ ) (Replying to @yudapearl) the other hands, the causal nature of the problem introduces no new nuance, so I assume traditional statistical technique of "on line estimation" could be borrowed to handle this problem. Standing to be corrected.

8.29.2021 10:50pm - (Replying to @artistexyz @VC31415 and @causalinf) I haven't kept track of this entire thread, still let me just interject that: 1. CIA is model dependent. 2. SUTVA has two parts: 2.1 non-interference and 2.2 consistency. 2.1 is an assumption needed for both PO and SCM, but 2.2 is a theorem, not assumption in SCM. I haven't kept track of this entire thread, still let me just interject that: 1. CIA is model dependent. 2. SUTVA has two parts: 2.1 non-interference and 2.2 consistency. 2.1 is an assumption needed for both PO and SCM, but 2.2 is a theorem, not assumption in SCM.

8.29.2021 4:33pm - Can we summarize this discussion by me admitting that: "Good grad econ progs (many in Europe, btw) do both->you (ie, I) may have oversampled the first type..". And, if possible, you admitting that "econ education should place greater emphasis on The First Law (& 2nd Law).
8.29.2021 6:55pm - (Replying to @yudapearl) If we can continue our conciliatory discussion, I would like very much to go into the *indispensability* of 'SCM-', relative to 'Cowles Comm-' style structural reasoning, and learn from you if I have missed anything in answering "What Kept the CC at Bay"

8.29.2021 4:01pm - (Replying to @yudapearl and @PhilHaile) Thought those linked *threads* were more general than just on the DAG aspect (I certainly didnt intend either to isolate only the DAG dimension), but there were many others discussion threads, e.g: Etc.

8.29.2021 3:31pm - (Replying to @sbuhai and @PhilHaile) DAGs were never the issue in my mind; answering important research questions was and is. eg. when is an effect estimable by regression? eg. How can we recover from "selection bias"? eg. How can we detect selection bias in RCT? eg. eg. .... "Agreement" on DAGs is/was irrelevant.

8.29.2021 2:35pm - (Replying to @EmreAkanak) There is! I haven't heard the Taliban threatening to destroy a neighboring state. They might, in time.

8.29.2021 2:24pm - (Replying to @PhilHaile) Agree. *Some/most* economists prefer to go from regression to causation. This preference was not present in the days of the Cowles Commission. But count the proportion of econ papers today asking when something "has causal interpretation".

8.29.2021 1:52pm - These dancing-prancing youngsters seem to be celebrating rather than "protesting" and, to them, the whole "border clash" seems more like an orgy. I wish Israeli youngsters could feel that happy and secured 20 meters from their neighbors' border fence.

8.29.2021 4:10am - (1/ ) Indeed, this productive period in the history of econ prompted me to ask: "What Kept the Cowles Commission at Bay?" in my ET paper The answers I gave also explain the tiny percentage of econ PhD's who can solve the toy problems I posed in that paper.
8.29.2021 4:10am - (2/ ) (Replying to @yudapearl) 2/ Among those "toy problems" we find the question that started this thread: When is a regression coefficient equal the causal effect it is meant to represent?". Modern economists can answer it in one sentence, while the old guards would lecture you to avoid toy problems.

8.28.2021 11:28pm - (1/ ) RFK Killer Granted Parole via @jewishjournal Influenced by the recent parole of my son's killers, in Pakistan, I submit that the prosecutor is abrogating his duty by not considering the message his decision sends to the thousands, perhaps millions of
8.28.2021 11:28pm - (2/ ) (Replying to @yudapearl) would-be killers, currently awaiting a license to vent their anger, or serve one of their "noble causes" by taking another innocent life.

8.28.2021 7:58pm - (Replying to @Axlamdax and @Grady_Booch) This was Russell, not Hume. Hume went counterfactuals, as narrated here: and especially in #Bookofwhy pp.264-5.

8.28.2021 5:44pm - (Replying to @Grady_Booch) The fascinating part of this paper is that it was written by someone (Hoover) who was not part of the causal revolution. We are anxiously awaiting an account of "History of causality in economics" written by a young Turk.

8.28.2021 7:30pm - (Replying to @artistexyz @VC31415 and @causalinf) Wow! Do you want to start WW-III ??

8.28.2021 7:23pm - Israel border with Gaza is on fire again. But these mothers and sisters will not be shown on NPR & NYT They represent the family of Barel Shulei, shot in the head while adhering to Israeli policy of using live fire only in cases of immediate mortal danger

8.28.2021 6:24pm - Sharing a wonderful gift received from the students at the Daniel Pearl Magnet High School, who wrote this about my beloved wife, Ruth: Thank you buddies!
In the link, if you click on "gallery", there are some endearing pictures

8.28.2021 5:44pm - (Replying to @Grady_Booch) The fascinating part of this paper is that it was written by someone (Hoover) who was not part of the causal revolution. We are anxiously awaiting an account of "History of causality in economics" written by a young Turk.

8.28.2021 5:26pm - Perpetuating the "most familiar" ends up conflating the world with the tools we build to interrogate it. The relationships you wish to observe are defined by your research question which, invariably, looks beyond regression.

8.28.2021 5:04pm - But this regression obsession is changing rapidly by young economists -- join the rebels.

8.28.2021 4:17pm - (1/ ) I've been thinking, why is it that economists prefer to go from regression to causation and not the other way around? I believe it goes back to the 1960-80's when an influx of statisticians flooded econ., for whom the world begins and ends with regression, the rest is frills.
8.28.2021 4:17pm - (2/ ) (Replying to @yudapearl) In contrast, for modern causal inference researchers the world begins and end with the processes that generate the data, the rest is just epi-phenomenon. For support of this theory, see Kevin Hoover's famous lament: "Lost Causes" or "Where have all the causes gone?"

8.28.2021 1:45pm - This is a very interesting way of posing the problem. Normally, in CI, we ask: "When would a causal effect be estimable by regression?" This paper goes the other way and asks: "When would a regression coefficient have causal interpretation?" I bet economists would love it!

8.28.2021 1:37pm - A better link to "Regression and Causation: A Critical Examination of Six Econometrics Textbooks"

8.28.2021 1:03pm - If you're an economist, then you probably mean this article, which examines how six Econometrics Textbooks conflate regression with structural coefficients. We still haven't seen a correct interpretation in any econ. textbook; please advise if you have.

8.27.2021 9:49pm - Amazing! The words that Bennett recited to Biden in their meeting today: are taken from Isaiah:60, which happened to be my Bar Mitzva's Haftara: "Your sons and daughters will come back from afar..." and which I can still chant like it (1949) was yesterday.

8.27.2021 2:57pm - A very disturbing article about Quora: a Q&A platform that we often reference on this channel. Since racists are super-motivated, their influence on such platforms is many times their size. Please watch out & report when you smell their presence.

8.27.2021 1:37pm - (Replying to @tkaiser_science) Not too challenging, if you are interested in applying it at the right level of abstraction, namely, the level at which psychotherapists communicate with each other.

8.27.2021 1:21pm - Thanks for posting. I believe Prof. Malik's views of "Foundation Models" reflect deep bewilderment, similar to the one that haunted us in our discussions. Glad to learn I am not the only one who is not critiquing FM but simply asking: What is it?

8.27.2021 4:58am - (Replying to @desai_pratik) Don't expect a definition, nor description. We have been trying to constructs one on this channel unsuccessfully. It appears to be one of those mysteries that will only be deciphered by future historians.

8.27.2021 4:50am - A wild guess: Afghanistan will not pull out, nor will Norway. The former out of conviction, the later out of spinelessness.

8.27.2021 4:37am - As a citizen of both countries, Ambassador Erdan speaks FOR me and TO me.

8.27.2021 1:10am - Americans should be aware of this new fund-raising industry, run by Taliban sympathizers, pretending to assist Afghani asylum-seekers. A good pre-cautionary measure: search Google for "American war crimes" under the solicitor's name. "Israel war crime" is also a good handle.

8.27.2021 12:37am - (Replying to @tdietterich) Confessing ignorance: What is LLM? I know other enterprises that "sucks oxygen" by overpromises, but not sure we mean the same ones.

8.26.2021 8:20pm - (Replying to @luckychucky85) Very good article on double machine learning, thanks for posting

8.26.2021 8:05pm - (Replying to @artistexyz and @cmnit) I didn't make the to Asimov. Perhaps we should ask FM users what associations the F meant to invoke.

8.26.2021 6:26pm - A warm welcome to a new paper on "bad controls" for economists and other rebels. One question: This is the first time I hear about "double machine learning"; who coined the name? and what feature makes it worthy of the title "double", as opposed to "single"?

8.26.2021 3:55pm - This still leaves us with an unresolved puzzle: What makes learning with "deep pockets" produce better performance than "shallow pockets"? We've ruled out "bigger data". So what remains? Could it be selection bias? i.e., run many more studies, select the successful, call them FM.

8.26.2021 2:55pm - (Replying to @KordingLab and @BruceTedesco) Only g*d could hand us an actual distribution, true, but every baby scientist since Bernoulli can figure out what performance can be expected when the number of samples goes to infinity, based on the concept of "distribution", before even having it.

8.26.2021 2:48pm - Multi-Congratulations to @bariweiss upon winning the LA Press Club Daniel Pearl Award for Courage and Integrity in Journalism. Bari adds a colorful new angle to what "Integrity in Journalism" means to journalists in the jungles of the 2020's.

8.26.2021 1:54pm - (Replying to @KordingLab and @BruceTedesco) My models are trained on more data than any FM can afford -- infinite data. I get it easily by assuming that we know the actual distribution P(*) behind the data. Am I using FMs ??? Are performance guarantees/limitations proven for my models applicable to Stanford FMs?

8.26.2021 7:15am - (Replying to @KordingLab) I begin to sense a definition of FM in what you are saying: FM is a learning algorithm that is trained in a unique way. It is not trained "de Novo" but "non de Novo". Can we elaborate on how "non de Novo" differs from "de Novo"?

8.26.2021 5:02am - (Replying to @cmnit) Before you can ask "why FM are so effective" you got to define (or at least describe) what an FM is, because if FM is defined by its successes you will be asking "why are successful days in Las Vegas more successful than bad days"

8.26.2021 4:58am - (Replying to @desai_pratik) Don't expect a definition, nor description. We have been trying to constructs one on this channel unsuccessfully. It appears to be one of those mysteries that will only be deciphered by future historians.

8.26.2021 4:31am - (1/ ) Today I've received a reply from Percy Liang, director of Stanford's Center for "foundation models" (FM) and, as promised, I'm reporting what I learned. The short answer is: they know it works surprisingly well empirically, but still don't know the mathematical principles.
8.26.2021 4:31am - (2/ ) (Replying to @yudapearl) I clarified that I'm not concerned about WHEN the model would be successful, but with "what is it?" Namely, Can we tell an FM from a fake FM? is FM defined by its successes? by its final structure? or by the training procedure it undergoes? I'll report back if/when informed.

8.26.2021 1:47am - I've just enjoyed listening to a thoughtful conversation with Stuart Russel on "what makes AI human compatible", which included an insightful perspective of the history of AI, as well as the current tensions between DL and GAI. Highly recommended.

8.26.2021 12:41am - (Replying to @holdmytowel @tarantoga3 and 3 others) understanding the procedures I concocted to understand the world is not understanding the world.

8.25.2021 10:52pm - Good question @JGreenblattADL , did you have a chance to look at this chapter of the curriculum: I got depressed when I read it. Imagine my grandchildren learning that their grandpa is a "right-wing" "settler colonialist" and an "oppressor of Palestinians".

8.25.2021 10:14pm - Depressing what they are doing to California education. You need to read the chapter on to gauge the immensity of irreversible damage inflicted upon our children, if this curriculum is allowed to be implemented.

8.25.2021 6:46pm - URGENT ACTION - Antisemitic Chutzpah of 'Liberated' Curriculum Exposed, Contact Jewish Caucus:
8.25.2021 6:46pm - (Replying to @yudapearl) What's happening now in California public schools will not stay in California but will corrupt the minds of your children wherever you live. Unless you take action and nip it the bud today, before the appropriation committee approves it. Racism to a first degree.

8.25.2021 6:26pm - I'd like to call readers attention to a new article posted on Stanford Encyclopedia of Philosophy, titled "Understanding", by Stephen Grimm. Why? Because now that we have a computational model of understanding (ie, CSM) we will soon understand what it is.

8.25.2021 2:23pm - (1/ ) If I was Bennett, I would never say: "There would not be resolution of the conflict for the foreseeable future" as he did here: Though it is true. I would say:"Peace? Of course! As soon as they say the words: 'equally indigenous'". Any risk saying it?
8.25.2021 2:23pm - (2/ ) (Replying to @yudapearl) 2/ i.e., Any risk the Palestinian leadership could pretend to agree with the 'equally indigenous' mantra and demand a sovereign state in return? No way! They can't utter those words anatomically, for their entire identity will collapse. There is ZERO risk for Bennett. Moreover,
8.25.2021 2:30pm - (3/ ) Moreover, just in case they utter it by a fluke of the tongue, or equivalent (in Arabic, or course), so much the better: a real peace may actually become a possibility, and Bennett would become a national hero.

8.25.2021 1:12am - One additional comment: My last sentence says: How CI can benefit from this enhanced mode of learning, which goes opposite to the motto of Benjo's talk: How ML can incorporate CI. Reason: CI already has the conceptual framework for causal thinking; ML may enhance, not replace it.

8.25.2021 1:00am - Speaking about AI and Machine Learning, good news comes from my alma mater, the Technion, which was ranked # 1 in Europe. Too bad only two of the Center's faculty are doing CI. Just imagine how they would rank if that numbers was 30?

8.25.2021 12:27am - (Replying to @heinz_r @UlrichJunker and @GaryMarcus) Thanks for posting. Illuminating talk, though there were only a few slides I truly understood. I have this weakness of absorbing mathematical theorems better than description of systems. I'll have to read the papers to see how CI can benefit from this enhanced mode of learning.

8.24.2021 7:37pm - Somehow, Squad members would never miss an opportunity to blame Israel for whatever their voters hate, making them look stupid, racist, and vacuous like the facts behind their accusations.

8.24.2021 7:18pm - (Replying to @havivrettiggur) Haviv, I stole some of your words Please forgive. I'll pay back.

8.24.2021 7:15pm - Someday, in an unexpected flash of insight, Cory Bush is going to notice how the words "Israel's apartheid" make her look stupid, boring and vacuous, like the facts behind those words. She will then switch to blaming Israel for Covid-19, which her voters could swallow, I'm sure.

8.24.2021 5:11pm - I agree that we should keep causality out of the equation. Still, I can't see the distinction between the "black" and "white" neural network. Is there a simple (3-4 variables) litmus test that Stanford's "foundation model" can pass and a "non-foundation model" cannot? ?????????

8.24.2021 5:00pm - Naive me believed we no longer have "unenlightened persons", at least not in public. Where do you find them?

8.24.2021 4:51pm - (Replying to @WahbaRachel @philipdogwater and @Eve_Barlow) Please use their family names: Zionophobes

8.24.2021 1:07pm - (Replying to @gottfriedmath) Confession: I enjoy re-reading it too, especially when I feel depressed (rare occasions) and I need historical perspective to cheer me up. It's really mind bugling how scientists have been struggling all those years with no language for causation.

8.24.2021 6:52am - What do you know! Google Scholar just whispered that my book "Probabilistic Reasoning: Networks of Plausible Inference", published 1988, has garnered 30,000 citations. A mighty challenge for #Bookofwhy to top next 30 years.

8.24.2021 5:18am - (1/ ) Here is a thoughtful new paper on Savage's Sure Thing Principle (STP) First, every paper on STP must be thoughtful; how else can one deal with deeply entrenched intuition that does not follow from logic? Second, it carefully analyzes the
8.24.2021 5:18am - (2/ ) (Replying to @yudapearl) many previous attempts to ground the STP on more basic principles, including my humble attempt Finally, it seems to extend the analysis to non-causal settings, though I fail to see how one can talk about "action" & "preference on actions" w/o causation.

8.23.2021 10:47pm - We are obsessed with our history because it is an empowering story, mighty worth getting obsessed with. I wish the world stops trying to make us forget it, we can't afford to breed more history-blank pathologies like Peter Beinart and JVP.

8.23.2021 2:42pm - (Replying to @holdmytowel @tarantoga3 and 3 others) That's what I've noticed, and was afraid to ask, not wishing to discourage the hundreds (literally) upon hundreds of PhD's working on data-centric explainability. Not to mention the dozens upon dozens of Gothic Cathedrals built to it by respectable universities.

8.23.2021 1:58pm - (Replying to @artistexyz @VC31415 and @causalinf) Sorry, I can't explain this type of "explainability". What I am missing is input-output descriptions and clear distinctions between causal and predictive notions ("contributes to" "responsible for" "regression" etc.) which do not mix.

8.23.2021 12:51pm - (Replying to @dbenyamin) It's only the tip. Now, that machine learning folks have shown some interest in belief propagation and causal thinking, we should see 300,000 citations in no time.

8.23.2021 12:47pm - (Replying to @caglar_onal) I agree with the ecumenical spirit of your post, but let's not conflate genetic lineage with history. The latter resides in the mind and governs our everyday thoughts.

8.23.2021 11:51am - "Israelis fleeing Palestine" - I've heard these words since 1947 (eg Azam Pasha) and each time I ask myself: "Why can't Palestinians understand that Israelis cannot "flee" from 3,000 years of history?" My answer: How can they? Theirs is only 100 yrs young, how can they?

8.23.2021 5:07am - What happens when farmers expect foxes to respect chicken rights, including female chicks, and post their expectations in the village square. Poor foxes.

8.23.2021 2:26am - (Replying to @maths_gee @seanjtaylor and @PHuenermund) There is a way, but it involves interventions, as noted by Woodward. However, Woodward's paragraph conflates "independence", "manipulation" and the physical law.

8.22.2021 11:26pm - (Replying to @Foreman1David and @FJnyc) Yes. Camera has a list of omissions and distortions here: I'm primarily fascinated by CNN's technique of showing the Palestinian side through the eyes of women and children and the Israeli side through soldiers and politicians - their babies have not eyes.

8.22.2021 11:11pm - (Replying to @DonnaRacheli and @LogicallyAI) This is what his brothers told David, when he showed them the sling and the five pebbles he intended to use against Goliath. In the end they were right, but it was David's slingshot that kept them alive to say: See, we told you!

8.22.2021 8:43pm - Well, this week's episode of CNN's series on Jerusalem was not as bad as last week's, see Sadly, the producer could not end it w/o giving the Palestinians a chance to spoil their cause again, by confessing (via JVP & BDS) their wet dream: end to Israel.

8.22.2021 6:03pm - (Replying to @CAMERAorg @CNN and @dmlitman) The CNN series on Jerusalem is disgusting not because it is one-sided but because the Israeli side describes honestly how it felt living those events while the other side decries the evil INTENTIONS Israelis must have had, still have, and must be punished for having. Disgusting

8.22.2021 4:39pm - (Replying to @soboleffspaces) Keen observation. This is indeed the first time that I see the term "invisibly" used in a technical report. I wish UCLA would allow me to start a Center around an "invisible model." Any connection to the invisible weavers of Hans Christian Anderson is purely linguistic.

8.22.2021 4:18pm - To remind readers of our first reaction to Stanford's "Center of Foundation Models", here is what I tweeted then (aug 4): I have hoped one of our readers would have a friend at Stanford to confide with and unveil to us why they are so excited about it.

8.22.2021 2:30pm - Shaked: Yamina to quit government if Lapid advances Palestinian state
8.22.2021 2:30pm - (Replying to @yudapearl) What happens if Biden asks Bennett to support a two-state resolution to the conflict? Bennet should answer: "Of course, as soon as they accept Israel, and tell it to their children" thus giving him 100 years to prepare.

8.22.2021 12:15pm - I sure did and, unfortunately, the questions I have remain hanging, the main one being: "Substance or hype?". I've written to Percy Liang, the Center's Director to suggest someone who can explain it to us in a principled-based vocabulary. Will report back if/when educated.

8.22.2021 12:11pm - "When we hear about GPT-3 or BERT, we're drawn to their ability to generate text, code, and images, but more fundamentally and invisibly, these models are radically changing how AI systems will be built." - CRFM Director Percy Liang via @StanfordHAI
8.22.2021 12:11pm - (Replying to @yudapearl) For readers who are still waiting anxiously to hear what "Foundation Models" are about, substance or hype, I haven't received an answer yet to my queries. But I've found this quote in today's email, which might give our readers a hint while exploring: "Substance or hype?"

8.22.2021 11:09am - I am flattered by your tweet, but it reminds me that my son, Daniel, when he copy edited my 1988 book "Probabilistic Reasoning", told me: Dad, you better take a class in basic journalistic writing; your sentences must be "clear and succinct." I'm still trying to heed his advice.

8.22.2021 2:25am - (1/ ) This new paper deals with the "nearest neighbor" (NN) and "causal modeling" (CM) interpretations of counterfactuals. I have remarked on several occasions that the former suffers from a fundamental problem of "representation". So, if there is a clash
8.22.2021 2:25am - (2/ ) (Replying to @yudapearl) between the two, I would scrap the former and go with the latter. Philosophers have found such a clash and this paper takes the bold step of scrapping (or modifying) the NN semantics in favor of causal modeling (ie our First Law). It is a happy day for philosophers and CI.

8.22.2021 1:25am - (Replying to @dpovey1 and @LogicallyAI) If it deters the next @GhostEzra from destroying as many lives as this guy did, then it adds positive color to the morality of AI systems.

8.22.2021 12:46am - One of the somber effects of commemorating annual tributes to Victims of Terrorism is for me and my family to view the murder of our son Daniel in the context of this global phenomenon, and to gauge our personal tragedy by progress made in fighting it -Dismal negative progress.

8.22.2021 12:02am - If @LogicallyAI deploys AI technology, then I am proud to be part of that technology. Their success in identifying #GhostEzra sheds a new positive light on the morality of personal identification systems.

8.21.2021 7:03pm - (Replying to @abritishjourno and @HinduHate) I am not familiar with the author mentioned, but I am not at all surprised that Rutgers may be insensitive to gaslighting a religious minority. Their spineless Chancellor had to apologize for uttering the word "antisemitism" w/o bribing the BDS cronies:

8.21.2021 6:10pm - (1/ ) I here provide two links that should help readers who recently joined us to catch up with ideas and conversation we have had in the past. (1) takes you to a searchable file listing all my 8,720 Tweets in the past 3 years. (2)
8.21.2021 6:10pm - (2/ ) (Replying to @yudapearl) takes you to another searchable file, listing all 508 papers and technical reports produced by the UCLA's Cognitive System Lab since its inception. Some have become classics, and some are gold nuggets lingering in mismatched journals, awaiting your attention. Enjoy.
8.21.2021 10:05pm - (3/ ) (Replying to @yudapearl) The 3rd link I can provide would be of interest to readers who wish to understand the latest attacks on Israel. It takes you to the opeds I wrote the past 20 years: where I explain why those attacks are morally despicale, not for being antisemitic, but
8.21.2021 10:05pm - (4/ ) (Replying to @yudapearl) exercising & inflaming a worse form of racism - Zionophobia - the fanatic denial of one people's right to self determination. Enjoy, explore, retweet, speak out and join the resistance.

8.21.2021 4:56pm - (Replying to @artistexyz) The use of causal DAGs is a notational way of distinguishing explainable from non-explainable systems. The key difference is the use of causal, extra-data information in some form.

8.21.2021 3:31pm - (Replying to @soboleffspaces) Good point, "effect modification" is another mode of explanation that enjoys formal definition, nuanced by the way the modifier is modified, be it by action or by selection.

8.21.2021 1:35pm - While Explainability may be too broad a concept to be formally defined, many modes of explanation do enjoy formal definitions E.g.,, necessary and sufficient causation, mediation and Causes of Effects

8.21.2021 5:47am - While debating Explainable AI, the logic of causation and the logic of curve-fitting I came across this joke about a professor of logic:, thought it would be relevant.

8.21.2021 3:51am - Lest the Simpson's Variation below is misconstrued as an indictment of Explainable AI, recall that Simpson's reversal disappears when explanations are generated by causal logic rather than curve-fitting habits. See,

8.21.2021 2:32am - @DavidHarrisAJC , I never pass through Rome without a visit to Arco di Tito the unassailable proof of a sovereign Jewish homeland 1951 years ago, ironically erected to mark the end of that sovereignty. An AJC video on Arco di Tito would be timely in 2021.

8.21.2021 1:41am - Simpson's Variation on a theme in Explainable AI:
User: Why wasn't my loan approved?
Syst: B/c you are a male.
User: What if I were a female.
Syst: It wouldn't get approved either.
User: So who gets a loan.
Syst: Applicants w/ unknown sex.
User: Come on?
Syst: I'm AI-explainable

8.20.2021 8:46pm - @Montreal_AI Correcting a link. An edited transcript of my talk at Montreal AI Debate 2 is posted here

8.20.2021 5:45pm - Israel 'prints' world's first 3D living malignant brain tumor
8.20.2021 5:45pm - (Replying to @yudapearl) Here is a video clips describing the implications and the team:

8.20.2021 2:21pm - (1/ ) (Replying to @pbleic @tylermenz and 5 others) There is another reason why viewing Simpson's paradox as a "special case of OVB" may be misleading. The words "special case" connotes a promise of finding the answer in the general case. However, to find the correct decision in Simpson's context requires causal analysis,
8.20.2021 2:32pm - (2/ ) (Replying to @yudapearl @pbleic and 6 others) of the story behind the data, something the literature on OVB does not provide, lacking graphical language. The pivotal role of graphs in deciding the correct decision is demonstrated in our Crash Course on Good and Bad Control:

8.20.2021 2:10pm - We are concerned for the safety of Daniel Pearl Fellow for 2019

8.20.2021 9:42am - (Replying to @MichaelALewis10) These two are typical examples of omitted and included bias, respectively. But the point is that, to determine which data gives the correct answer, the aggregated or disaggregated, we need to conduct causal analysis, something the literature on OVB does not provide.

8.20.2021 9:34am - (Replying to @chr1sa) Sorry, but I do not understand the question, perhaps because I am not familiar with any "traditional scientific notation" suitable for handling causal relationships. Hints = ?

8.20.2021 9:25am - I would not say that Simpson's paradox is a special case of "omitted variable bias". Why? Because bias may come from either "omitted variable" or "included variable", and the data does not tell us which, until we conduct causal analysis on the story. See

8.20.2021 2:19am - It has been 30 days since I lost my beloved wife, Ruth, my light, my guide and my comfort. In her memory, I am posting the song "Try not to Cry" which expresses my thoughts and my promise to her: I will try. Song by 14 yrs old Uziya Tzadok.

8.20.2021 1:59am - (1/ ) This new paper, from the DoWhy group of Microsoft Research would be of interest to ML folks who wish to understand the role of extra-data assumptions in causal inference, as well to CI folks who take seriously the prospects of automatic generation of
8.20.2021 1:59am - (2/ ) (Replying to @yudapearl) priority list of potentially useful assumptions that a user may feel comfortable defending on scientific grounds.

8.19.2021 5:51pm - (Replying to @seanjtaylor @analisereal and @MatheusFacure) I was/am merely asking for example of such "other representation" and its "usefulness". No, there is a limit to the subjectivity of comprehensibility. Certain modes of representation are universally more comprehensible that others.

8.19.2021 5:51pm - (Replying to @seanjtaylor @analisereal and @MatheusFacure)

8.19.2021 5:03pm - (Replying to @VC31415 @instrumenthull and 6 others) Would love to get the link

8.19.2021 4:53pm - Pablo's quote (below) is indicative of the dismissive/exclusive attitude that has dominated the "quasi-experimental approach" in economics since 1995, and is still pursued in some academic circles (see we now see signs of econometric renaissance.

8.19.2021 4:28pm - (Replying to @instrumenthull @eliasbareinboim and 5 others) I have not seen simultaneity or shape restriction being excluded from the structural framework (SCM). Have you? See Causality p.215-17 for simultaneity, See pp. 389-91 for shape restriction (Linearity). Only insecure disciplines are engaged in excluding new tools and notation.

8.19.2021 4:14pm - (Replying to @MatheusFacure @analisereal and 3 others) I am reading this discussion to be about whether DAG-less PO can be an insightful "alternative" to DAG+PO. IF we all agree that it is an impossibility, we are all in agreement, and there is no need for PR. Are we?

8.19.2021 4:04pm - (Replying to @grecostatistics @alexpghayes and 3 others) Your perception of the field is not perceived that way by some ppl who keep on arguing "who needs this extra baggage, we can do it all with Bayesian stat if only we define a PO variable Y_x and assume that we can express assumptions with such variables."

8.19.2021 3:50pm - (Replying to @pablogerbas @instrumenthull and 4 others) Moreover, "In our work (...)" includes everything under the sun: social science, health science, statistics, biostat. etc.

8.19.2021 3:42pm - Being humbly constructive, no snarky, is showing that DAG-less PO cannot be an "alternative perspective" to DAG+PO, hence one cannot argue for DAG-less PO in the name of "multiple perspectives".

8.19.2021 1:30pm - If I may be of assistance here, the reason statistics texts fail to ask the "right questions" is because most "right questions" are causal, and standard texts fail to introduce vocabulary for asking causal questions, not to mention tools for answering them.

8.19.2021 2:03pm - (Replying to @holdmytowel and @BEBischof) Don't blame them. They couldn't ask causal questions b/c they didn't have a language to do so. Who you should blame? The stat professors who in 2021 still say: "We do not need special vocabulary, it is all Bayesian statistics". They condemn Stat to another century of stagnation.

8.19.2021 1:51pm - Good question. May I suggest starting with: The go to the links provided, following your curiousity.

8.19.2021 1:22pm - (Replying to @MatheusFacure) I am yet to see an example of a "perspective not visible at first" that is "unlocked" by a language different than structural models. Why? because all "other perspectives" that I know are derivatives of structural models. But I am dying to see an example if you can share.

8.19.2021 1:13pm - The mantra "both answers are right, but to different questions" was statisticians' old trick of avoiding causal questions. Today we need to understand why it is that NO DRUG can be bad for a man, bad for a woman and good for a person. There are no "two right answers" here.

8.19.2021 12:11pm - Thank you #SIM for dedicating your Special Issue this month to the memory of my beloved wife Ruth, whose light, warmth and wisdom still guide every neural activation in my brain. May that it guides our planet towards a better future. YZB.

8.19.2021 11:52am - @economeager , I am gratified to know that the causal interpretation of Simpson's paradox has found traction in LSE. In my blog post: "Simpson's Paradox: The riddle that would not die." I ask: what is it that prevents ppl from accepting this interpretation?

8.19.2021 11:10am - When I read: "graphical representation of causality is my weak spot" I hear someone saying: "I can't see through the fog, let me put on my eyeglasses". Causal inference starts where knowledge resides, not with knowledge pre-cooked by an extra-terrestrial shef.

8.19.2021 4:00am - I love your Deming quote: "Without data you're just another person with an opinion." To which I would add: "And with data only you're just another data-scientist with no opinion."

8.19.2021 3:43am - (Replying to @ML_Theorist) Not sure we need 2,3,4, but 5 is extremely important (ie, "actual causation".) The reason I question the necessity of (2) is that I do not find explanatory sentences such as: "Because we assumed Dirichletian priors .." to be very illuminating.

8.19.2021 3:27am - (Replying to @LuigiBiagini) What precisely is the problem? The goal of AI is almost the same as the goal of economics -- to understand the world around us so we can control it. The only difference is that, for AI-ers, "understanding" means represent knowledge & inference in a machine, to prevent cheating.

8.19.2021 2:11am - Before we say "explainable AI" we must decide WHAT is it that we wish to explain. Are we about to explain the function that the system fitted to the data? or are we about to explain the world behind the data? Science writers seem unaware of the difference.

8.19.2021 1:57am - (Replying to @WvanAmsterdam @eliasbareinboim and 2 others) Limiting the word "effect" to binary contrasts is too constraining. Specifying the entire curve p(Y=y|X=x,do(A=a)) for all y,x,a is what "effect" means, especially in non linear, non-binary variables. With this understanding, I would label p(*) the x-specific effect of A on Y.

8.19.2021 1:02am - More Orwellian news. The Jewish Voice for Peace is opening a fund-raising office in Kabul, headed by Ariel Gold. Next, Qatar just announced the establishment of Edward Said professorship for Jewish Studies at the University of Kabul, to be held by Peter Beinart.

8.18.2021 12:12pm - (Replying to @JadePinkSameera @Lester_Domes and 12 others) Why? Is there a shortage in stat textbooks ? Or, you mean a text that will prepare stat students for modernity? Out of rung-1 of the Ladder?

8.18.2021 12:06pm - (Replying to @analisereal @seanjtaylor and 4 others) Thanks. I was going to mention this book, indeed, because it deals with the general problem: Given a SQ (statistical estimand) in some form, find an estimator that would be optimal in some sense.

8.18.2021 9:17am - (Replying to @yskout and @CT_Bergstrom) You are absolutely right. Science writers are so eager to demonstrate that we have "two alternatives" with opposite conclusions. God knows why they stop here and do not tell readers that we now have the tools to decide which is more plausible. Poor Simpson.

8.18.2021 9:04am - (Replying to @CasualBrady @seanjtaylor and @MatheusFacure) I see a snip of the First Law there, am I right? Pedagogically speaking, do you agree that it deserves the title First Law?

8.18.2021 2:51am - Never underestimate the mental acrobatic, even creativity of a Zionophobic Jew

8.17.2021 11:15pm - (1/ ) (Replying to @seanjtaylor @MatheusFacure and @CasualBrady) Let's see. What makes an empiricist well-read on identification and counterfactuals different from a statistician who is NOT well-read in CI, wishing to estimate some statistical quantity SQ from data? I believe it is merely the fact that SQ typical of CI tasks is different

8.17.2021 11:35pm - (2/ ) (Replying to @yudapearl @seanjtaylor and 2 others) from the SQ's that are estimated in non-CI tasks. The former includes: backdoor formula, IV estimands, DiD, front-door formula, transport formulas, etc. I dont know a book/paper that covers all SQ's that emerge from CI analyses; books usually cover some and not others.

8.17.2021 9:06pm - I wish I could join @seanjtaylor in recommending @matheusfacure 's book, but I can't find the First Law of Causal Inference in it (, so how can I? As to @CasualBrady 's notes, the chapter on counterfactuals is still in the making; it has great potential.

8.17.2021 7:03pm - Anyone who participated in the discussions on this educational channel should be able to submit a paper to this WHY-21 workshop, with the exception, of course, of those who refuse to climb the Ladder of Causation. #Bookofwhy

8.17.2021 8:22am - Remember the Open Letter that 55 USC professors sent to USC leadership Well, they received a reply: Note: No mention of Israel or Zionism. If I were at USC, I would still be wondering if I am a criminal or a desirable professor.

8.17.2021 5:05am - (Replying to @Foreman1David) Interesting analogy. You can't build a nation if the nation is not committed to building itself. It took Israel 30 years to build its infrastructure (1917-1947), same 30 years that PA has had since Oslo, and which it wasted on denying its neighbor, instead of building itself.

8.16.2021 10:29pm - Turkish delight! Hilarious! If only it were a joke.

8.16.2021 4:42am - Why Durban IV must be boycotted - opinion
8.16.2021 4:42am - (Replying to @yudapearl) Another uplift on a day of infamy: reading BHL's poetic piece on why Durban III is a disgrace to our generation.

8.16.2021 4:07am - The Taliban spoiled, not ruined my day, because the innocent smiles on the proud faces of Israeli kids greeting their Gold Medalist Linoy Ashram made me an optimist again, for just two minutes, true, but what more can one expect on a day when millions of Afghans stop smiling.

8.15.2021 9:05pm - I'm all open to "tendencies", but I can't understand why it is called "a tendency" rather than "betrayal of core values", and how this betrayal manages to lure universities to grant it "academic department" status. Beyond me.

8.15.2021 6:25pm - Leaders of India and Israel Share Messages of Friendship on India's Independence Day
8.15.2021 6:25pm - (Replying to @yudapearl) While leaders do their act, let me send a personal congratulation to all Indian readers of this channel on India's Independence Day. When our son, Danny, was assigned to the WSJ bureau in Bombay, he wrote to us with pride: "I'm going to cover the world's largest democracy!"

8.15.2021 6:10pm - When the Jew-Bashers Are Jews
8.15.2021 6:10pm - (Replying to @yudapearl) Another personal disappointment from President Biden: His appointment of pseudo-Rabbi Sharon Kleinbaum to the US commission on International Religious Freedom.

8.15.2021 4:48pm - (Replying to @xjfhdndj) You are saying in essence: Western feminism is anything but feminism. I agree. Western feminism is interested in whatever leverages populist emotions but feminism. "Palestine is a feminist issue, the Taliban is someone else's, perhaps the dpt of cultural relativism, not us."

8.15.2021 1:52pm - This may sound super-naïve, but I had a dream that UCLA department of Gender Studies posted a "Statement" (on University Website) saying: "Today, we stand in solidarity with Afghan women, whose gender/human rights have been taken backward 1,500 years or more". Naïve dream.

8.15.2021 1:28pm - Lapid decries passage of Poland's `anti-Semitic' restitution law, recalls charge d'affaires from Warsaw via @JNS_org
8.15.2021 1:28pm - (Replying to @yudapearl) This may sound far fetched, but Poland's latest bill may affect me, personally, in some way. My grandparents owned a house in the town of Kielce. They perished in Auschwitz in 1941, and I haven't inquired on its status since -- too painful.

8.15.2021 7:20am - Taliban Enter Kabul, Promise `Peaceful Transfer of Power' via @TheMediaLine

8.15.2021 7:11am - (Replying to @ncphi) Good suggestion.

8.15.2021 7:09am - The PhD Dissertation of Ang Li is now posted on:, Titled: "Unit Selection Based on Counterfactual Logic", it should be of interest to folks developing Recommender-type systems, A/B testing and (frankly) everyone doing ML+CI. Congratulations! @Ang_UCLA

8.14.2021 7:23pm - (Replying to @yudapearl) @Bakari_Sellers , I wish to commend you for standing up for Israel and Zionism against the toxic rhetoric of Cornel West. I have advised West in 2015 to retire, since his words no longer carry the weight they once did, Evidently, he did not heed my advice.

8.14.2021 4:17pm - I am re-posting this video panel on ML and Causality that took place 2 years ago: It's fascinating for me to compare the perspectives that ML folks had then with what they think now. Much has been changed, but some of the unwise reactions still exist.
3.1.2019 1:40am - I came across this interesting article about Causality in Machine Learning which also links to a recent panel on causality It is interesting to see how young ML researchers react to the new hype. Very encouraging #Bookofwhy.

8.14.2021 2:28pm - (Replying to @BarElitzur) What would be helpful is a formal definition: Def. xyz: Given an SCM M, event C is said to be a Twisted Mackie Cause (RMC) of event E if the following holds in M: (1)...(3) .

8.14.2021 2:10pm - To be honest, I have not expect this from Biden either; I voted for him on the basis of the fortitude and principles he espoused in the campaign. Will the Biden I voted for wake up?

8.14.2021 1:48pm - Bakari Sellers Denounces Cornel West as an 'Anti-Semite' for Saying He's Tied to `Very Right-Wing Jewish Lobbyists' via @mediaite
8.14.2021 1:48pm - (Replying to @yudapearl) I wouldn't dignify Cornel West with the title "anti-Semite", and certainly not "progressive". He is plainly a Zionophobic racist, for whom Israel exists only in the pockets of right-wing lobbyists, void of human beings, void of history and dreams.

8.14.2021 3:07am - This paper on "Estimating individual-level optimal causal interventions" is truly about "individual-levels", no imposters: But I am having hard time understanding the approximation used. Would appreciate help.

8.14.2021 2:07am - I don't know why everyone is jumping at poor @RepRashida . Her purpose is not to pass legislation but to use her platform to smear and malign one tiny, beleaguered nation she hates. Her success is measured by the amount of dirt she manages to spread in the Halls of Congress.

8.14.2021 1:17am - Broken link? Let's try this one: Modelling linguistic causation - #ScholarAlerts

8.14.2021 1:12am - Some of the issues you are raising are covered by the "Theory of measurement" (eg Foundations of Measurement, Krantz, Luce, Suppes, and Tversky (1971/1989/1990). ) which deals with mappings between arithmetic operations and physical operations (eg mixing things).

8.14.2021 12:50am - (1/ ) Anyone who is serious about explainable AI must think seriously about ways in which world knowledge licenses linguistic choice in the production of explanations. This paper paper argues that Structural Causal Models can form the basis for such choices.
8.14.2021 12:50am - (2/ ) (Replying to @yudapearl) I feel uncomfortable however with the emphasis on Mackie's INUS condition, which is not uniquely defined in SCM (Causality, Sec. 10.1.4

8.13.2021 5:15pm - (Replying to @voidmstr) "Umitsadenu Od Yareem Anachnu PO" "And our steps will some day proclaim: 'We were here'". My fingers tremble when I type these words. Here is Theodor Bickel singing it, in Yiddish:

8.13.2021 4:41pm - (Replying to @yudapearl) Their banner did not say: never again. Because they knew it will occur again. Their banner said:" Anachnu PO" (We were here), thus becoming an inspiration for generations to come, to break fear-driven silence and stand up to injustice.

8.13.2021 10:21am - (Replying to @DrBobGoldberg and @purplemeheart) Warm? I believe my Asian roots go deeper than any of those in the department of Asian Studies. Can they recite poems that were written 3,000 years ago in Asia? I can: "Bereshit Bara Elohim..."

8.13.2021 9:56am - France's Macron to boycott UN racism conference on anti-Semitism concerns
8.13.2021 9:56am - (Replying to @yudapearl) What made my day? France still shows signs of rational thinking; remnant of its glorious past.

8.12.2021 7:12pm - The only thing I would add to this article is a collegial message to the Chair of UCLA Department of Asian-American Studies, Natalie Masuoka @purplemeheart , asking if she is not embarrassed by her department juvenile "statement". (And if she accuses me of "yellow washing")

8.12.2021 1:43pm - The difference is important. Anti-Semites actually believe that Jews are wicked, deceitful and violent, whereas @RepRashida & @Ilhan know that Israel is progressive, honest and humane; So why do they want her dead? Tribal honor, religious zeal and perhaps deeper still.

8.12.2021 7:49am - (Replying to @EinatWilf @giladerdan1 and @HolocaustMuseum) I don't believe the @HolocaustMuseum depicts faithfully the role of Britain in preventing Jewish emigration to Eretz-Israel, not only in Haifa but all the way to the Danube River and the ports of Romania. And, of course, the role of the Mufti behind Britain's action.

8.12.2021 6:18am - These are not anti-Semitic conspiracies but warped minds of @RepRashida & @Ilhan who see "violence" in everything Israel does (drinking, breathing, existing) and victimhood in everything Palestinians do (rockets, terror, denial, eliminationism). Warped Neural Nets in Congress.

8.12.2021 5:56am - (Replying to @roireichart) Congratulations! Under normal circumstances, a paper using do-operator authored at the Technion would be one I should understand. But this one presents a notational obstacle on my way: Absence of task description in terms of conditional expectations, as in

8.11.2021 2:03pm - In a letter to UC President, @JGreenblattADL supports UCLA professors who protested the politicization of academia and the criminalization of Israel and its supporters. See Thanks you @ADL

8.11.2021 1:43pm - Thanks for anointing me with a black mustache, but I can't understand these comics - I grew up with little red riding hood and, even she was speaking English. The only utterance I understood was "When are all these curve-fitters going to learn to learn?" Still wondering: When?

8.11.2021 3:28am - (1/ ) A brilliant rebuttal! UCLA administration sends us such a statement twice a week, and I am dying to ask: What's the purpose? It either "affirmed platitudes to which we can all subscribe, or, more menacingly, it asserted controversial and arguable positions
8.11.2021 3:28am - (2/ ) (Replying to @yudapearl) as though they were axiomatic certainties." Unfortunately, I dont have Glenn's courage and, like the rest of UCLA faculty, we smile to each other and hope for the day when axiomatic certainty would not necessitate "sophomoric nostrums." I bet Loury is a great econometrician.

8.11.2021 1:58am - Not every fashion (or "in vogue") carries wisdom. I am trying to focus on principles rather that fashion, knowing that the latter is often a product of ego-centered personalities, stale traditions, isolationist organizations (eg NBER) and just bad habits.

8.11.2021 12:34am - It's more depressing to see not one dissenting voice from the faculty of the Gender Study department. Not one to come forth and complain: "I wasn't asked to sign" or "The Emperor has no clothes"

8.11.2021 12:10am - (1/ ) Thank you Aaron for giving voice to USC faculty. The thing to watch now is how the President responds to faculty members who are responsible for USC's academic reputation. The litmus test is whether she uses the words Zionism and Israel, or continues to hide in the bunker of
8.11.2021 12:10am - (2/ ) (Replying to @yudapearl) 2/ "fighting antisemitism." The worse response would be to ask the offending departments to add a tiny footnote to their websites saying: "opinions expressed to not represent the University", thus preserving the damage and encouraging other BDS operatives to hijack departments.

8.10.2021 7:59pm - Evidently, Palestine is a feminist issue after all! I was almost led to believe it is a populist slogan used by BDS-hijacked Gender Study departments to justify their academic status. (As in USC, for example, No! "Palestine is a feminist issue".

8.10.2021 2:43pm - (Replying to @danieltsinis) Solution manuals are available upon request from

8.10.2021 2:40pm - (Replying to @ben_golub and @lewbel) Yes, because it applies to EVERY individual, not "on the average". It is in fact counterfactual restriction. It is not excluded from SCM/DAG analysis, but it is classified correctly, distinct from structural assumptions.

8.10.2021 2:33pm - (Replying to @deaneckles @ben_golub and @lewbel) Strictly speaking "nonparametrics" stands for "No restrictions on the functions or on the error distributions". See Causality. But the motivation is not fear of parameters per se, but fear of the indefensible. If you have a parametric assumption that is defensible, by all means.

8.10.2021 2:20pm - (Replying to @lewbel) I was led to nonparametrics by commonsense, not by pseudo-revolutions. The logic was: Structure is more defensible than numbers. Here is an example: ppl tend to mistake "monotonicity" with "spend more on the average", it's in fact "with no exception". Data quality is unrelated.

8.10.2021 9:47am - (1/ ) My eyes got glued to this review of CI in forensic science: perhaps because it claims that, lacking a gold standard, forensic medicine is inherently different from establishing a diagnosis in clinical medicine. In view of what we know today about causes
8.10.2021 9:47am - (2/ ) (Replying to @yudapearl) of effects and probabilities of causation (eg,,, I am not sure the dichotomy is as sharp as the review claims.

8.10.2021 2:08am - (Replying to @BaskervilleJudy @FeechPeach and @KevinKay500bee) Indeed, the word "antisemitism" does not appear in the USC Letter. On purpose! To prevent the administration from taking refuge in the "antisemitism" debate. What is emphasized instead is (1) the criminalization of Israel's birth and (2) its effect on students.

8.10.2021 1:28am - There is some wisdom to this fashion. Non parametric identification is based on what we assume holds true in the world, functional restrictions are based on what we wish would hold true so that we get identification. The fashion is not against restrictions but for transparency.

8.9.2021 11:11pm - (Replying to @AlessandroStru4) Taking political position, especially on issues you know nothing about, is indicative of content-free depts. The academic content of Gender Studies dpts. should indeed be re-examined for accreditation. If my dpt decides to sell cosmetics one day, someone would surely notice!

8.9.2021 10:33pm - (Replying to @microsamonomics) The university is guilty of blindness and inaction, true, but the departments (in reality, just a few loud-mouths in the department) went as far as criminalizing their own students (and own colleagues) in the name of the University.

8.9.2021 9:44pm - (Replying to @la_cosgrove and @seanjtaylor) Not sure I replied to your last and important tweet. Can you help me identify the phrases that make people feel I am "aggressive" or "unkind" or anything less than "good faith". I am a slow but faithful machine learner.

8.9.2021 6:08pm - 1/50 top professors at USC just released this open letter to the USC leadership:, The new phenomenon -- departments endorsing political pledges that have nothing to do with their educational or research mission -- is spreading across the country
8.9.2021 6:08pm - (2/ ) (Replying to @yudapearl) 2/ like a brush fire, and is threatening higher education at its very core. The fire has intensified to the point where it is no longer raging among students; it has now ignited key professors who take it as an unaccepted climate of personal harassment on their own campus.
8.9.2021 6:08pm - (3/ ) (Replying to @yudapearl) We are witnessing the collapse of higher education, while university administrators look the other way, hoping for the fire to subside once they reaffirm their century-old "commitment to fighting antisemitism"; namely a century-old license for inaction on campus Zionophobia.

8.9.2021 12:29pm - (Replying to @soboleffspaces) I'll have to look it up. But there is a slight nuance of interpretation there. PS_01 :=P(Y_1 =1|Y_0 =0) stands for: "the proportion of people who would die if forced to a certain activity, among those who would not have died if *forced not* to do so"

8.9.2021 5:00am - (Replying to @radagaisus and @nntaleb) I would like to understand first if the asymmetries in @nntaleb 's book bear any relations to causal directionalities.

8.9.2021 12:39am - The PhD Dissertation of Carlos Cinelli is now posted on: Titled: Transparent and Robust Causal Inferences in the Social and Health Sciences. Congratulations @analisereal

8.9.2021 2:49am - (Replying to @BobbiePoPS and @analisereal) I was just trying to better understand what you were saying. Can't read w/o eye-glasses.

8.9.2021 2:20am - (Replying to @PardonMusings) It's getting more comical when it comes to Israel: "ethnic cleansing" "apartheid" and, most ridiculous of all "colonialism".

8.9.2021 1:07am - (1/ ) The story has a tragic coda. We asked: "What made our teachers so good?" The answer is: A man named Arlozoroff, who signed an agreement (1933) with the Nazis to allow 60,000 German Jews (with their possessions) to emigrate to Eretz Israel. He was
8.9.2021 1:07am - (2/ ) (Replying to @yudapearl) assassinated on the beach of Tel Aviv in 1933, possibly (though not certain) for his negotiations with the Nazis. Israel gained a tremendous boost in technological and academic infra-structure, of which we were direct beneficiaries. Our teachers were all intellectual giants.

8.8.2021 5:37pm - (1/ ) An anecdote about the smallpox vaccination and my high school teacher. In my introduction to Causality I told the story about our teacher Dr. Feuchtwanger who first told us about the smallpox vaccination controversy as a logical fallacy. Last week,
8.8.2021 5:37pm - (2/ ) (Replying to @yudapearl) I talked to Daniel Kahneman who asked: I also had a teacher named Feuchtwanger, are they the same? It turns out we happened to go to the same high school in Tel Aviv, two years apart 1947-9, and shared same science teachers. Lesson: Good teachers live forever.

8.8.2021 12:37pm - It didn't occur to me that my mockery of the "logic" and "vocabulary" of Cory Bush would be taken seriously as an endorsement or worse. I should be more explicit next time. She represents the dangers of where America is heading.

8.8.2021 10:54am - The #Bookofwhy was written before Covid-19 became the talk of the day. And the smallpox vaccine controversy really took place in 19th Century France. The arguments were very similar. A message to Stat educators: Absent mathematics for causation (ie DAGs) they will repeat again.

8.8.2021 10:39am - It's a strange phenomenon that happened to me too. Having been defending #Bookofwhy in the past three years against all kind of critics, from extreme right to extreme left, it feels so refreshing to re-read the raw text and listen to the music singing: "How true! How true!". Try!

8.8.2021 10:29am - (Replying to @ncphi) This follows from probability theory: P(y|x) = SUM_z P(y|z,x)P(z|x) It's just the law of total probability P(y) = SUM_z P(y|z) P(z) applied to P(y|x), instead of P(y).

8.8.2021 4:00am - Hezbollah is no longer deterred by the IDF - analysis

8.8.2021 3:01am - (Replying to @GarhNauti) Extremely worried, and lost in thoughts.

8.8.2021 2:16am - Artificial intelligence can now be recognised as an inventor after historic Australian court decision

8.7.2021 7:23pm - Wasn't it Karl Marx who said: "Whatever exists must perish" ?? He was probably right, but I love my old books, the equations I once proved, the freedom to prove more, and the rescue team that came over when I called 911.

8.7.2021 5:52pm - Funny. I heard UNRWA officials claim: "Not true! We do not encourage this belief" instead of stating explicitly and publicly what they do believe should be the fait of that land "from the river to the sea." Quick to deny, too scared to state.

8.7.2021 4:15pm - Olympics: Rhythmic Gymnast Linoy Ashram wins gold medal for Israel
8.7.2021 4:15pm - (Replying to @yudapearl) I am ecstatic. She truly represents Israel, the "against all odd" country. After fumbling her hoop rotation yesterday, she unleashed her inner strength.

8.7.2021 3:47PM - I can listen to her all day long; her logic inspires me, the vocabulary, the mentality, and the vision of where America is heading.

8.7.2021 2:54pm - An article titled "What's (successful) extrapolation would surely wet the appetite of every red-blood ML theorist. Alas, instead of defining successful extrapolations by causal logic, the author takes us back to pre-causal era of model-free speculations.

8.7.2021 2:36pm - (1/ ) Readers ask how "actual causes" affect "policies". Here is an example. In an interview last week on MSNBC, Robert Malley said, "America is less safe because Iran has a more expansive nuclear program and because it accelerated and intensified its regional activities"
8.7.2021 2:36pm - (2/ ) (Replying to @yudapearl) Malley's actual cause was: the Trump administration's maximum pressure campaign and the killing of Soleimani. Conclusion: Reverse that policy, minimize pressure, maximize deference, appease the Ayatollas with cash, ignore their attempted kidnapping & blame the "actual cause."

8.7.2021 11:51am - (Replying to @AnothrAtheistJu @HillelIntl and @ADL) It seems the reporter also noticed the absence of key words in the public statement, did some digging on his own, and received the usual "not every" reply from ADL. Now I am doubly worried about this initiative ending up where others did.

8.7.2021 3:14am - We're in receipt of a new book Deceitful Media: Artificial Intelligence and Social Life After the Turing Test, by Simone Natale. AI folks should not be insulted by the word "deceit," intelligence is indeed unique -- the only way to fake it is to have it.

8.7.2021 2:01am - For faculty and students embroiled in the latest Zionophobic campus madness, this new initiative by @HillelIntl & @ADL may be a game-changer. However, the absence of the words Zionism and Israel from their statement makes me tremble in doubt & discomfort.

8.6.2021 7:14pm - (Replying to @mgaldino and @rsilveira79) Nothing against PO, it's the blind brother of SCM.

8.5.2021 11:12pm - Glad you alerted me to this source of confusion. By "theoretical limitations" I mean mathematical results on what can and cannot be done with data alone. For example, you cannot tell what effect intervening on X would have (on Y) by just watching samples of (X,Y) pairs.

8.5.2021 11:01pm - (Replying to @mike_ivanov) We cannot escape causality when we talk about generalization across domains, because the notion of "domain shift" is causal, like intervention, and cannot be properly handled by data alone, e.g., the domain can shift while the distribution remains unaltered.

8.5.2021 10:18pm - Congratulations are due to our new UCLA colleague, Aditya Grover, who just won the 2021 ACM SIGKDD Dissertation Award Aditya's research is centered around machine learning, probabilistic modeling, representation learning, and sequential decision making.

8.5.2021 9:00pm - I have written to the directors of the Stanford's Foundation Model Center, asking whether they can suggest someone from the Center who can teach us the principles behind the new methodology. I hope to hear from them.

8.5.2021 6:25pm - Thanks for resurrecting this Quora post (June 2019) which, untypical of me, is still resonating with my current views. I would only change the counterfactual example to read: "If only it did not rain yesterday," to dissuade people from thinking they can compute it using RL.

8.5.2021 3:03pm - What gives me the audacity to say something meaningful about a system trained on gazillion samples, something I havn't seen done before? Ans. I know something about infinite samples, because CI is inference from distributions. Whatever fails at infinity would fail on gazillion.

8.5.2021 2:02pm - (Replying to @MariaGlymour @ProfMattFox and 2 others) You are absolutely right, some economists do not know what their IV conditions mean, or whether the needed conditions hold in a given situation. Here is an illuminating discussion

8.5.2021 12:26pm - (Replying to @MariaGlymour and @ProfMattFox) Yes, DAGs offers a meaningful language for encoding and scrutinizing assumptions (there might be others) and, once in DAG form, we can derive conditional exchangeability conclusions if they are needed for identification. What we cannot afford is direct judgment of cond. exchang.

8.5.2021 10:28am - (Replying to @MariaGlymour and @ProfMattFox) One tiny grain of salt, and we would be in agreement. In order to "figure out what possible assumptions you might adopt.." you must understand what those assumptions mean, so they must be encoded in a meaningful language, not as cryptic conditional ignorability statements.

8.5.2021 8:54am - (1/ ) This just in. A new successful paradigm for building AI systems has emerged, called "Foundation Model". According to its inventors, it works as follows: "Train one model on a huge amount of data and adapt it to many applications." Not only is it
8.5.2021 8:54am - (2/ ) (Replying to @yudapearl) 2/ seen "as the beginnings of a sweeping paradigm shift in AI", but a whole Center has been erected in its honor, dozens of prominent researchers, post-docs and PhD students has joined its staff, and an interdisciplinary symposium has been announced. We, foot-soldiers in the
8.5.2021 8:54am - (3/ ) (Replying to @yudapearl) trenches of AI research are asking, of course: "What is it?" or "What is the scientific principle by which 'Foundation models' can circumvent the theoretical limitations of data-centric methods as we know them, especially those that hinder generalization across environments?"
8.5.2021 8:54am - (4/ ) (Replying to @yudapearl) I have tried to pin point the principle, but ended up in failure; the limitations as we know them are not mentioned in the papers I examined, nor are concepts such as "causation" "Data Fusion" "Transportability" "data-centric" etc. Can some readers enlighten us?

8.5.2021 8:49am - Put differently, messy data are not a license for fuzzy logic. @ProfMattFox @MariaGlymour

8.5.2021 8:40am - You will never hear me say that "artificial intelligence is not actually intelligent". It is, if done right. It isn't, if handcuffed by data-centric thinking. @mims @wsj

8.5.2021 8:27am - (Replying to @EpiEllie and @HarvardEpi) Revisit our assumptions - by all means. Revise our assumptions - by all means. But coming up with conditional exchangeability judgments "based on the data we have" (instead of "based on the revised assumptions"), I am yet to see mortals capable of such judgments.

8.5.2021 8:16am - (Replying to @suicuneblue) I am quite ready to accept the dream of training gazillions on bajillions, but not before I see it working on 3 x 3. Moreover, I become twice as skeptical if I can prove it can't work on 3 x 3.

8.5.2021 2:26am - Nice road-map for epi researchers. One comment: step 4 is cognitively formidable, it can be obtained instead from step 1. This is something that epi-education is still lacking. @EpiEllie @HarvardEpi

8.5.2021 2:02am - I will be convinced if someone shows me how the method works on a toy example, where we know the correct answers. For example, the 3 models in Fig. 3 of Note that equality of distributions does not mean equality of environments; tough for data-centrists.

8.4.2021 7:39pm - (Replying to @ftzo and @PyroAi) I think your definition of teleological modeling could be sharpened by specifying the type of questions you want the "framework" to answer.

8.4.2021 2:12pm - (Replying to @DJMinded) Indeed, the way my friends in Israel have been oppressed through 73 years of rejection, denial, strangulation and violent attacks is unprecedented in the history of mankind. Even dreams would not recreate it. Unimaginable.

8.4.2021 1:16pm - (Replying to @CAReyesProfe) You keep on using the term "same thing" which I have not heard from Stephen or anyone else. What I heard is "the 2nd wrong cannot undo the 1st", which I do not believe you deny.

8.4.2021 1:09pm - (Replying to @DJMinded) HMMM... I would suggest a different strategy: examine the facts and the ropes behind the facts, then use logic and commonsense. But, I agree, never label Zionophobes "Anti-Semites"; a title reserved for people with unfortunate upbringings, which the former do not deserve.

8.4.2021 7:32am - (Replying to @DJMinded) And I have never heard a viable argument for the apartheid fiction from anyone who was not, at least to some extent, either temporarily blind, or morally deformed. So, how are we going to reconcile our two diverse experiences?

8.4.2021 7:01am - (Replying to @CAReyesProfe) If you see a sense in which the "new" racism CAN solve the "old" racism, define the two and enlighten us. But you cannot deny that the "new" attempt to solve the "old" problem has a racist element to it, perhaps benign, perhaps inevitable, but racist nevertheless.

8.4.2021 6:06am - The New Racism Won't Solve the Old Racism
8.4.2021 6:06am - (Replying to @yudapearl) Bret Stephens is one of the few remaining thinkers on the NYT staff; his words both capture and shape ideas. But his courage to say what needs to be said is unmatched.

8.4.2021 4:44am - Remembering Ruth Pearl, Mother Of Slain Journalist Daniel Pearl
8.4.2021 4:44am - (Replying to @yudapearl) Thank you Ari Shapiro for re-sounding Ruth's voice, a voice of strength and hope we have found impossible to replace since she left us two weeks ago.

8.3.2021 8:48pm - Just in: A profound victory for the rational wing of the Democratic Party, proving that AOC's loud-mouthed squad is an empty noise maker, not a vote getter. "A coin in an empty tin-can makes an awesome noise" says an Aramaic proverb. Will Nancy Pelosi get the message?

8.3.2021 6:41pm - Clarifying: The nostalgic paper I wrote about was:, which got lost in my retweet. Its thrust is: Who needs do-operators, counterfactuals etc. if we have RCT's data to tell us all about interventions. It fits on Rung-1/2 of the Ladder in #Bookofwhy

8.3.2021 11:41am - This paper is a hymn of nostalgic yearnings for classical statistics and its causality-free vocabulary. But why? Do statistics' achievements justify the nostalgia? Is causality-free language more scientific? Don't scientists communicate cause and effect when no one is looking?

8.2.2021 8:59pm - The truly worst part of machine learning is that remedies for the problems you listed (i.e."selection bias") are available (eg #Bookofwhy, but ML practitioners and theorists who are too busy building cathedrals for machine learning,

8.2.2021 10:35am - (1/ ) This is sadly true. And it comes up each time a colleague asks: "Who needs CI? Statistical decision theory is sufficient!". I take them to their latest decision tree, point to an arrow emanating from a decision node and ask: "What probability is this?". The answer I get is
8.2.2021 10:35am - (2/ ) (Replying to @yudapearl) invariably P(outcome| D). Once I convince them that the correct answer is P(outcome|do(D)), which is not equal P(outcome| D), Operational Risk Management gains a Causal Inference devotee. But science grinds slowly. See

8.1.2021 6:50pm - Readers ask why Danny does not smile. My guess: He felt silly in the rented bow tie and tuxedo. TBH, I felt silly too, but it was my induction to the National Academy (1995) and I didn't want to break the protocol. I changed.

8.1.2021 6:13pm - They ranked him 6.6 on "difficulty". For me it looks at least 600, though I haven't tried it yet.

8.1.2021 4:18am - Israeli gymnast Artem Dolgopyat wins Olympic gold medal in Tokyo Games

8.1.2021 3:45am - (1/ ) Your @amjepi primer on DiD could improve its motivation section. Your readers have surely read #Bookofwhy, so they know that, given a DAG, every question about effect size can be answered algorithmically by do-calculus. This raises the question "Why DiD?". Namely, what type of
8.1.2021 3:45am - (2/ ) (Replying to @yudapearl) problems necessitate the use of DiD methods, that is, why couldn't they be handled by standard methods of do-calculus. Has any reader asked this question? #EpicPartner @societyforpei #epitwitter @causalinfer

8.1.2021 3:17am - (Replying to @RWJE_BA) Our belief that our experiment is close to "ideal" relies on assumptions about the procedures taken and, more importantly, on a logic which transports the credibility of our conclusions to the veracity of those assumption.

8.1.2021 2:00am - (Replying to @RWJE_BA) Try one in "experimental causality" which is not transformed to assumptions about the procedures producing the data.

8.1.2021 1:54am - (Replying to @RWJE_BA) To make your point, you should give an example of a "genuine solution" which is not merely a "transformation".

8.1.2021 1:52am - The interesting thing about this "Bad Week" is the remedies proposed by the authors, which are just re-namings of failed remedies to the fundamental problem of data-centricity It seem like none of the authors read #Bookofwhy #ai #deeplearning

8.1.2021 1:14am - (Replying to @RWJE_BA) This is precisely what "solving" means in science; transforming one problem to another which is based on a more plausible set of assumptions or on more credible sources of data. #Bookofwhy

7.31.2021 11:14pm - I happened to violate this rule, having fallen in love with a theory that answers new questions like no other theory. And I intend to remain in love with her even when younger competitors answer more questions, remembering what she accomplished in her youth.

7.31.2021 11:04pm - (Replying to @RWJE_BA) Where you see notation others see logic. Lacking notation, logic would still be in its pre-Boolean state of disarray.

7.31.2021 6:49pm - (Replying to @grecostatistics) We care about this question for one reason only. The argument that "do(x) is outside statistics proper" is still being used today to deter statisticians from using it, and many are unfortunately deterred by it, as if "statistics proper" offers an alternative. It doesn't.

7.31.2021 5:21pm - (Replying to @RekaSundaram) The applications are all over, what we need is a methodology. E.g., We are receiving dozens of reports on experimental studies on Covid-19 patients. Is there a way of combining them to say something more than just what each study says in isolation?

7.31.2021 5:11pm - (Replying to @PhilosopherMD1) It does "count"; of course. Now try to strip all problem-specific acronyms and express the general rule of "counting" formally. Do we have the language to express it, especially if study-1 differs from study-2 ???

7.31.2021 4:35pm - Put another way, the language of standard statistics does not allow us to take experimental findings from past studies and leverage them to draw conclusions about effect size in a new study. Curious: Does Evidence-Based methodology offer such a language?

7.31.2021 4:23pm - (Replying to @RWJE_BA) Yet the languages that statisticians lacked is one that would allow scientists to articulate either judgmental assumptions or experimentally provable findings toward the estimation of a new effect size. I wonder if Evidence-Based scholars have acquired such language.

7.31.2021 8:38am - (1/ ) Can "traditional statistics" handle "effect sizes?" If we include Neyman-Rubin in "traditional statistics" and interpret "Can" to mean "Can, in principle", the answer is Yes. However, if we take "traditional statistics" to be represented by: Pearson, Fisher, Chochran, Tuckey,.
7.31.2021 8:38am - (2/ ) (Replying to @yudapearl) Breiman, Friedman,...+deceased presidents of ASA, RSS...+authors of stat texts+..., and if we interpret "Can" to mean "Capable of handling a simple problem in 2 weeks time," I would bet 100:1 on "NO!". Reason: They lacked a language to articulate the assumptions needed for
7.31.2021 8:38am - (3/ ) (Replying to @yudapearl) estimating effect sizes, and it takes about 2 weeks to learn such a language, be it "potential outcomes", DAGs, SCM, or equivalent. Why haven't the giants bothered to learn any? My 1993-9 email is full of reasoned excuses, but the most common one has been: "It takes us out of
7.31.2021 8:38am - (4/ ) traditional probability calculus.." or "Is do(x) a concept in some new theory of probability? Define it! (actual quote)". As T. Kuhn said: "Never ask a scientist to jettison the paradigm that has been the basis of his/her life's work (paraphrased) you will never be forgiven." (Replying to @yudapearl)

7.30.2021 12:04pm - (Replying to @RevDocGabriel and @kareem_carr) What happened? Are there still statisticians left? I thought they now call themselves machine leaning experts.

7.29.2021 3:53pm - Agree with @lewbel . The culture of "more control" is so deeply entrenched in the DiD literature that, unless someone writes a DiD-specific crash course, it will continue to misguide the unwary, and encourage the "bad controllers". Any volunteers? @causalinf , @EconBookClub

7.29.2021 3:40pm - `We will win this war'--Ruth Pearl, in her own words via @jdforward
7.29.2021 3:53pm - (Replying to @yudapearl) Sharing my wife's life story, in her own words.

7.29.2021 12:24am - No one screams "antisemitism"; the scream is: "thoughtlessness". If Ben & Gerry think Israel can defend itself better without controlling the West Bank, it is a valid military opinion. But their decision to criminalize in public those of opposite opinion is careless at best.

7.28.2021 7:42am - While most authors mention Cox and Efron's quest for: "data descriptions that are potentially causal," only few (I counted 2) note that this aspiration has been fulfilled in the 20 years that passed since Breiman's paper.

7.27.2021 10:45pm - Thank you Rob @Foodaism for remembering Ruth and rekindling her light in our memories.

7.26.2021 10:16pm - Hillarious. Economists who fear "endogeneity" in the 21st century must be living in the 20th. Some would rather die of polio than take the vaccine they vow to avoid. Like this vaccine: or these vows:,

7.26.2021 8:03pm - UC Davis Faculty Group Argues Departments Condemning Israel Violates School, State Policy
7.26.2021 8:03pm - (Replying to @yudapearl) Evidently, UC Davis has been infected by the same virus that diabilitated the UCLA dpt. of Asian-American Studies. Does UC President have the time to vaccinate the other depts and quarantine those tested positive?

7.26.2021 6:51pm - (Replying to @simonwiesenthal) Indeed, I often wonder if the UCLA dept of Asian-American Studies has any Asian or American faculty member in it. I have never met an Asian or American scholar who would use as foul a language as the one posted on their Hamas-inspired website.

7.26.2021 3:25pm - (Replying to @kareem_carr) I would buy "Less thinking" (which is the essence of formal logic") but I would replace the "comprehensive framework where questions can be asked" with a "framework where some questions can be asked and others, the important ones, cannot". I am hinting at

7.26.2021 2:13am - (Replying to @kareem_carr) Hey Kareem, what was the result of you survey among Bayesian fans? Genuinely curious.

7.26.2021 2:09am - It was gratifying to see how the ladder of causation can organize and guide decision making even among rank-and-file practitioners.

7.25.2021 10:04pm - (Replying to @whywouldthatbe) Many politicians, many governments and many institutions have had many things to hide. And they did a good job.

7.25.2021 8:43pm - Palestinian: "This is just another 'White Washing'" Israeli: "What have we done wrong now?" Palestinian: "Breathe!"

7.25.2021 8:10pm - (Replying to @jordanfrank) According to those "professors", students that belong to targeted groups should simply renounce their mother milk and join the barking herd. Some do, like @PeterBeinart and @Sethrogen , others go into hiding. Administrators love it, education suffers.

7.25.2021 7:24pm - Amazing discovery! I haven't realized a video of that lecture exists. Too bad the slides are hazy. It would be interesting to see how my views changed in the past 20 years by comparing it to this one: which includes Generalizability and Missing Data.

7.25.2021 2:14pm - Interesting thread. Reminds me of my Lakatos Lecture at LSE 2002, where I considered my works on SCM to be a playground for ideas in scientific epistemology.

7.25.2021 5:28am - After 49 years Israelis killed at 1972 Munich Games remembered in opening ceremony
7.25.2021 5:28am - (Replying to @yudapearl) An honorable ceremony at the Olympics that ends 50 years of politically motivated denial and coverup.

7.25.2021 12:57am - We had a long discussion in this channel on what what Evidence-Based practices means, or should mean. What I learned is: EBM must be model-blind else it's not evidence based. I therefore can't relate to @educationweek paper, which advocates models for practice.

7.24.2021 9:10pm - Readers of @AmericanHeartJ might get the impression that every Cardiovascular Study requires its own analysis of covariates, colliders, mediators etc. Not so. Chap 7 of @Bookofwhy shows how to skip these details and get direct answer to any research question, using do-calculus.

7.24.2021 7:11pm - Sounds like taken directly from the archives of my own university, 2021: "what was most sinister was not just the threat of violence but the pusillanimity of the university authorities in bowing to it. (Attacks on Jewish students and on Jewish professors).

7.24.2021 6:28pm - Hail to Germania! But until we see France on the list we cannot consider contemporary Western civilization worthy of its enlightened reputation.

7.24.2021 4:36am - Thank you @AsraNomani for invigorating Ruth's life in your story:

7.24.2021 4:10am - My beloved wife was laid to rest on Thursday, next to our son, Danny, overlooking San Fernando Valley. I am grateful for the flood of condolences and support that you sent my way, which I wish I could acknowledge it in person. Will be back soon with renewed explorations of "WHY".

7.20.2021 7:21pm - Dear readers of this educational channel. I've lost my dear wife this afternoon, my northern-star and my college sweetheart. Need to take a few days off. Hope to be back.

7.20.2021 11:56am - Trouble is, these poor souls at the UCLA dpt of Asian-American Study do not see it as politics; they honestly believe it is their inner-most moral duty to criminalize other faculty members to fulfil their existential calling.

7.20.2021 11:45am - (Replying to @Dem_Bitz and @UCLA) I am still trying to find out what "Yellow washing" is, before signing: "A proud yellow-washer".

7.19.2021 10:30pm - (Replying to @rina31420183) Are you kidding? They think it is the most noble statement that was ever posted on their website. Such statements, in the name of the most noble of causes, are usually posted to cover up for empty academic content.

7.19.2021 9:54pm - (Replying to @amgenova) Your alma mater is in a state of dismemberment. Its leadership fears its herd, and the herd has gone amok. Hope? Next year will tell.

7.19.2021 9:01pm - (Replying to @erikbiz) The morally objectionable aspect of BDS is it's very aim: to demonize Israel to destruction (see ). E.g., the Statement of UCLA AAS dpt maligns the very creation of Israel as “seventy-three years of settler colonialism, racial apartheid, and occupation”,

7.19.2021 8:38pm - When a student union is hijacked by BDS cronies, students suffer but the university continues to function. When an academic department is hijacked, the university is dismembered, marking the end of higher education. A group of us decided enough is enough:

7.19.2021 6:44pm - Thanks for telling us about the historical roots of @HaagenDazs_US . Next time my wife unwraps her favorite chocolate bar its going to taste twice better, especially considering @benandjerrys capitulation.

7.19.2021 12:04pm - (Replying to @jasonhartford) A 3-variable example will be extremely helpful, especially if taken from Fig. 3 of where we know theoretically and transparently what can and cannot be done, which estimator is biased and which is not.

7.19.2021 3:42am - I am a slow learner, please forgive. How do we represent the input and how do we tell the machine about possible shifts? Say we have 3-4 variables, like in, and we want to use "equivariance" on any of the examples there. Can we?

7.19.2021 2:16am - (1/ ) I was under the impression that we are in possession of only one mathematical object (SCM) capable of converting causal information to generalizable predictions. Your tweet alerts me to the existence of a guiding principle called "equivariance", that I may have missed. Can
7.19.2021 2:16am - (2/ ) (Replying to @yudapearl) you describe conceptually how to use this principle in simple prediction tasks under shifting environments? What input is required? How is it represented? And what is the logic by which it promotes generalizability. No fancy equations needed, just a conceptual tweet would do.

7.18.2021 11:11pm - If the height is intentional, I would claim "algorithmic unfairness"; -- Data Science is the only one on high heels. If it isn't, I would claim "selection bias;" -- systemic absence of an underrepresented minority: Causal Inference folks.

7.18.2021 10:44pm - (Replying to @artistexyz @lewbel and 2 others) The frequentist-subjectivist debate is orthogonal to causal reasoning. It resides in a bygone era when people argued the meaning of probabilities with zero interest in the process that generated the data. Bygone!

7.18.2021 9:46pm - Am I the only one puzzled whether this treasure hunting journey into the mysteries of cosmology and sub-atomic physics is truly necessary for achieving AGI or even generalizable predictions? @wellingmax @GarryMarcus @ceobillionaire @montrealdotai @kenneth0stanley

7.18.2021 8:30pm - (Replying to @lewbel @artistexyz and 2 others) In your story it is the mental activity of "finding out about B" that caused A, not "B itself". To allow for mental causation, agent specific variables such as "intends" "desires" "knows that" etc. must be part of the model, in the form of multi-SEMs, one per agent.

7.18.2021 4:22pm - (Replying to @EliSennesh @TheMossadIL and @KenRoth) Thank you for alerting me to @KenRoth dive into the science of causation in his attempt to evade responsibility for causing the rise in antisemitic incidents. Deep inside he knows better; history will remember him.

7.18.2021 3:54pm - I have commented on Granger Causality here

7.18.2021 3:30pm - (1/ ) @KenRoth shares direct responsibility for the rise of antisemitic incidents. The causal logic is elementary: If Israel is as criminal as @KenRoth labors to paint her, then Jews who love Israel and support her with money, words and prayers, must be complicit in those crimes.
7.18.2021 3:30pm - (2/ ) (Replying to @yudapearl) And any feeble-minded who wishes to express an outrage at those crimes would naturally target those Jews. Intellectuals who spread combustible fuels knowing that lunatics are running around with lit matches cannot be totally exonerated from responsibility when fire breaks out.

7.18.2021 12:15pm - (Replying to @lewbel @artistexyz and 2 others) It couldn't be "debatable" because Granger causality is a probabilistic notion and we know that no probabilistic condition can be necessary (nor sufficient) for a causal relation. So, I would label it "orthogonal" rather than "debatable".

7.18.2021 12:24am - (1/ ) My attention to this clips by Max Welling was drawn by a couple of quotes such as: "If you have this causal structure of the can generalize much better". The bulk of the conversation however impressed me as a desperate search for something
7.18.2021 12:24am - (2/ ) (Replying to @yudapearl) better, a holy grail, that no one knows what it looks like. I kept thinking: Do they know that a miniature replica of a holy grail is available (SCM, of course) and that blind search is no longer necessary? I bet the next step in DL would be to take an ordinary causal model,
7.18.2021 12:24am - (3/ ) (Replying to @yudapearl) call it "inductive bias", have it guide the function-fitting capabilities of Neural Nets, decorate it with quantum terms, and declare victory of neural computation over symbolic AI. Why not just follow the hybrid strategy recommended here:

7.17.2021 10:55pm - Thank you @MeghanMcCain for narrating the story of Malki Roth in a time when justice is losing its meaning to political payoffs.

7.17.2021 9:12pm - Any public health instructor (AKA econ. stat & political science) who can solve the homework problems in Primer would be able to stand in front of a class and smilingly answer all causal questions w/o fumbling (unlike my professors). Solutions available.

7.17.2021 8:40pm - When I was 6, the sight of community leaders, Rabbis, the town's mayor, my teachers etc sitting on a 5" bench, ashes on their foreheads, crying their hearts: Jerusalem! Oh, Jerusalem! was one of the most memorable sight of my youth. It made the fall of Jerusalem a short-term fall

7.17.2021 8:14pm - (Replying to @peterlorentzen) I truly believe our gurus know in advance that their way will come out looking worse in any well-designed study in which we have an objective criterion of performance. It is for this reason that they insist on "real life" problems, where such criterion are fuzzy.

7.17.2021 4:19pm - Nice idea, the problem is that comparing "pedagogical approaches" entails (1) instructors able to participate and (2) agreeable criteria of success. Try to author a final exam to measure effectiveness and the gurus will object: "No toy examples, real-life only!" Back to darkness.

7.17.2021 1:23pm - (Replying to @peterlorentzen and @causalinf) Our 2013 survey of economics texts is here: But the two I regard as "harmful" are: (1) Angrist and Pischke, and (2) Imbens and Rubin (2015) Reason: They address causality explicitly and blow it.

7.17.2021 9:46am - (Replying to @quantrad @RamanDutt4 and @GlockerBen) Since causality if a property of nature and deep learning is a man-made tool to interrogate nature, the right question would be: Can we understand the role of deep leaning in capturing causality? Not the role of causality in deep learning.

7.17.2021 9:34am - (Replying to @artistexyz @VC31415 and @causalinf) I think you should include a footnote saying the time ordering + data still do not uniquely specify a causal structure.

7.17.2021 9:21am - Sorry for the semi-faulty link. Try this one Definition 3.2 gives a formal definition of "intent".

7.17.2021 9:14am - (Replying to @Davi_simon and @causalinf) Beautifully said "Great times we live in." If only students could comprehend and appreciate the unique greatness of our time, and stop listening to their outdated textbooks, more greatness would be awaiting us.

7.17.2021 1:06am - I feel for BLM and other organizations caught between the moral causes for which they were formed and the political ideology to which they sold their souls. The former says: Cubans are our brothers and sisters, the latter says socialism (and AOC) can't do wrong.

7.17.2021 12:56am - (Replying to @dugvudys and @Vijay_Ramjattan) I am trying to understand if the language itself carries racist messages or the people who teach it .

7.16.2021 7:37pm - (Replying to @emanumiller and @VeracityThe) Note that Ms. victimhood @HalawaTala who claims it was a moment of uncontrolled anger proves her current Zionophoic racism hence unfitness by labeling you "pro-Israeli mob". It does not occur to her that Israelis too have feelings and deserve the respect she demands for herself.

7.16.2021 6:13pm - (Replying to @FilipeAMoura and @BretWeinstein) It's the best compliment my vocal coach has ever received.

7.16.2021 4:49pm - (Replying to @causalinf) Hey this calls for celebration. I bet it was my blurb that made econ. instructors curious. Listen: "Cunningham's new book would probably shock economics instructors.... The benefits will outlast the shock." I love it. Time for celebration. Cheers.

7.16.2021 4:35pm - (Replying to @quid_agis_2713 and @TheConsultant18) Naïve me, thought "intent" is measured by the number of bleeding babies photos Hamas manages to produce for Al-Jazeera, NYT and UP. We had a causal definition of "intent" here, but I doubt lawyers would go for it in our generation.

7.16.2021 7:58am - (Replying to @Laconic_doc @pash22 and @BretWeinstein) Agree, I should have said "weighted average". But the point about not accounting for commonalities and disparities is still valid. Meta-analysis makes sense for aggregating statistical parameters, not causal quantities.

7.16.2021 7:52am - (Replying to @YallaGaza and @realRahulRK) Permit me to terminate this exchange. Those who defines what others are w/o listening to the others, are called "Orientalists" (remember?) and I have no patience educating them.

7.16.2021 7:40am - (Replying to @YallaGaza and @realRahulRK) You really have to learn a lot about Jewish peoplehood. The Jews you mentioned are currently thriving in Israel, secular and free, having ZERO connection to those countries, and immutable connection to each other: Common history and common longing to the same landscape.

7.16.2021 6:52am - (Replying to @YallaGaza and @realRahulRK) The more you talk the more I hear traditional Zionist ideas coming from your side: "peace & security & coexistence as EQUALS". Very encouraging. One missing ingredient is: "Equally indigenous". Can you pronounce that? Once you do, you will be elevated to "Zionist at heart"

7.16.2021 6:35am - (Replying to @Princeton @EPrinceton and 3 others) A personal impression: The audio of "The Book of Why" is truly superb. I sometimes listen to it just to marvel at how the music adds color to abstract ideas. The reader/actor has done an admirable job.

7.16.2021 6:26am - (Replying to @YallaGaza and @realRahulRK) I encourage you to continue to unveil to the world your state of mind, why Israelis have a point when they despair: "We have no partner" and why I begin to doubt if Palestinians will establish a State in my lifetime. One cannot demand that which one denies one's neighbor.

7.16.2021 5:58am - (Replying to @RekaSundaram and @Vijay_Ramjattan) So the discrimination was against non-English speakers, not inherent in the English language itself?

7.16.2021 5:50am - Honored to see #Bookofwhy among the chosen, but feel uncomfortable seeing it squeezed between "everybody lies" and "storytelling with data"- two practices I oppose. You can't tell a story with data and I don't see any reason to lie if you confess the limits of your methods.

7.16.2021 5:39am - (Replying to @YallaGaza and @realRahulRK) In my life time I've see only one "Apartheid entity" = Zionophobia. The one "entity" that denies homeland to one and only one people. This is "Grand Apartheid", standing unassailable above all imitations, fabrications and accusations.

7.16.2021 5:28am - (Replying to @Vetinari100 @pash22 and @BretWeinstein) Elaboration = ?

7.16.2021 2:55am - This is the most realistic Bayesian Network I've seen. It proves that causal diagrams can say things even NYT reporters and CNN anchors cant grasp -- the ropes behind the data. It should be made into a rally poster: "Don't count the rockets - count their purpose".

7.15.2021 7:27pm - (Replying to @lenlayton and @BretWeinstein) No need to call. I've been here for two years with "patient education" as my banner. If you want to challenge the hegemony of randomized trials, see I'll be glad to look into other flaws in RCT, perhaps they can be proven irredeemable or shown repairable

7.15.2021 5:44pm - (Replying to @SRMandeles @Yale and 2 others) I don't this it's Yale undergrads who attacked you, it's sounds more like the Malaysian Twitter-Army recruited to harass unwanted Tweeters. I would blame @Yale administration for not thinking what capitulation to BDS cronies can do to @Yale reputation.

7.15.2021 7:48am - (Replying to @DavidDeutschOxf) Thanks for reminding me of this one: “Hamalbin et pnei chavero barabim, k'eelu shafach damo.” “Whoever humiliates another person in public is considered as if he spilled his blood.” Vowing to arm myself with this teaching before diving into the Twitter swamp again.

7.15.2021 7:02am - Too early for celebration, @mehdirhasan , the question cleverly excluded antisemites like yourself. Let me rephrase the question a bit and OOPPS! Mehdi Hasan and his lefti Zionophobes would beat right-leaning antisemites hands down.

7.14.2021 11:19pm - (1/ ) Last week, July 4th, marked the 75th anniversary to the Kielce Pogrom; a tragedy that still defies comprehension: One year after the Holocaust, in a quiet town named Kielce, 42 Jews are killed by their Polish neighbors, on account of a blood-liable rumor
7.14.2021 11:19pm - (2/ ) (Replying to @yudapearl) I have a personal memory of this event because my aunt survived the killing and, later, when she arrived at Israel, shared her experience with us. While Poland is still trying to digest what happened, my aunt (departed) has dozens of grandchildren building a brighter future.

7.14.2021 9:23pm - (1/ ) A mini-victory at the LA Board of Education today: Note the difference: Instead of denouncing a slippery "antisemitism," as all Boards love to do, they went a far as denouncing "the rise in anti-Semitic, anti-Jewish, and *anti-Israeli rhetoric*
7.14.2021 9:23pm - (2/ ) (Replying to @yudapearl) and hate-motivated crimes." Watch out @IlhanMN ! "anti-Israel rhetoric" will take up every neuron in your brain. I wish our University Administrators could gather the courage to include theses words in their meaningless condemnations of campus hostilities. Perhaps even
7.14.2021 9:23pm - (3/ ) (Replying to @yudapearl) include "Zionophobic rhetoric" on the list, to be perfectly clear where the virus rests? I remain hopeful.

7.14.2021 3:24pm - (1/ ) Not only is this quote true and penetrating, it is as important to articulate it verbally for the many Jews abroad who feel it in their heart but remain speechless when it comes to convey it to their peers, children or adversaries. Interesting, it echoes what Herzl said in the
7.14.2021 3:24pm - (2/ ) (Replying to @yudapearl) 1st Zionist Congress (Basel, 1897): “Zionism is a homecoming to the Jewish fold even before it becomes a homecoming to the Jewish land.” See []

7.14.2021 3:26am - (Replying to @richard_landes) I'll be missing George Galloway - a man of peace. 3:26 AM · Jul 14, 2021·Twitter Web App

7.14.2021 3:04am - (Replying to @HumzahAlkindi and @nntaleb) And your tweet is the best thing that can happen to an author, especially one who aspires to see a quantum shift in the way causality is taught. Thanks.

7.14.2021 2:04am - (Replying to @realRahulRK) The problem is indeed not HOW they protest but WHAT they protest: the very existence of a homeland for 8 million human beings. Academic boycott, of course, is a different beast but, again, it is the last item on their agenda.

7.14.2021 12:51am - Replying to @realRahulRK Boycott is the last item on BDS agenda. Their goal is defined by their founders, eg. Omar Barghouti see Their tactic to implant the equation "Israel = On Trial" in students mind.

7.14.2021 12:37am - (Replying to @DavidEspinoza01) So did Linda Sarsour, who would have become Secretary of State in Bernie's cabinet - a great joy for the Jewish people and all lovers of peace.

7.14.2021 12:33am - A "White Man's Burden" -beautiful description of the Oslo "peace process". When I visited Oslo, 2014, I couldn't understand why my hosts (all tall Norwegian scientists) would not stop cursing Israel in every breath. After few pints of beer it came out: We spoiled their generosity

7.13.2021 11:37pm - Cornel West Resigns From Harvard, Partly Blaming School's 'Hostility to Palestinian Cause'
7.13.2021 11:37pm - (Replying to @yudapearl) The "Palestinian Cause" has become a fig leaf for covering academic incompetence and moral bankruptcy. I have dealt with West's morality here: "The human mind is a funny machine, Dr. West". Today I wish you a happy retirement in some Zionophobic elsewhere.

7.13.2021 10:42pm - Jihad Slavery Abounds, 27 Years After Our New York Times Exposé:
7.13.2021 10:42pm - (Replying to @yudapearl) I know Dr. Charles Jacob personally, and I admire his integrity and relentless commitment to human rights. His description of the on-going slavery in Afrika and why progressive organizations allow it to strive is heart wrenching and outrageous. It begs for wide exposure.

7.13.2021 1:16pm - (Replying to @EricAuldSoclSci The classic paper is "knowledge compilations map" by Darwish and Marquis But Darwish has written many papers on this topic since. I also had some thoughts, at adolescence:

7.13.2021 6:41pm - (1/ ) Professor Evan Morris, of Yale, is trying an educational approach, assuming that Yale students are facts-&-reason-minded as they were before the Orwellian YCC propaganda Since Yale Daily News (campus newspaper) would not published his article, please
7.13.2021 6:42pm - (2/ ) (Replying to @yudapearl) forward it to any Yale student you know. It is a pity to see a highly reputable campus go up the flames of ignorance and herd thinking. I care, because I am sort of 2nd-hand alumnus of Yale (honorary doctorate, 2018), and would hate to see it slide down irreversibly. @Yale

7.13.2021 5:32am - (1/ ) This new paper in Psychological Review develops a computational model of how humans judge causation in physical situations It presents several difficulties to a reader of my background: (1) The acronym CSM is too close to SCM - inviting confusion.
7.13.2021 5:32am - (2/ ) (Replying to @yudapearl) (2) Hard to tell what the input & output are. Video of scenarios? (3) I keep on thinking about the Oxygen-Match-Fire problem; how would it answer the question: what caused the fire? (4) What's the role of data? My problems, true, but still problems.

7.13.2021 5:02am - One of the most moving video clips I've seen in weeks. The Knesset responds to @ShirlyPinto speech in her own language.

7.12.2021 11:59pm - (Replying to @JansenKornelius @JeanClaudeFox2 and 2 others) It's a really good book.

7.12.2021 11:57pm - The problem with University administrators is that they fail to appreciate the damage done to their students careers, once stigmatized as thoughtless morons, ready to swallow any false claim fed to them by propagandists such as those behind the YCC action. @Yale @SaloveyPeter

7.12.2021 11:31pm - (Replying to @RWJE_BA and @JadePinkSameera) It's a beautiful example. Theoretically, plausibility judgment is at the heart of every exercise in modern sensitivity analysis. Practically, it's the only argument that actually convinced a society of smokers to think twice.

7.12.2021 3:56pm - (Replying to @RWJE_BA and @JadePinkSameera) If we were to allow "evidence-based medicine" to rule medicine, the tobacco industry would still be ruling our lungs. They stopped in the 1970's because we listened to "plausibility judgment" (derived from Cornfield inequality) which, evidently, does not belong in pure EBM.

7.12.2021 3:25pm - (Replying to @ShMMor) This is sadly the heart of Israel's choice in this age of Orwellian morality. Do we prefer moral exile on physical exile? Some of my colleagues are so shocked by the former that they are willing to re-take the latter. Luckily, 95% of Israelis would not hear of "physical exile".

7.12.2021 2:06pm - Finally, a visionary with a coherent, humane, just and practical solution to the Israeli-Palestinian conflict. Who said 2-state is the only solution?

7.12.2021 1:51pm - See why I feel left out for not being at this Rally?

7.12.2021 6:37am - (Replying to @noatishby) Thanks @noatishby for retweeting this revealing post by @sumayaawad . But saying BDS is antisemitic takes away from its genocidal character, as expressed here
7.11.2021 11:30pm - Thanks for reminding us indeed that Palestinian call for boycott started 1936 by Hitler's confidante, al-Husseini, whose riots of 1936-39 compelled Britain to prevent Jews from reaching safety, sealing their fate. It was the first boycott-induced genocide in history. Hail to BDS!

7.12.2021 5:24am - Great discovery: a 3000 yr old inscription of Yerubaal (Gideon). It's like finding an old cousin whom we followed since 2nd grade, and for whom we sang songs of praise. Gee, he really exists, alive, and sends us greeting from the birthplace of our history.

7.12.2021 5:08am - (Replying to @RWJE_BA and @JadePinkSameera) I bet part of the reason you "de-emphasize" clinical experience and physiological theories is that they are not well formalized and, even if they were, most EBM people wouldn't know how to use it. This is what modern CI contributes.

7.12.2021 4:11am - (Replying to @RWJE_BA and @JadePinkSameera) If you rely ONLY on RCT you would be missing out on two of the three components of EBM: Physician experience and Patient preferences, both are based on judgment, not on RCT. Formalizing these two is what modern CI is all about, with lots of progress to show.

7.12.2021 3:35am - (Replying to @victorianoi and @BigThingsConf) A beautiful song. I first sang it in 1962, in my guitar class, in the style of Los Paraguaios, and it never left my lips.

7.12.2021 12:14am - Please help me retweet this reminder 20,000 times in memory of my grandfather who, in 1939, was prevented from joining us in Tel Aviv after British emissaries prevented Jewish refugees from sailing the Danube, on their way to the Black Sea and then to Haifa.
7.11.2021 11:30pm - Thanks for reminding us indeed that Palestinian call for boycott started 1936 by Hitler's confidante, al-Husseini, whose riots of 1936-39 compelled Britain to prevent Jews from reaching safety, sealing their fate. It was the first boycott-induced genocide in history. Hail to BDS!

7.11.2021 11:30pm - Thanks for reminding us indeed that Palestinian call for boycott started 1936 by Hitler's confidante, al-Husseini, whose riots of 1936-39 compelled Britain to prevent Jews from reaching safety, sealing their fate. It was the first boycott-induced genocide in history. Hail to BDS!

7.11.2021 10:53pm - (Replying to @JadePinkSameera) The DAG should come from domain experts, not from a methodologist like me, whose interest lies in learning general principles, not in specific diseases. It is only from canonical problems that we learn such principles. No toys -- No principles.

7.11.2021 10:41pm - (Replying to @HumzahAlkindi and @nntaleb) Whose nonsense? I don't have any!

7.11.2021 7:03pm - My My, this is the third time in my life that I kick myself for not learning Spanish. The first was when a learned to sing "Malagueña Salerosa". The 2nd, when I lectured in Madrid 1980's. And now, dying to learn how you pronounce counterfactuals in Spanish.

7.11.2021 6:30pm - His name was Yehuda Guetta, like my name at birth, Yehuda; it could have been me gunned down by an angry Palestinian, and it could have been my parents forced to see @RashidaTlaib pouring crocodile tears over bricks and stones w/o mentioning my name. His name was Yehuda.

7.11.2021 5:51pm - It was a hot day, so I am told by friends, but it did not stop the thousands who came from all over the US to declare: (1) No Fear and (2) We Love Israel.

7.10.2021 10:17pm - Please remind me what the problem was with Lord's Paradox that necessitated reliance on D-separation.

7.10.2021 9:37pm - (1/ ) (Replying to @JohanDH2O @ShMMor and @PeterBeinart) Fair question: What gives me the audacity to speak in the name of a community? I believe it is the almost total support my writings and speaking have evoked in the past 20 years on many aspects Jewish identity and history. E.g., True, this is still a
7.10.2021 9:58pm - (2/ ) (Replying to @yudapearl @JohanDH2O and 2 others) 2/ one man's voice. But for a community that does not have a Pope to excommunicate members, and that has had a painful history of false messiahs maligning its character for selfish rewards, consensual moral support may constitute sufficient authority for speaking in its name.

7.10.2021 9:26pm - Less than 12 hours remaining for the No Fear Rally, co-sponsored by the Daniel Pearl Foundation. For those unable to attend in-person, the rally will be broadcast via livestream on multiple platforms: You tube live: and Facebook

7.10.2021 7:32pm - (Replying to @JohanDH2O @ShMMor and @PeterBeinart) Please suggest less condenscending way of telling @PeterBeinart to stop using his Ex-membership in a community as a license to indict the community for crimes he enjoys imagining. How else can a community tell an Ex-member: "Stop speaking in our name."

7.10.2021 6:05pm - Spectacular Second-Temple era building unveiled in Jerusalem
7.10.2021 6:05pm - (Replying to @yudapearl) For an Israeli-educated youngster like me, who was brought up believing that the "Wailing Wall" depicts the essence of Jewish history, any discovery of "non-Wailing" normalcy becomes an exhilarating source of inspiration.

7.10.2021 4:53pm - This #NOFEARRALLY shows that Israel is back to being the most uniting force among American Jews, recognizing that violence and hostilities are but symptoms of a more serious disease, Zionophobia, that has spread to vital organs of society: media, academia and Congress. Be there!

7.10.2021 4:49pm - To readers who believe statistics aims is to visualize data, here is a masterful visualization. As to conclusions, to me it says that Cyprus is invitingly uninhabited.

7.10.2021 4:11am - (Replying to @PHuenermund @VC31415 and 2 others) The non-triviality in path-specific effects ( ) resides in identifying such effect from non-experimental data, not in computing effects once the functions are given.

7.10.2021 12:33am - (Replying to @VC31415 @causalinf and 2 others) Reading the main text, I'm trying to share your enthusiasm but can't understand the research question: "how much attribution would node 1 have received if edge (1, 2) had not propagated the change at node 1?" This sounds like "path-specific effect", isn't it trivial to compute?

7.10.2021 12:25am - (Replying to @martin_frasch @causalinf and 3 others) Many thanks for posting

7.9.2021 10:37pm - (Replying to @JamesGrace111) I have not seen any elaboration on this problem. But I have been so immersed in causal inference that my blindness to progress on probabilistic problems should not be taken as a reliable measurement.

7.9.2021 1:48pm - Thank you, @SenJackyRosen for you clear voice in combating anti-Jewish hatred and violence. Please permit me to add that condemning a symptom without naming the virus is hardly effective. There is a virus in DC that need be named: Zionophobia. @DavidHarrisAJC , @AJCGlobal

7.9.2021 1:28pm - Well put, @ShMMor , I have been waiting for someone to tell @PeterBeinart up front that he is no longer a *member* of the Jewish people. He has excommunicated himself by disavowing the one bond that tie us together: Jewish history and its dream for normalcy, culminating in Israel.

7.9.2021 5:49am - One of the thrills of working in causal inference as long as I have is seeing lingering problems resolved. One such problem,, on the complexity of extending a partially directed DAG has been waiting 30 years, and just got resolved:

7.9.2021 1:33am - (Replying to @ben_golub) Decryption=?

7.8.2021 11:34pm - (1/ ) The lively discussion concerning statisticians problem with SEM behooves me to qualify my earlier statement. Not *all* statisticians are confused about the meaning of SEM. The younger researchers I meet on the air have no problem accepting the causal interpretation of SEM as
7.8.2021 11:34pm - (2/ ) (Replying to @yudapearl) articulated in To encourage and assist these researchers I am retweeting my "SEM Survival Kit", written more that 20 years ago which, in view of recent postings, seems more relevant than ever.

7.8.2021 4:56pm - I wish I had a stronger background in game theory to appreciate this paper which connects Shapely Value to graphical causal models. Perhaps some of our economics insiders could enlighten us? @causalinf @VC31415 @lewbel @PHuenermund

7.8.2021 3:17pm - These debates will go on for another century, treating SEM as the elusive elephant in the parable, waiting for blindmen to opine: "I like this, and I hate this.." I ask: What is the mental barrier that prevents statisticians from seeing the whole elephant? One theory:

7.8.2021 1:38pm - We are in 2021; one hundred years after Sewall Wright used structural equations models (SEM) to encode