This report documents our conclusion that international collaboration in science and technology is a fundamental paradigm for advancing knowledge and solving global problems. It urges the United States to take a leadership role in advancing international collaboration, and presents new models to achieve this mission.

In the United States, the Administration and Congress clearly support international collaboration in science and technology. International collaboration is viewed as a good way to leverage U.S. investment in science and technology, to position U.S. industry to capitalize on research breakthroughs, and to train U.S. researchers to collaborate with foreign colleagues to solve global problems. Further, the free exchange of ideas is viewed as a positive force for democratic change.

A clear and visionary plan of action regarding international collaboration in science and technology can be an essential component of national policy to address physical issues (e.g., global warming) and socio-political issues (e.g., economic liberalization, drug trafficking, rising standards of living around the world, increase of democratic governments). International collaboration in computer science and engineering must be an essential component of this plan.

Advances in computer science and engineering, more than any other discipline, are responsible for the internationalization of science. Advances in these areas have created a global information network that supports distributed computing and collaboration. Future advances will reduce or eliminate remaining barriers to collaboration due to time, space and language, and support a global scientific enterprise. Quite simply, the disciplines of computer science and engineering are central to international collaboration. They are creating and advancing the essential infrastructure that is used to conduct the business of science.

In October 1997, scientists in areas of computer science and engineering participated in a workshop on international collaboration sponsored by the National Science Foundation. The goals of the workshop were to: a) understand the benefits and barriers to international collaboration; b) examine current policies and programs; and, c) propose new means to increase transnational research and realize greater benefits for the U.S. and participating countries from such activities.

The participants concluded that the benefits of international collaboration justify extraordinary efforts to increase and sustain international research projects. The main benefits of these projects are: (a) the creation of new knowledge and technologies; and, (b) their application to global problems that threaten health, peace and prosperity. The main barriers to international collaboration are the lack of coherent and consistent policies and plans to promote and support such activities. The report reviews the historical context that produced current policies. This is so current policies' motivations can be contrasted easily with the barriers they pose to international collaboration.

The report concludes that the U.S. must take a leadership position in creating mechanisms for international collaboration in science and technology. This would be a significant step toward achieving many national goals. Specifically it would contribute to: a) developing a more secure, healthier, wealthier (prosperity depends on clean air and water, economically viable neighbors), and better-educated citizenry; b) advancing scientific knowledge; and c) finding solutions to global problems.

The United States has one key agency that is well positioned to lead this effort. It is the National Science Foundation. We conclude that the computer science and engineering community should be challenged to work with the NSF, and to form bridges from it to other agencies, to create programs leading to increased international cooperation on science and technology projects.

The report proposes several models to stimulate and sustain international cooperation in science and technology, namely:

International Science Foundation
The research community and NSF and should establish an International Science Foundation with participation by all nations. The new institution should have as its main goal advancing knowledge and developing technologies to solve global problems. A single organization with this mission can help remove many of the current barriers to international collaboration inherent in bilateral relationships, and develop standard procedures to initiate and sustain multinational projects.

New Computer Science Division for International Collaboration
The National Science Foundation should establish a new division within the Directorate for Computer and Information Science and Engineering (CISE) to support international collaboration. This would underscore a new national commitment to a world-orientation in building computer and intellectual infrastructure. Substantial budgetary resources should accompany the new division. Plans should ensure a focus on significant international goals where achievements are possible due to advances in computer networking. [A new CISE directorate is the appropriate administrative step because computer science plays an essential role in developing the infrastructure and tools that support the global scientific enterprise.]

International Super Centers
International Super Centers should be formed to solve specific global problems. These would be virtual collaboratives focused on physical centers of excellence. Associates would use high bandwidth networks to communicate. New technology will be in place here to enable the worlds best scientists and engineers to work together to solve global problems. [Developing them will provide computer scientists and engineers an invaluable testbed for design and test of distributed networks; distributed computing; and, collaboration technologies.]

Develop Next Generation Infrastructure and Tools
Fund explorations to advance the tools and infrastructure for international collaboration. [International projects provide the best possible environment for computer scientists to propose and evaluate advances in distributed computing, communication and collaboration.]

Expand Existing Programs in Latin America and Elsewhere
Increase the amounts for, and durations of, collaborative research projects. In developing and economically depressed countries, the U.S. should subsidize joint research projects. [We include a general conclusion that investment in international collaboration is the best form of foreign aid.] Move current NSF funding for joint research projects from modest levels [designed to stimulate initiation of collaborative research projects in new areas] to more substantial efforts. To attract the best researchers in the U.S. and other nations these programs must include follow-on funding with a joint review scheme to sift out the most successful projects.

The radical changes in information and communication technologies sweeping the globe during the past decade are having a profound effect on the way people learn about and interact with the world, on the way business is conducted, and the way scientists and engineers do research. We are moving in rapid steps toward an age of a single, global economy and a single, global scientific enterprise.

Although professional and national competition is as vital as ever, cooperation and shared goals are gaining increasing importance. Evidence shows that international scientific collaboration leverages national investment, and can produce great benefits. International efforts to protect ecosystems and the natural environment recognize and underscore the importance of setting objectives higher than individual gain in order to provide for societal gains. This spirit of international cooperation is reflected in common market efforts, trade agreements, international recognition of intellectual property rights, and international efforts at educational reform.

The Administration has emphasized the importance of international collaboration in recent policy statements. In his address to the United Nations on October 12, 1998, Vice President Gore proposed five new challenges in a "Declaration of Interdependence". He challenged the world community to:

• improve access to technology so everyone on the planet is within walking distance of basic telecommunication services by the year 2005;
• bridge language barriers by developing technologies with real-time digital translation so anyone on the planet can talk to anyone else;
• create a global knowledge network of people working to improve the delivery of education, health care, agricultural resources, and sustainable development, and to ensure public safety;
• ensure that communications technology protects the free-flow of ideas and supports democracy and free speech; and
• create networks that allow every micro-entrepreneur in the world to advertise, market, and sell products directly to the world market.

Today, we can only glimpse the changes that advances in information and communication technologies will bring. But as scientists who are involved in the development of these technologies, we need to look ahead and begin to create frameworks that will be successful in guiding and managing these changes, and harnessing their potential to advance science to meet global objectives as we enter the next millennium.

While the United States maintains a leadership position in many areas of information and communication technologies, federal agencies supporting computer science and engineering research have been slow to consider and implement new policies to support international collaboration. Given the global changes in communication, commerce and collaboration driven by the growth of the Internet, now is an excellent time to consider whether new policies are necessary, and if so, what issues an integrated, comprehensive policy must address.

As a start toward understanding the internationalization of science and technology, and specifically of research on computer, communication and information science and engineering, the NSF funded a Workshop on International Collaboration in Computer Science (WICCS9297). The goal of the workshop organizers and participants was to arrive at a deeper understanding of the benefits of international collaboration, the costs and risks associated with these activities, and to recommend ways to increase opportunities and optimize future benefits.

• What are the main benefits, barriers and risks associated with international scientific collaboration with respect to individual investigators, to the interests of the U.S., and to the global community?
• What does the global landscape of international collaboration look like today, and what might we anticipate in the future?
• What are the existing opportunities for U.S. scientists? What is the special role of computer science and engineering in international collaboration?
• What steps should the community in general, and the NSF in particular, take to guide rather than be swept up (or aside) by the forces of globalization?

This report summarizes the main insights and recommendations of the workshop. It also responds to recent developments subsequent to the workshop including new international initiatives, and recent activities of the U.S. Congress and White House. The report also includes many beneficial contributions by members of the research community who did not attend the workshop.

The main conclusion of this report is that international collaboration in computer science is vital to our national interest. This is so because international participation is required for solutions to global problems that threaten our welfare and security. Indeed some global problems could even impact the long-term survival of humanity.

Moreover, computer and information science and engineering provide infrastructure necessary to conduct the business of science. Today the essence of harnessing diverse international talent is computing and communications. Therefore the report recommends that the CISE community be challenged and given means to take leadership in promoting international collaboration in computer science and engineering. International CISE collaboration must include sufficient investigators to positively address those global problems whose solutions depend on advanced CISE technologies.

Progress toward these goals can be materially advanced by NSF activity. For example, a programmatic assertion that international CISE collaboration is of highest priority would be very significant. Similarly, the process of supporting international interaction would benefit greatly from definite standards, new research funding initiatives, and significant programs at the level of major NSF divisions. These should be focused on CISE topics, and oriented to stimulating and supporting international collaborative CISE research.

The remainder of this report is organized as follows. In the next section, we briefly review the changing landscape of U.S policy regarding international collaboration in science and technology, and the nature of the NSF=92s participation in support of international collaboration in CISE. In the following two sections, we explore the many benefits of international collaboration in CISE, as well as the barriers to international collaborations. In the final section, we submit a set of recommendations for increasing and optimizing international collaboration, while overcoming some current barriers to such activity.

Until quite recently, science and technology policy in the United States was driven by national security needs. The critical role of science and technology in winning World War II underscored the importance of national investment in science and technology. In a 1993 report, the House Committee on Science provides this summary:

In 1997, the Speaker of the House of Representatives charged the House Committee on Science to develop a new science and technology policy for the U.S. The letter included the following motivation for this task:

In Computer Science and Engineering, the vast majority of international collaboration occurs at the level of individual investigators, or through collaborative relationships between research laboratories or academic institutions. The NSF's Division of International Programs (INT) coordinates international programs for graduate students, postdoctoral and junior investigators; joint research projects for more senior scientists and engineers; and supports planning visits and international workshops. In many cooperative research projects INT works with other NSF research divisions to support internationalization of projects supported by those divisions.

The NSF's INT division provides a vital role in stimulating and supporting international collaboration in computer science. The focus is on exploring collaborative opportunities through workshops and visits, and supporting initial collaborative partnerships for several years at the level of an individual investigator grant. INT's goal is to stimulate international collaboration, and to fund new cooperative projects for a period of two or three years. Activities supported by INT require direct cooperation between U.S. scientists and foreign scientists and their respective institutions, and cost sharing by each side. The exact requirements of joint research projects depend upon the bilateral agreements between the U.S and its partner countries.

In the past two years, the NSF=92s funding of computer science related research has increased dramatically, with major initiatives such as the $68 million Knowledge and Distributed Information initiative. For the first time, the NSF funding of computer science and engineering research exceeded funding by DARPA. Moreover, the NSF has sponsored an increasing number of programs encouraging interdisciplinary efforts, multiple institutions, and technology transfer from laboratories to classrooms. These new programs reflect changing national priorities, and the growing realization that advances in computer science and engineering are vital to our national interest. These new initiatives and programs have not included an international component.

The United States and European Union (EU) are currently working together through joint efforts to support international collaboration in digital libraries and multilingual information management. Both the U.S. and EU support substantial research efforts in these areas, and are seeking to accelerate progress by leveraging on existing work and the many benefits of international collaboration.

It is clear that there is great interest in international collaboration in computer science by members of the U.S. and EU research communities in digital libraries and multilingual information management. These activities have culminated in an agreement between the United States and the European Union to support collaboration in multilingual language technologies.

New initiatives are also underway between the U.S. and Latin America. International workshops in Brazil and Mexico, sponsored in part by the NSF, have led to joint initiatives to support international collaboration in all areas of computer science. A joint NSF/CONACyT program with Mexico has helped support a dramatic increase in support of computer science research in Mexico, and has resulted in notable success stories resulting from joint research funded through this initiative. International collaboration between the U.S and Brazil is now supported through a joint NSF/CNPq initiative, which provides up to $200k of support for U.S. researchers to collaborate on joint projects of up to three years duration with their Brazilian colleagues.

The White House and Congress have identified international collaboration in science and technology as a national priority. The research community, through this report and other efforts, has an exciting opportunity to help identify new opportunities and models for international collaboration.

This section concludes that there is a compelling case for making international collaboration in CISE a national and international priority.

As the dominant species on this planet, it is our responsibility to support biological well being (ourselves and other creatures), protect future generations and foster global health and prosperity. Threats to our health, prosperity and future include: terrorism; drug-trafficking and abuse; compromises of computing infrastructure integrity and security; international monetary instability; spread of AIDS and other diseases; poverty and starvation; natural resource depletion; environmental pollution; bio-diversity losses; global warming; and natural disasters. Each of the above ultimately impact everyone, and are therefore global problems.

The scientific community plays an essential role in solving global problems because its work leads to understanding their causes, and the ability to evaluate potential solutions. Such work consists of defining the problem space, collecting and analyzing data, and proposing and testing solutions. The best progress in this work requires scientists from many nations: global problems require global solutions, and global solutions require global participation.

Climate change is an example of a global problem that requiring widespread participation. The consequences of even small increases in global warming -- fewer than five degrees Centigrade -- are losses of life and property.

The applicability of CISE research to monitoring and managing global weather follows from the need to collect, process, understand, predict and evaluate (enormous amounts of) information at large numbers of world sites where unique resources (people, laboratories, sensors, computers, actuators, etc.) are located. The following lists ten sample areas of CISE research essential to world weather and other global tasks:

1. World-level networking. Size, heterogeneity, software demands and hardware requirements of large global networks present unique CISE research problems. (This is needed to enable seamless access to weather information across the world from both sensors and databases.)
2. Archiving and processing massive amounts of data. Coping with real-time data collection rates. (Processing accumulations of weather data is needed to address global warming, climate trends.)
3. Intelligent data retrieval. Means to support timely selection and distribution of data. (Obtaining weather information and directing it to the people and locations where it is needed.)
4. Distributed processing of information. Means to enable local processing and global sharing and analysis of data. (Weather information and its analyses possess value apart from its geographic source.)
5. High-performance computing. Means to provide the necessary computational power for detailed simulation beyond the ability of any existing supercomputer. (Needed for weather simulations across the world.)
6. Knowledge understanding and data mining. Methods to search, detect and explain phenomena from aggregated data sources. (Weather-relevant issues are present in the massive resources present in existing databases.)
7. Computer interfaces and language translation. Technology to allow people of varied cultures and languages to work and collaborate effectively. (This is essential to create world weather management infrastructure, and to support more effective air traffic control.)
8. Software engineering. Standards to enable worldwide software development. (Practices to support interfacing software modules developed in different countries, particularly for issues that bridge political borders, e.g. weather, natural resources, and biology.)
9. Network-based machine operation. Means to support remote operation of sensors, actuators and instrumentation from anywhere in the world. (Seismological and weather measurements require new network modalities.)
10. Intelligent sensor and instrumentation technology. Means to allow data to be collected anywhere and anytime. The range of such technology goes from the extremely small (nanomachines) to the uniquely large (e.g. arrays of large antennas and satellites) and encompasses a multitude of CISE topics. (As in 9., weather and other issues require new modalities.)

The above contains central CISE topics. Research activities on these themes are under way in universities, industries and national research laboratories around the world. However, global warming presents problems of unprecedented scale involving many millions of information sources networked across the world. Each source could provide a massive and continuous stream of data. Likewise, the system could require feedback control in ways that could depend on the evolution of the data. The complexity of coordinating, managing, maintaining and programming such an infrastructure dwarfs that of any system in existence today. To collect, store process and manage these data requires infrastructure involving large constituencies and capable of surviving geopolitical phenomena. The global warming problem can be solved only through novel structures ensuring world participation of specialists in computer and information science.

In the United States, the European Union, Japan, Korea, the former Soviet Union, and to a lesser extent elsewhere, governments and industries have spent billions of dollars advancing information and communication technologies. Joint initiatives have been created that combine the resources of government, industry and academia, and massive infrastructure has been put in place. Researchers have been educated and trained to solve problems in computer science and engineering. International collaboration can leverage this investment, building upon the infrastructure developed, the research breakthroughs achieved and new technologies developed by individual nations and groups of nations.

International collaboration builds upon the education, training and experience of tens of thousands of engineers and computer scientists, the lessons they have learned, and the discoveries they have made. The diversity of the talent pool produces synergistic results: Scientists from different countries approach problems with diverse skills, attitudes and perspectives that stimulate novel approaches and creative breakthroughs. Providing scientists from different countries the opportunity to work together not only increases the talent pool, but also changes and enhances it. The potential synergy of the hybrid talent pool could greatly accelerate the advancement of knowledge, the creation of new technologies, and the solutions to global problems.

Current NSF CISE initiatives, such as Digital Libraries, Educational Innovation and Knowledge and Distributed Information (KDI), will create new technologies, tools and paradigms for scientific collaboration and discovery. International collaboration extends the scope of these activities to overcome challenges of physical distance, time zones, languages, cultures, computing environments, etc. Moreover, the scientists in different countries who collaborate to solve global problems test the effectiveness of new tools and technologies. Thus, international collaboration is a valuable laboratory for research, development and evaluation of new paradigms for scientific discovery and collaboration.

In addition to their role as participants in scientific inquiry, computer scientists and engineers play a special role in science and technology: they provide the infrastructure and tools used by the scientific community to conduct research and advance knowledge. This includes the infrastructure and tools used by the international community to collaborate and solve global problems.

While international collaboration in computer science is happening today, it is not now focused on study and solution of coordinated global problems. Better tools, infrastructure and focus can greatly improve our chances of understanding and solving global problems. There is thus a compelling need to overcome individual, group and national barriers to promote and support international collaboration in computer science and engineering.

New paradigms for collaboration and discovery require new infrastructure. Understanding the infrastructure needs and implementation strategies in an area that is evolving as rapidly as computer science and engineering is itself a major research challenge. It is both inefficient and risky to address problems of distributed computing and collaboration locally, and hope that these solutions generalize to support international collaboration. Similarly, in disciplines other than computer science, it is equally inefficient to develop international cooperation mechanisms and policies without any involvement by the CISE community and hope that they are the most efficient in leveraging information processing and communication techniques.

Projects that require international collaboration provide necessary training and experience for scientists, engineers and managers who must compete in an increasingly global marketplace. The graduate students who participate in global projects will gain competitive advantage, both for themselves and the companies that will employ them. These fortunate students will master the complexities of working in distributed virtual environments with people in other cultures; gain valuable insights about the limitations and capabilities of current technology; and discover opportunities for commercialization of new technologies. If the U.S. is to maintain world leadership in information and communication industry, it must train students to work and compete in a global arena.

In the past five years, the high technology sector has contributed the largest share to our national growth, about double the national average. Growth in high technology is perhaps the most prominent among the factors contributing to recent U.S. economic prosperity.

A main tenet of U.S. international science and technology policy is to promote our economic well being. "While we must remain ever vigilant and militarily strong, the need to maintain economic strength has taken on primary importance today. We now recognize more clearly than ever that economic strength facilitates not only a strong defense, but promotes other societal needs, such as social and political stability, good health, and the preservation of freedom." (1998 House Report, page 4)

The Asian financial crisis underscores the extent to which the economies of individual nations are interrelated. International collaboration in computer science, computer information, and computer engineering offers a potential positive force for global financial stability by enabling developing nations to become both consumers and producers of high technology.

Supporting foreign research provides an excellent way to exercise U.S. policy abroad. Supporting research and education in economically depressed and developing countries could represent one of the least controversial and most efficient means of foreign aid. Such support would reach some of the most respected and influential people in targeted nations, provide means for them to collaborate with U.S. researchers, and contribute to the careers of globally aware and U.S.-friendly leaders worldwide. Foreign aid in the form of direct support for research will help less developed countries build centers of excellence, computing infrastructure and industries, and help bring developing countries into the information age. This aid can take the form of computing equipment and software produced by U.S. companies, thereby establishing markets for the U.S., and providing motivation for U.S. industry to participate in foreign research initiatives .

U.S. support of research in foreign countries through international collaboration is good foreign policy. Many countries economically depressed and developing countries have highly capable and motivated scientists and students who lack the resources to achieve their potential. Scientists in these countries are unable to participate in large international projects because they lack infrastructure, tools and training. With the ever-decreasing costs of computers, and the ability to collaborate over long distances using computer networks, relatively small investments in foreign research can produce enormous returns. Given appropriate resources and guidance by their colleagues in the U.S., scientists in these countries can establish viable research and development efforts, enabling them to participate in future multinational projects. In addition to producing advances in science and technology, investment in foreign research increases the hybrid talent pool, builds lasting professional relationships and earns good will for the U.S.

The impending transition of most of the world=92s nations into a global information society brings with it an urgent need for engineering leadership in the definition of standards, international protocols and mechanisms for trade, commerce, intellectual and cultural exchanges. In case the U.S. does not provide this leadership, other technologically advanced countries will seize this unique historical opportunity and reap the benefits.

Historically, information technologies have been strongly influenced by scientific discovery and interaction. Scientists often blaze the technology trails that industries later leverage and expand into publicly accessible infrastructures. This paradigm of technology transfer and the fact that in order to lead in technology deployment it is essential to be involved in its development are strong arguments for the early (pre-competitive) involvement of US scientists in international CISE research efforts.

The benefits described in this section indicate an urgent and vital need to accelerate international collaboration in computer science, computer information, and computer engineering. Providing the means and motivation for scientists to collaborate effectively across national and political boundaries will create the infrastructure and synergies needed to advance knowledge, create new and improved technologies and provide solutions to coordinated global problems that cannot be solved locally. In addition, international collaboration will train U.S. researchers to participate in global projects, and position U.S. industry to both provide and participate in the development of new technologies to support international efforts.

The main barrier to international collaboration in computer science and engineering is lack of U.S. policy that promotes and supports international cooperation. Although international collaboration is now identified as a national priority, policies and programs have not yet been put in place to translate this priority into a plan of action. For example, within the National Science Foundation=92s Computer and Information Science and Engineering (NSF CISE) directorate, procedures for funding international collaboration differ from country to country, when such opportunities exist at all.

International collaboration in science and technology is the province of the Department of State. Congressional testimony by several experts noted that State is poorly provisioned to promote international collaboration or take advantage of opportunities that arise. According to Congressional testimony, the Department of State has failed to provide the necessary vision, leadership and support for international collaboration in computer science. The 1998 House Report cites testimony that "scientific expertise and commitment is severely lacking within the Department of State." Admiral Watkins noted that "leadership there always seems to be lacking in both timely enthusiasm and technical qualifications."

International Collaboration in CISE has a relatively weak constituency within the NSF. Programs within CISE are driven by a constituency of managers and researchers who support and lobby for the importance of their programs. Each program has a clear focus for advancing knowledge and creating new technologies in areas of national importance. Given this organizational structure, it is difficult to motivate support for international collaboration, since funding is perceived as a "zero sum game," with support for international collaboration necessarily eliminating support for other programs. Supporting international collaboration is robbing Peter to pay Paul, and Peter is better armed.

Support for international collaboration is also relatively weak in the U.S. computer science academic community. It was surprisingly difficult to get scientists to participate in the workshop on international collaboration in computer science. Fewer than half those invited to WICSS agreed to attend. Many who agreed to attend dropped out in the weeks before the workshop. In contrast, attendance at the 1997 NSF Interactive Systems Grantees Workshop (organized by one of the authors of this report) was about 95%.

Researchers understand where the "action" is within the NSF, and invest their time in activities that promise to deliver the greatest amount of funding. The NSF divisions in CISE science have relatively large budgets to support research, and relatively small budgets to support international collaboration. The NSF=92s approach to international collaboration provides seed money to stimulate collaborative activities, such as travel, workshops, visits, postdoctoral training. A limited number of collaborative research projects are also supported at modest funding levels. Because the NSF does not provide resources to fund major efforts involving international collaboration, these efforts are not taken seriously by the research community. Whether intentional or not, the NSF sends a message to researchers in computer science and engineering: International collaboration is encouraged but is not a priority.

Funding of research in computer science and engineering in the United States is guided by national priorities, such as the desire to maintain technological superiority in areas of computing and communication. National priorities are often in conflict with the outcomes of international collaboration=97shared knowledge and technology=97since research advances and new technologies are no longer "owned" by the U.S. This conflict is compounded by industry participation in federally funded projects (stimulated by agencies=92 requirements of matching funds), since industry participation is often motivated in part by the opportunity for competitive advantage in the marketplace. These same barriers to international participation are present=97to greater or lesser extent=97in all countries.

For international collaboration to be truly effective, collaboration must occur between national governments and their agencies. Even when all parties share the same vision, existing rules and regulations (immigration, taxation, etc.) complicate the process. Government bureaucracy may be frustrating to researchers and program managers within the U.S., but the effects of government bureaucracies attempting to work together on international programs are truly enlightening.

The Workshop on International Collaboration in Computer Science began with a session entitled "The Good, The Bad and The Ugly," in which participants shared feelings and stories about international collaboration. A clear consensus emerged from our work that the benefits of international collaboration outweigh the difficulties and hardships often encountered. U.S. researchers' motivations range from the desire to collaborate with the best people in the world, to the desire to help talented colleagues in less prosperous countries who cannot realize their creative potential because of insufficient resources.

There are significant barriers to international collaboration in computer science. Because there is a clear, important and urgent need to increase international collaboration, the computer science research community seeks to facilitate international collaboration, and to remove current barriers to such activities. This will require proposing, examining and testing new models for international collaboration. We present a set of recommendations for consideration. These share the common theme that the goal of new programs and models is to dramatically increase opportunities for international collaboration.

The research community would like the NSF to consider establishing a new division in CISE: International Collaboration in Computer Science and Engineering (ICCSE). Having a CISE division devoted to promoting and supporting international research efforts removes the most serious barriers to international collaboration. The division would have significant resources; would develop a strong constituency with the CISE community; and would become an effective partner with the NSF=92s International Division.

Establishing a new division within CISE would demonstrate the high priority, commitment and focus that international collaboration requires. Divisions within the NSF have status and stability. Formation of a new division within the NSF CISE is a significant event; it underscores the importance that the U.S. and its CISE research community assign to international collaboration. It sends a clear message to other governments and research agencies that the U.S. assigns high priority to initiating and supporting international collaboration in computer science, and establishes the NSF as the point of contact for initiating new programs.

An appropriate mission of the new division would be would be to serve the national interest by advancing scientific knowledge in computer science and engineering through international collaboration. The division, for example, could promote and coordinate establishment of international "super centers" to solve global problems. Additional goals might be to support training of U.S. researchers through participation in international projects; to help establish global computing and communications infrastructure for international collaboration in science and technology; and to support the development of new technologies for solving global problems through international collaboration.

An important activity of ICCSE would be to seek out opportunities for international collaboration in computer science. Having NSF personnel working in different regions of the world would do this most effectively. NSF should consider opening new offices in different regions of the world, or assigning NSF representatives to embassies in areas where U.S. participation in science and technology is lacking and needs to be increased.

Experience shows that establishing joint initiatives between countries requires a great deal of time and effort. Champions must be identified within the funding agencies and research communities in two countries, workshops must be proposed, organized and funded, and the many details of a joint program worked out. Given the time and effort involved, it is probably not practical for the U.S. to establish joint funding initiatives with all countries. Even if this could be a done, joint programs exclude collaboration by scientists from more than the two participating countries. Establishing programs between three or more countries may be possible, but the difficulties probably increase exponentially, and political sensitivities may arise if other countries feel that they are being excluded.

Because a policy of forming bilateral agreements is inefficient (and perhaps ineffective) in promoting internationalization of science and technology, the research community proposes formation of an International Science Foundation, with participation by all nations. This organization would work like the NSF, accepting unsolicited proposals from scientists, or proposals in response to programs and initiatives developed and offered by the Foundation. All proposals, however, would require collaboration by scientists from two or more countries. This model is already followed within the European Union where a variety of research programs have been established with the explicit requirement of international collaboration. These programs have been driven by the creation of the European union. However, they also illustrate the efficiency and effectiveness of an infrastructure that can stimulate, evaluate and fund international collaborative research while satisfying the political and scientific goals of the participating countries.

We believe that the notion of an International Science Foundation has great merit, and could overcome the major barriers to international collaboration. We also believe that this kind of organization will inevitably be created and the time is now for the U.S. to play a leadership role in its definition. In this context, the recommended creation of the ICCSE division is also an important step towards the establishment of an International Science Foundation. The foundations of operational procedures, mechanisms for pooling of international funds and awarding grants can be explored by the ICCSE and International Program divisions of NSF and then transferred to the proposed International Foundation. It is beyond the scope of this report to conceptualize this in detail. If the idea of an International Science Foundation is well received by the research community, we recommend that a separate workshop be organized to consider its principles and organization, perhaps with internationa l advisors.

Distributed international centers of excellence provide an excellent model for international collaboration. International super centers could be established to advance science and technology in areas of computer science, such as digital libraries, multilingual information management or virtual environments for collaboration, or to solve global problems.

The NSF should increase the number and scope of joint research programs in developing countries. While we have argued that bilateral agreements are not the most efficient means to promote internationalization of computer science, they can be very effective in specific countries, and should be pursued at increased funding levels until more ambitious multinational initiatives can be implemented. Establishing such programs requires identifying researchers in the U.S. and partner nations who are willing to organize workshops to promote joint research programs.

It is important to find ways to subsidize to foreign institutions to support joint research efforts. We have made the case that this is good foreign policy. In many countries, financial resources are simply not available; amounts as low as $30k to $50k can support students, postdoctoral fellows and faculty in other countries. Presently, this is very difficult to do within the NSF. This investment would enable U.S. investigators to utilize unique resources, and to leverage CISE funding. In many countries, small U.S. dollar investments can make a big difference.

The benefits of U.S. participation in joint research programs in developing countries are illustrated by the success of the NSF / CONACyT program. During the last five years, the NSF and Mexico=92s CONACyT have sponsored three workshops to promote collaboration in computer science between the U.S. and Mexico. Prior to these workshops, there was no division or program within CONACyT responsible for funding computer science. Mexican computer scientists were forced to submit proposals to programs supporting research in Chemistry, Biology, Physics, etc. The NSF / CONACyT workshops, organized by a Principal Investigator at UTEP, with the encouragement of CISE Program Officers of the NSF, helped mobilize the Mexican computer science community. At the most recent workshop, a representative of CONACyT reported that funding computer science is now a top priority within CONACyT. CONACyT has now funded the establishment of centers of excellence at several universities in Mexico.

We recommend that joint research programs, such as those established by the NSF with federal funding agencies in Mexico and Brazil, be initiated in other countries. The amount of the awards in these programs must be increased to attract top U.S. researchers, and agencies should find ways to provide follow-on support beyond initial grants. A seed grant is a good way to initiate collaboration, but there must also be mechanism to support successful collaborations, so they can realize their potential. Follow-on funding is necessary to attract top researchers to international programs. The model could be the successful SBIR program.

A model for supporting international collaboration among different countries or funding agencies that is reasonable, consistent and understandable by researchers and program managers world wide is highly desirable. Ideally, it would support training international research teams on the basis of scientific goals and complementary expertise. It would enable groups to seek funding for their work by submitting research proposals to a single agency with the ability to review and fund the proposal. The details of what agencies in different countries should be involved, their deadlines and requirements and other idiosyncrasies should be hidden from the researchers by the agency that receives the proposal.

Funding agencies have a major organizational role to play in this context and should reexamine and reform their inner workings and formal relations with foreign counterparts. Their international funding units should capitalize on the new interest and high priority placed in international collaboration and global efforts. In order to seize this opportunity, they must re-evaluate and revise policies to remove barriers and constraints that were imposed by Cold War attitudes and policies. They have a major responsibility and much of the necessary competence and experience in developing an international collaboration model that reflects the CISE needs in the 21st century.

Many in the CISE community do not know about existing international opportunities. Funding agencies should market existing international activities and opportunities more aggressively.

Funding agencies should actively encourage researchers to take advantage of opportunities for international collaboration when they can show that such collaboration is appropriate and desirable. In such cases, concrete measures should be used to quantify the benefits of collaboration.

The NSF should develop better reporting mechanisms regarding INT proposals, research results, the investigators and their international collaborators. The information should be collected and made available readily to the CISE community.

It is also important to make the CISE community aware of international activities, such as those of the European Union, including RACE, PACE, ESPRIT, etc. Some of these multinational programs require the participation of universities, industry and government labs, and serve as interesting models for international collaboration.

The importance of international collaboration must also be marketed within funding agencies. Leading officers should encourage and empower program directors to take advantage of existing mechanisms to stimulate and fund international collaborative research.

Further, we recommend that a small delegation of workshop participants make a presentation to leading officers of funding agencies to underscore the importance of international collaboration, and the recommendations herein.

Finally, a follow-on workshop should be held in about one year, to examine the outcomes of the first workshop, and to continue to find ways to support international collaboration in CISE.

Funding barriers to international collaboration could be overcome in several ways.

• Congress should be lobbied for additional support for international programs. The good news here is that 1998 House Report notes that "The importance of stable funding for large-scale, well-defined international science projects should be stressed in the budget resolution and appropriations process."
• By involving other divisions within the NSF such as INT.
• By involving other government agencies whose participation is needed to solve global problems such as the DOE or NASA.
• By enlisting the support of industry. Since U.S. software, hardware and communications industries are dominant in many areas, they are well positioned to provide the resources needed to support international collaboration, and to capitalize on technological advances.

International collaboration requires significant infrastructure. Development of infrastructure is essential to successful research. The development of tools to enable international collaboration is a necessary step.

The international networking infrastructures are essential for both CISE research and international scientific collaboration. Efforts are needed to investigate and develop new classes of applications that explore these infrastructures. Examples of such applications include tele-immersion and virtual collaboratories.

The potential benefits of international collaboration justify extraordinary efforts to produce new programs and initiatives, and to explore new models for collaboration and funding. Coordination of resources contributed by other countries and international organizations will have major impact on world problems.

We gratefully acknowledge the contribution of Dr. Rita Rodriguez, whose energy and vision motivated the workshop and final report. The report benefited greatly from revisions by Allen Klinger, and input from Jerry Engel, Robert Frederking, Hynek Hermansky, Eduard Hovy, Judith Klavans, Rick Schlichting, Don Towsley and Benjamin Wah.