Contents

Executive Summary
0. Introduction
1. Vision
2. Intellectual Mission of CIS
3. Possible Organizational Structures
4. Recommended CIS Structure:  A School Without Borders
5. CIS Funding
6. Possible CIS Components
7. Summary and Future Plans
Appendix 1 - Possible CIS Common Opportunities
Appendix 2 - Enabling Technologies
Letter from the Department of Computer Science
Letter from the Institute of Statistics and Decision Sciences
Endorsement of the Department of Computer Science Industry Advisory Board

Information Technology (IT) will have a profound impact on our lives in the coming century. The information revolution is well underway. Changes in how we live, work, create, and play will be pervasive, cutting across traditional boundaries and systems. In this dawning information culture, those who can bridge historical barriers are creating wealth and opportunities at an astounding rate.

As we look to the future, we see many challenges brought on by the IT revolution. Even today, there is a great shortage of information technology workers, particularly those trained across traditional disciplinary boundaries. Progress is limited by the rate at which we can research and develop new technologies, and these new technologies are creating societal impacts that challenge our legal, policy, and ethical systems.

Duke University is in a unique position to be a leader of the IT revolution. We have mastered education and research, foster interdisciplinary collaborations, and value diversity. We have a world-class medical center, renowned business and law schools, and significant strengths in the computation and information sciences in the Pratt School of Engineering and in the College of Arts and Sciences. At the present time however, the University does not have an organization that focuses these strengths on the issues of information technology. We believe IT will become so pervasive that a new organization is required to address IT educational and research issues, many of which occur at the boundaries between traditional Duke departments.

It is hard to imagine that any major university will not fundamentally reorganize its structure to address IT within the next few decades. Some visionary institutions have already made this bold move. The demand for academic participation in the IT revolution is high, and it is evident that financial resources are available to those institutions that can innovatively adapt. But the window of opportunity is narrow. The question for Duke is can we respond soon enough to capture the available investment opportunities and become a leading force in the information age, or will we become an also-ran that follows the lead of other institutions.

We propose that Duke University respond to the looming IT challenges by undertaking a broad, campus-wide study related to the establishment of a new organizational structure to meet the needs for education and research in computation and information sciences. In this initial planning document, we review possible organizational structures and propose a novel mechanism we refer to as a "School Without Borders" in Computation and Information Sciences (CIS) as a starting point for the study. CIS would have the necessary budgetary and planning authority to realize its academic mission, but without a separate admissions process. Undergraduate students, for example, would earn CIS-related degrees and take CIS courses from their homes in the College of Arts and Sciences or in the Pratt School of Engineering. We see the synergy provided by CIS as providing substantial value to the existing schools at Duke.

There are five goals for CIS presented in the proposal:

• To provide a broad-based critical mass for addressing the education and research challenges of the IT revolution.
• To be agile and effective in responding to needs imposed by rapidly advancing technologies.
• To reduce the duplication of effort and expense of addressing IT issues dictated by the current organizational structure.
• To be inclusive of those interdisciplinary activities and external relations necessary for leading academic institutions in the information age.
• To facilitate the understanding of the impact of technology on society.

If the University does not initiate a timely and effective response to the pending information revolution, it will find itself increasingly marginalized as other institutions capture the available opportunities and funding first. A bold move to establish CIS will place Duke on the leading edge of the IT revolution.

0. Introduction

Information Technology (IT) will be one of the key factors driving progress in the 21^st century--it is dramatically transforming the way we live, learn, work, and play. Advances in computing and communications technology will create a new infrastructure for business, scientific research, and social interaction. That infrastructure will provide us with new tools for communicating throughout the world and for acquiring knowledge and insight from information.

It is now quite evident that the information revolution is progressing at a remarkable rate. Computing and information technology has driven about 1/3 of economic growth in the U.S. since 1992, and the pace is accelerating. Many of the founders of new high-technology companies have not yet graduated from college, and new IT careers are appearing at established companies. The Internet and the Web are perhaps the most visible aspects of this change, but it is pervasive, touching nearly every field and discipline, from computational techniques in the physical and biological sciences, to new interactive media in the arts. Statistical issues, and the associated needs for new models, methods and algorithms, permeate many aspects of scientific, technological and social investigation. This revolution has already brought fundamental social change that surely will become even more profound.

Being an informed citizen in the Information Age requires knowledge of computing systems, global communications networks, and interactive information resources. The requisite level of knowledge goes beyond simply being comfortable with computing tools. It requires the ability to apply computational ways of thinking to design, to writing, to experimentation, to artistic expression, and to problem solving--to the very core of human intellectual activity. Just as higher education requires writing skills that go beyond the mechanics of sentence and paragraph structure, the information age requires scientific skills that go beyond the mechanics of programming and the use of software packages. In the information age, our ideas are no longer constrained solely by what is physically realizable, but by what is computationally realizable. For example, a chemist is now able to search more effectively for new compounds by modeling them before ever going into the lab, and engineers are able to design new lifesaving biomedical devices significantly more cheaply and faster by modeling. Nearly every discipline is changing, not just because of new tools but also because of new computational ideas and paradigms.

The rapid and fundamental changes of the information age pose significant challenges as well as opportunities. In academia, one pressing challenge for the faculty is to stay ahead of, or in many cases to catch up with, the students. More students are seeking to combine informatics with a liberal education in the arts and humanities, to prepare themselves for jobs that increasingly require both technical and creative skills. Another pressing challenge in academia is the need to attract and retain outstanding faculty in core computation, statistics, and information science disciplines. Students with bachelors and masters degrees in these areas are commanding starting salaries higher than what some tenured faculty members are paid. Moreover, these industry jobs often provide the kind of intellectual challenge that traditionally attracted people to academic careers. While universities cannot match the financial opportunities in industry, we can and must strive to create an environment that is intellectually more stimulating. Information Technology will be central to the concept of the university in the 21^st century. Universities are in the "information business", and so cannot help but be profoundly changed by a science/technology that is dramatically changing the processing of information. There will not be a "great" university with a less than great information technology capability. The grand challenge for universities like Duke is to discover how to teach and develop new conceptual models of the computational world.

New fundamental understanding is required to make optimal progress in the IT issues that are essential for solving critical national problems in areas such as fundamental science and engineering, the environment, health care, and government operations. A new administrative structure is needed that will allow Duke to lead the way in developing the computational and information sciences and to shape their myriad effects on society. We need to augment our knowledge base and to educate the workforce needed to realize the value of information technology. In this planning document we discuss the scope of what can be done in the information sciences at Duke and suggest ways to do it. We urge the University to begin a wide campus-based study of the issues and opportunities.

1. Vision

The well-accepted axiom that IT will permeate all aspects of society suggests that it is crucial for Duke to strategically address information sciences as a coordinated University-wide effort. By establishing an agile new structure, which for purposes of this document we refer to as Computation and Information Sciences (CIS), Duke University can make a bold statement about how to enable innovative cross-disciplinary efforts to address current and future IT problems.

It is difficult in the current environment to achieve the flexibility necessary to provide meaningful cross-disciplinary educational programs, since students are trained within the constraints imposed by traditional departments, and special arrangements to the contrary require substantial ad hoc efforts. The U.S. Government has studied the impact of the Information Revolution and has reported its findings in the Presidentís Information Technology Advisory Committee (PITAC) Report, issued on February 24, 1999. The PITAC report envisions a dramatic shift from the traditional discipline-specific degree programs provided by current university structures. More and more, companies (e.g., in biotechnology and geographic information systems) are seeking IT workers with expertise in more than one discipline or technology, and such training comes from collaborations that are difficult to maintain in the current environment.

At this juncture, the University could continue with the current practice of addressing IT needs by piecemeal departmental-specific approaches and by a proliferation of interdisciplinary centers and certificate programs. However, the cost and duplication of effort in this scenario would be staggering. Although centers strive to bring together researchers from different departments to address important research issues, the current lack of infrastructure and the rigidity of the overall system generally limit their impact, and ironically the effect of centers can be to compartmentalize researchers further. Those centers that do function well on a research level usually have a limited education component.

Duke could instead take the lead in shaping administrative structures to address compelling IT needs. Visionary universities such as Cornell, Carnegie Mellon, Penn State and Georgia Tech have already done so. We borrow much in this planning document from a recent campus-wide study at Cornell University (available at http://www.cs.duke.edu/~dept/CIS/Cornelldoc9911.html on the World Wide Web). The CIS structure we advocate for Duke would be well equipped to provide the coordinated long-term planning and infrastructure support needed for IT education, research, and external relations. CIS has the potential not only to coordinate those common IT research needs that transcend many traditional campus boundaries, but also to break down historical boundaries between departments and schools by providing a focal point for efforts revolving around the development and application of IT.

A well-organized CIS structure will facilitate collaborations by providing a coherent mechanism for infrastructure, planning, and development. That type of coordinated environment is crucial for interdisciplinary centers to fully realize their mission in harmony with individual departments. Through CIS, Duke will enhance its position to contend for major grants focused on interdisciplinary research and education. In a variety of ways, CIS will prepare a firm foundation for those cross-disciplinary research and education contributions of IT needed in the next century. Without this strong program in information technology, Duke faculty members' ability to compete for grant funding will be seriously impaired. Once the CIS forum for collaboration has been established, the potential for future interdepartmental contributions is greatly multiplied.

2. Intellectual Mission of CIS

The mission of CIS is to facilitate interdisciplinary programs that address those IT challenges common to many fields in science, engineering, medicine, business, law, and the humanities. There are great needs for transcendent IT research and education that can only be met by a proactive CIS. A crucial component is the development of the enabling technologies, which will not only provide the needed tools for proposed endeavors, but will also open new paths previously unseen. Duke is well positioned to take the next step to excellence in this fundamental domain.

Based on discussions of IT issues with many administrators, faculty, and staff from a wide variety of departments and organizations, we have identified the interdisciplinary opportunity areas in the table below as key challenges for CIS. This list is not meant to be exhaustive, but rather to illustrate the magnitude and scope of the area. The opportunities span the entire campus. The common intellectual threads running through them involve collaborative research in computation and exploration of massive amounts of information.
 

Common CIS Opportunities	Campus Groups Impacted
Computational Biology and Genomics	Biology, Chemistry, CS, Engineering, ISDS, Math, Medicine, NSoE
Modeling for Computer Simulation	Biology, Chemistry, CS, Cognitive Neurosciences, Engineering, ISDS, Math, Medicine, NSoE, Physics
Information Management & Exploration	Business, CS, Economics, BME, Libraries, ISDS, Math, Medicine
e-Business	Business, CS, Economics, ISDS, Law, Math, Medicine, Public Policy
Education Technology	Business, CS, Engineering, ISDS, Math, Medicine, OIT, Psychology, Trinity College, Center for Instructional Technology

In order to make progress in these opportunity areas, we have identified the following five enabling technologies as important areas of concentration:

• Networking & Ubiquitous Computing,
• Intelligent Systems,
• High-Performance Computing & Algorithms,
• Imaging & Scientific Visualization, and
• Computer Security & Privacy.

The intellectual aspects of CIS are its most crucial, and therefore we feel that it is important to elaborate on each of these five opportunity areas and five enabling technologies. We devote the entire Appendix to describing some of the research needs and cross-disciplinary impacts of each of opportunity areas (Appendix 1) and enabling technologies (Appendix 2). Although these issues are technical, they also bring to the fore important questions of ethics, privacy, security, and societal impact that relate to a variety of disciplines in the humanities and social sciences.

3. Possible Organizational Structures

We believe that Duke should create a central home for computing and information research and education, spanning the entire campus. Such a home would serve to bring together experts in CIS around the common opportunity areas and technologies described above in Section 2 and the Appendix. Such a home would provide fertile ground for emerging research and scholarly activities across campus. Such a home would further provide a framework for creating new courses, new concentrations, and eventually new majors to better serve the educational needs of our students--who increasingly seek to combine computation and information sciences with their disciplines of interest.

There are many possible ways to structure an initiative in computation and information sciences, but the optimum approach is clearly to establish an environment where there is a natural tendency for the "right" things to happen. In this section, we examine the possible mechanisms at Duke for a CIS structure. We draw heavily from a recent major campus-wide study at Cornell University on the "information revolution". Much of this section comes from the Cornell Report (available on the World Wide Web at http://www.cs.duke.edu/~dept/CIS/Cornelldoc9911.html).

There are a number of traditional ways in which to create such a home. Four broad classes of existing structural models are common at academic institutions:

A second critical problem with the department model is the necessity of maintaining the strength, identity, reputation, and visibility in those departments that are possible core activities. Adding a broad range of people to the departments that contain these core activities could easily be perceived as diluting their strength with "soft" or "applications" work. Thus we do not see the common home for broader activities fitting inside any single department. On the other hand, we do not believe that having multiple activities in departments scattered across the campus works well either, because of the substantial structural barriers to the development of broad collaborations and cross-disciplinary courses.

The main advantage for the research center model (#2) is that it can support cross-disciplinary research that would otherwise be difficult to undertake. One problem with this model is that it usually has an insignificant undergraduate educational component, and cannot easily fill this academic role. Some new academic endeavors naturally require starting with scholarship and teaching, rather than research efforts. At many academic institutions, and especially at Duke, there has been a proliferation of centers that often pull faculty in different directions. Except in a few notable examples such as the Pratt School of Engineering's Center for Emerging Cardiovascular Technologies, the typical result is that centers do not fulfill their chartered mission. The center model is not an effective model for Duke unless there is a logical organization within which the centers can operate as designed.

The division model (#3) could serve as an interesting interim approach, but in its common form--a management brokerage between chairs and dean--has shortcomings in visibility and effect. Only in a divisional structure with real resources and authority (e.g., budget, faculty lines, etc.) vested in the division director will it be possible to avoid difficulties in coordinated hiring, in setting directions, and in building integrated cross-cutting curricula. Unless a structure has resources to bring to the table it can get little accomplished in terms of building ties. Moreover, a division often serves to exclude those not within it, and our major emphasis is to avoid that. The lack of external visibility would inhibit the potential of major capital fundraising.

The main advantage for the traditional school model (#4) is that it is the de facto way of creating broad-based academic endeavors that involve both research and teaching. A college or school has external visibility, financial resources, faculty lines, and a structure for creating programs of education and research within it. The main disadvantage for this model is that it can be divisive. It is doubtful that a traditional school could truly reach across the campus, building joint programs with units from the arts, humanities, engineering and sciences. In practice a traditional school creates barriers to cross-disciplinary activities unless they happen to fall entirely within the school and much effort is needed on a case-by-case basis to cross these barriers. A second critical drawback with the traditional school model is that it creates additional complications for undergraduates. Moving between schools is difficult for students to do, and it is hard to imagine that this would change. Adding another traditional school would simply add to this difficulty.

In the preceding paragraphs, we have discussed various pros and cons of several traditional models, none of which are best for Duke. If Duke is to become a leader in the IT revolution, the new model we create must adequately address the following five organizational issues that have emerged from the previous discussion.

Specifications for Success:

4. Recommended CIS Structure: A School Without Borders

The more farsighted and agile alternative that we recommend is to establish a new innovative organizational structure for a School of Computation and Information Sciences (CIS), with a core of resident faculty and joint appointments of faculty from traditional campus departments. CIS would have budgetary and planning authority and would address infrastructure needs in a coordinated way. But we envision CIS as a school in a nontraditional sense--a "school without borders"-- in that existing schools like Engineering and Arts & Sciences would offer their own degree programs making use of the CIS framework. In this sense, our notion of a "school without borders" is similar to the sort of fluid structure at Cornell called the Faculty of Computing and Information. Cornell's new structure, like the one we propose, is designed to break down barriers and build synergy across campus in the computing and information sciences. Another possible model is the Carnegie Mellon University School of Computer Science which is highly respected for its outward view and creating close and effective ties with several departments, including statistics, philosophy, management and policy. Significant information science activities emanate from that consortial arrangement, including large NSF funded centers and new graduate programs. This is precisely the kind of development we're looking for in CIS. Though the CMU model does not include statistics and other groups in the school administrative structure, the interconnections and resource flows in these consortial activities are strong and semi-transparent. Not only are their intellectual themes similar to those we propose for CIS, but we also there are useful lessons to be learned here.

CIS could be the home of those core "units" with IT focus. Possible examples include the Department of Computer Science, the Institute of Statistics and Decision Sciences (ISDS), the Department of Mathematics, an Institute for Computational Science & Engineering, and the Center for Multi-Scale Modeling and Distributed Computing. At this point itís important to note that this suggested organization is only a proposal. Many issues related to departmental identities, the transition from the status quo, individual faculty concerns, etc. must be addressed and resolved at the University level prior to the development of an implementation plan. While we realize that the kind of structural change embodied in this proposed CIS is not easy to undertake in a university, we believe that the speed of change in the world makes it critical to explore new, more flexible and more inherently interdisciplinary structures. ISDS is already an inherently interdisciplinary organization, applying statistical methodologies to the solution of problems in many fields. There are valuable lessons here that can be expanded to encompass CIS operations.

The Information Revolution is likely to reorder the reputations of many major research universities. Those with the most effective response will reap enormous benefits as society reacts to this revolution. The goal of our recommendations is to permit Duke to enhance its excellent reputation in computing and information sciences. Like all distinguished universities, Duke must continually transform itself to adjust to rapidly changing realities.

CIS would also include centers to bring together faculty on timely research issues such as the Center for Geometric Computing, a new Center for Computational Biology and Genomics, a new Center for Information and Society, and a new Center for Technology Transfer. Within the relatively small organizational structure of CIS, centers such as those mentioned above have a good chance of realizing their full potential as centers of interdisciplinary excellence and can help develop needed interdisciplinary graduate programs that compliment and complement departmental programs. The successful model of the Pratt School of Engineering's Center for Emerging Cardiovascular Technologies will be very beneficial here. In contrast, the current situation at Duke is that the competing demands between departments and centers hinder the efforts of most IT-related centers, thus keeping them from realizing their mission. CIS will provide coordinated planning so that the core departments and centers will reinforce one another, and it will provide common infrastructure that is extremely difficult for individual departments and centers to justify on their own, much less sustain.

Many traditional application-specific interactions between participants would continue to occur as they have in the past, but with more emphasis and effect. CIS will focus its attention and resources on those IT issues that cut across traditional campus boundaries and where the potential impact is enormous. The concept is to bridge interdepartmental barriers in order to address those issues that require truly interdisciplinary effort. A "consulting" model will be included in the service mission of CIS and will provide an important campus-wide resource.

In addition to its research contribution, CIS will provide students with an interdisciplinary educational foundation unmatched anywhere else on campus for learning new skills that will be in great demand in the 21^st Century. We already see demand for these skills, but acquiring them requires tremendous personal initiative to overcome the rigidity of the current structure. The result is that very few individuals make this extraordinary effort. CIS will fill the interdisciplinary education void by producing relevantly-trained students much more effectively at both the graduate and undergraduate level. CIS is meant to be an organization capable of addressing rapid change by training students with the coordinated theory and skills needed to adapt to and ultimately to lead that change.

In order to broaden the educational opportunities for our students, we expect that CIS would develop and offer new undergraduate majors, minors, or concentrations that none of its individual units could manage alone. These new offerings would include areas such as computational science, statistical computing and algorithms development, intelligent systems, information systems and management, biostatistics and bioinformatics, cognitive studies, and other programs that fall within the focal areas identified above. We believe that the CIS should also play a leading role in developing introductory computing, statistics, and information sciences courses that serve the needs of students at large. This program would have the goal of bringing fundamental ideas from computation and information science to students in all disciplines, while at the same time respecting the different educational contexts of the individual disciplines. New courses developed along these lines could play an important role in Curriculum 2000, but no single department would have the scope or resources to undertake the effort alone. These activities will be coordinated by the new Center for Technology Transfer (CISTech) that we discuss later.

CIS will also open up new opportunities for graduate education, especially in computational science and engineering, Bayesian statistics theory, bioinformatics, and intelligent systems. It is currently difficult to attract and matriculate new students in truly interdisciplinary areas since the requirements they must satisfy are the union of the requirements of two or more departments. New joint graduate programs will flow naturally as a result of enhanced CIS research collaborations and the coordinated planning mechanism. Industry is especially concerned with broad interdisciplinary issues and wants desperately to be aligned with universities in their academic training programs. Industrial interaction will be a key component of CIS through CISTech, where we can take advantage of our current experience in industrial relations at Duke. We expect as a result significantly new collaborative opportunities and models for graduate education involving internships and joint academic-industry oversight.

An exciting aspect of CIS is a new Center for Information and Society, which will provide an arena for a broad range of interdisciplinary studies in the humanities and social sciences. The center will provide researchers with direct contact to the scientists driving the technologies, and conversely these interactions can play an important role in guiding the development of new technologies. For example, machine learning and statistical methods will be key approaches for studying the effects of genes on individuals' traits and susceptibilities, and the same technologies can be used to "mine" relevant facts and associations from the massive data in electronic business applications. Both of these applications have many ethics and privacy implications that we need to understand, and conversely such an understanding will help the scientific potentials of such technologies to be realized in practice in a productive way.

In the remainder of this section, we look at the five requirements specified at the end of Section 3 and examine how the proposed CIS structure meets those needs:

5. CIS Funding

The creation of CIS would initially involve the transfer of the faculty lines of the core departments from Arts & Sciences to CIS using a formula arrangement. An advantage of having CIS as a school without borders, as opposed to one of the first three traditional structures discussed in Section 3, is the heightened potential for fundraising. Taking the lead in identifying information sciences as a fundamental school will increase visibility and position Duke well to raise funds to support its efforts. There are several possible donors in the information technology arena. A recent prominent visitor to the Department of Computer Science commented that a new building for Computer Science would have been an easy sell for a major donation, but that such a donation wasn't needed given the newly constructed LSRC. A new building or contiguous space to house the core components of CIS would be a major plus, and a major gift could make a new building or contiguous space happen. Endowing a new School for Information Sciences could be a compelling incentive for a major new gift. Recently, a number of major gifts have been given to interdisciplinary research organizations at other universities, but none in the area of computation and information sciences. There is a very narrow window of opportunity for this investment. If Duke hesitates, other universities will surely capture the available funding opportunities first.

The Department of Computer Science has an effective Industry Partners Program that it can expand to the school level to provide industry contacts and support. CIS will establish a forum for technical information exchange and supply relevantly trained students to a community that places very high value such training. Strategic investments by Duke in this unique resource will attract significant resources to Duke from both industry and government granting agencies.

6. Possible CIS Components

We list below a possible set of Core Units of CIS as well as the campus entities closely related to CIS. The diagram that follows gives a graphical picture of the overall structure and interactions. To maximize the contribution of CIS, this structure must be flexible to adapt to the specific needs of each particular interaction. CIS would manage an infrastructure to support effective individual department and organization interactions. A novel core component of CIS is the Center for Information and Society, which is designed to illuminate the societal impact of the IT revolution and to provide an intellectual arena for social science and public policy researchers to interact with and collaborate with CIS researchers. The Center for Computational Biology and Genomics would serve as a natural linking point with the Bioinformatics Consortium involving Statistics, Mathematics, and Computer Science.

Possible CIS Core Units (Departments and Centers):

Department of Computer Science
Institute of Statistics and Decision Sciences (ISDS)
Department of Mathematics
Institute for Computational Science & Engineering
Center for Computational Biology and Genomics
Center for Geometric Computing
Center for Multi-Scale Modeling & Distributed Computing
Center for Technology Transfer (CISTech)
Center for Information and Society

Sciences (Biology, Chemistry, Physics, Psychology)
Pratt School of Engineering
Medicine (Biostatistics and Bioinformatics, Genetics, Medical Informatics, Microbiology, and Radiology)
Fuqua School of Business
Social Sciences (Economics, Public Policy, Philosophy)
NSoE
Law
Humanities
Office of Information Technology (OIT) and Duke Libraries

Institute for Genome Sciences and Policy (Bioinformatics, Modeling)
Center for Cognitive Neurosciences
Center for Advanced Computers & Communications
Center for Nonlinear and Complex Systems
Bioinformatics Consortium

7. Summary and Future Plans

In this planning document we propose that the University begin to view Computation and Information Sciences in a comprehensive manner and as an innovative organizational entity. We have used the term "School Without Borders" because of the unique type of structure needed. Related notions, such as a "Faculty" as in Cornell's case, could also prove appropriate. Our suggestion is that the Computation and Information Sciences are pervasive now and will continue to be so.

We recommend that a University study be undertaken to address Computation and Information Sciences. We have proposed CIS as a starting point to guide this study. The committee should have broad representation from all parts of the campus. We see the following three major issues as important to address in such a campus-wide setting:

While the planning and implementation of any new academic structure takes time and careful deliberation, we believe that some kind of reorganization that recognizes the centrality of CIS is inevitable. If we begin to plan now, Duke can seize the advantage. Some universities are already ahead of us, and their best practices can be studied to help design a setting appropriate for Duke. Clearly there is excitement here to share with our alumni, trustees, benefactors, and friends. A timetable for planning and ultimately implementation will help convince everyone that we are serious and that the goal is attainable.

Duke must achieve an intellectual "buy in" by the entire academic community. In this initial planning study, we have just begun this process by touching base with a substantial number of possible participants. Initiatives proposed by other groups of faculty related to CIS will prove enormously helpful in refining the ideas presented here. We particularly need to articulate clearly our relation with and collaboration with the Pratt School of Engineering and that Schoolís new Dean. Important ties already exist between Computer Science and the Pratt School of Engineering, especially in Biomedical Engineering and Electrical and Computer Engineering; and these ties should be nurtured and expanded. There have been (limited) interactions between ISDS and the Pratt School, though ISDS has established and ongoing collaborations with biomedical engineers and radiologists in statistical image modeling, which is moving to a new level. Other established research collaborations exist between ISDS and signal processing groups elsewhere, in statistical time series modeling in communications signal processing and related areas (though not, seriously, with Duke EE). These are two of several relevant areas for potentially significant interaction. There are possible new engineering ventures in material science, optics, and maybe "engineering biology"; CIS can make a substantial contribution to these and others with the right collaborative arrangement. Similarly, in addition to forging intimate ties between the computing, statistics and mathematics departments and groups, the CIS framework will embrace renewed and stronger ties with various areas of the natural and environmental sciences, and with the real-world challenges of information technology in the medical domain. Ideally, the units of Computer Science, Statistics, Mathematics, and Computational Science would be closely aligned, either directly as integral CIS departments or at least in close proximity if not all were core units in CIS. The links to the Social Sciences and Humanities should also play a strong role in any new structure.

After an intellectual "buy in" has been achieved, it is natural to expect that a financial "buy in" is possible. Any reorganization implies financial rearrangements. Part of a new organization will require donor support; fortunately, the visibility of such an endeavor as is proposed will help CIS. There will need to be careful "formula type" negotiations since a separate undergraduate admissions process is not proposed. The budget process and the attendant responsibilities will need detailed articulation. Financial decisions in a university environment where there are many fundamental and exciting ideas are never easy; if the idea belongs to everyone, they are possible.
 
 

Appendix 1

In the following paragraphs, for each of the five opportunity areas identified in Section 2, we describe some of the relevant campus-wide IT educational and research needs and the cross-disciplinary impact. We discuss the five enabling technologies in more detail in Appendix 2.

Although these issues are technical, they also bring to the fore important questions of ethics, privacy, security, and societal impact that relate to a variety of disciplines in the humanities and social sciences. Such studies will have an intellectual forum in the new Center for Information and Society.

Computational Biology and Genomics: Advances in the computational aspects of biology will require bringing together computational, physical, and biological scientists into a stimulating and nurturing environment for research, development and training of a new type of scientist--one who can incorporate theory, simulation, and experiment to expand our understanding of modern biological problems. Students need an intellectual environment for considering problems that transcend traditional disciplinary boundaries, and they need training opportunities with mentors in different disciplines. Needs cover a significant range of areas including the broad categories of advanced computational methods & tools, biomolecular structure and function, imaging and dynamics, mathematical modeling of biosystems, and medical and genomic informatics. Opportunities are significant in the computational, informational, and biological challenges created by the increasing flow of information about genes and proteins identified by large-scale genome sequencing. There is great interest in developing the ability to use computational analyses to make biological predictions about the structure and function of genes and proteins.

Several existing ISDS projects are in areas related to cancer genetics, and can be classified as bioinformatics research. New projects in defining clinical phenotypes based on large-scale genetic expression profiles, and in plant genomics focused on protein expression, typify the kinds of research frontiers we currently see as defining the statistical and computational challenges in the next several years. In the near term, increased activity at the interfaces between DUMC and several Arts & Sciences departments--to generate new relationships, encourage collaborative research projects, develop co-operative funding mechanisms for new faculty and research personnel, provide the base for new graduate, postdoctoral and undergraduate programs--will provide the core bioinformatics infrastructure that will help to establish the road map towards a Genomics institute at Duke. The current initiatives in genetic expression analysis and plant genomics involving ISDS are examples of the kinds of activities that are needed on a much broader scale.

The information systems that result from this work will allow integrated analysis of biological data. Biologists will then be able to navigate through the complicated, highly-distributed bioinformation and obtain new insights. The genome annotation process, for example, involves the integration of highly distributed information stored in archival databases. There is great interest in developing the ability to make biological prediction from computational analyses so that predictions can be made about the structure and function of genes and proteins. Statistical methods, machine learning, geometry, and database methods must be combined together in order to accomplish accurate molecular modeling and to develop techniques to store and search large databases of proteins. No single discipline alone is sufficient to solve these problems.

Modeling For Computer Simulation: Modeling is an important component of the virtual laboratories of the future. Scientists can use their models to simulate the design and testing of new structures in a rapid and inexpensive manner. Modeling involves making assumptions about the data and then processing parts of the data to give the best answer possible to questions within the scope and validity of the model. For example, in mathematical modeling, we often assume that the data specify an equation set; then the task becomes the computer solution of the set of equations. Furthermore the solution to the equations might be hard to see, and often the solution must include computer visualization techniques. The study of making such a mathematical model, resulting in an equation set, and the study of the algorithms and hardware platform to solve the equations and the final output representation are important components of Computational Science and Engineering.

Equation formulation is not the only type of modeling. Other examples include statistical modeling, where the data are assumed to exhibit a particular probabilistic pattern, and geometric modeling, where the data correspond to a geometric structure. Much interdisciplinary statistical modeling is nowadays inherently based on simulation--deterministic and stochastic simulation of complex mathematical and statistical models, and utilizing simulation technology for model fitting and exploration. Computational and statistical issues and challenges in large-scale computer modeling cut across many areas of science, engineering and socio-economics--currently topical examples, and subjects of fundamentally interdisciplinary approaches, include large-scale simulation models in transportation networks, micro-simulations in demographics, and simulations of physical processes such as subsurface groundwater flows in porous media. Statistics is deeply implicated in such simulation modeling programs--in studying and representing relationships among possibly high-dimensional parameters that are input to numerical models, for synthesis of uncertain data from possible many different sources, for the identification of input-output relationships, for regularization and formal statistical solution of often radically ill-posed inversion problems in matching real data with simulation outputs, for inference and sensitivity studies, and--quite critically--for assessment and evaluation of simulation models based on often quite limited data.

Since there are many types of modeling, the subject of modeling for simulation (simulation meaning the computation required to explore the model) will of necessity become increasingly sophisticated. In a generic sense the scientist will work with an information technologist and formulate an "axiom system", which in working terms means the rules for engagement to attack the data and pry out the relevant information. The answers are, of course, only as good as the model, and hence experimental model verification will also play a large role in this computationally intense arena.

Information Management &Exploration: Scientific research, business processes, and government programs are generating ever-increasing quantities of data. Terabyte databases are common. Dealing with such massive quantities of data is an important and perplexing problem faced by many organizations. Both the movement, storage, and access to the data, and extracting and presenting meaningful information from the data require significant research. Ideally we would like to access, query, or print any book, magazine, newspaper, video, data item, or reference document in any language by simply clicking the mouse, touching the computer screen, talking to the computer, or blinking an eye. We would like to select among modes of presentation: data, text, images, or audio. Information can be referenced and derivations can be incorporated in many new ways, adding value and revealing insights through networked and software-enabled tools.

The transformation to this vision of the future requires significant improvements in data access methods and data mining techniques. High-performance file systems and tools will be needed to help individuals locate information and present, integrate, and transform the information in meaningful ways. Systems will require interfaces accessible both to experts and novice or infrequent users regardless of physical ability, education, or culture. Multi-modal human-computer interaction technologies are needed including speech, touch, and gesture recognition and synthesis. There are research requirements for topics ranging from network reliability and bandwidth to scalable software support and high-performance computing, and robust, reliable, and secure ways to deliver--and to protect--critical information. Challenging issues regarding dissemination of information in electronic form--including privacy, copyright, intellectual property rights, and realistic business models--remain important policy and research topics.

In addition to forming core challenges to statistical science, data exploration typifies problems of information science that are strikingly interdisciplinary. Data mining of large-scale data bases (in medicine, finance, marketing, etc) and internet resources has emerged as a subfield of both computer science and statistics, but that cuts across areas of mathematics, business and socio-economics also. The development of artificial intelligence tools in knowledge discovery in data bases and for manipulating and analyzing problems of uncertainty for informed decision making has led to a major subfield of networks/graphical modeling in computer science that is almost wholly based on Bayesian statistics. The computer science/artificial intelligence communities refer to this domain as Bayesian network modeling.

Future requirements for electronic medical records and health-system intranets will lead to increased reliance on the national infrastructure for communications, data sharing, and direct provision of care at a distance. Privacy and knowledge repositories are important research topics.

e-Business: The Internet and electronic business are revolutionizing traditional business processes. Companies can now be reached easily by their customers, regardless of location. They can receive immediate customer feedback and rapidly adjust marketing strategies or product inventories based on that feedback. Consumers can shop for the best products, services, and prices from the convenience of their hotel room, home, or office. Electronic purchases can be made securely, providing suppliers and retailers with immediate access to cash generated by sales, and providing consumers with automated statements detailing spending and purchases that allow for improved personal financial management.

Electronic communication is also dramatically changing how commercial transactions between companies are conducted, how digitally-based goods and services are distributed, and how retail sales are made. Companies are using information technology to get closer to their customers and suppliers. Technology is also helping to reduce paper work and purchasing costs by streamlining the acquisition process and by allowing companies to more efficiently find the best suppliers. Reliability of the communication networks, computers, and business applications are vital to the success of U.S. companies.

The use of IT, in particular the growing popularity of the Internet, and the emergence of global economic commerce has introduced a series of important and complex policy issues, such as privacy, intellectual property rights, the emergence of information "haves" and "have nots," and the impact of state and national regulation. For the most part, neither of these issues nor the resulting debates have been properly informed by leading IT research. Social scientists may be able to make important contributions to these ongoing debates by conducting empirical research on the impact of IT and the social policy necessary in the next millennium. Such research can help IT researchers identify potential technical solutions for addressing these policy issues (e.g. new metadata tagging standards and micropayment technologies for managing intellectual property).

Education Technology: In the not too distant future, any individual will be able to participate in on-line education programs regardless of geographic location, age, physical limitation, or personal schedule. Students will access repositories of educational materials, easily recalling past lessons, updating skills, or selecting from among different teaching methods in order to discover the most effective style for that individual.

In education, information technology is already changing how we teach, learn, and conduct research, but important research challenges remain. In addition to research to meet the scalability and reliability requirements for information infrastructure, improvements are needed in the software technologies to enable development of educational materials quickly and easily and to support their modification and maintenance. We know too little about how best to use computing and communications technology for effective teaching and learning. We need to better understand what aspects of learning can be effectively facilitated by technology and which aspects require traditional classroom interactions with the accompanying social and interactive contexts. In CIS, effective teaching will most likely consist of a combination of education technology and personal contact. This hybrid approach will take advantage of Duke's expertly trained faculty and is something no "electronic institution" can provide. Duke Libraries' Center for Instructional Technology is already exploring many of these issues, and we believe that the potential for a partnership between CIS and the Center will be a significant resource for the University.

We also need to determine how best to teach our students the powers and limitations of the new technologies and how to use these technologies effectively in their personal and professional lives. A workforce literate in information technology will be critical for ensuring that our Nation is prepared to meet the challenges and opportunities of the Information Age. Those who employ our graduates will certainly expect us to provide an education in information technology itself, including the ability to conceptualize and solve problems (not just teach how to use spreadsheets). We must emphasize the ability to adapt to and harness rapidly changing technologies of the future for both IT majors and for students who only want exposure to a few IT-related courses.

Appendix 2

In order to address the opportunities discussed above in Appendix 1, there are intellectual challenges for CIS in several enabling technologies. We describe in detail the five enabling technologies highlighted in Section 2. They are common to most if not all the opportunity areas, and it is important that they be addressed with coordination and collaboration. As a testimony to the importance of these enabling technologies, the Pratt School of Engineering is also planning to invest much of its future faculty development and resources in the same areas.

Networking & Ubiquitous Computing: The advent of the Internet and the World Wide Web has drastically increased the demand for network speed and bandwidth. We are rapidly moving to a world where every electronic consumer device costing more than $20 will contain an embedded processor and wireless networking capabilities. The hardware and software that connects future computing resources will be responsible for providing reliable, efficient, and scalable communication among the billions of devices (and people) with network connectivity. It is hard to imagine any field of endeavor that will not be increasingly dependent on the capability of such networks. As two compelling examples, consider emerging applications in telemedicine and e-business that require real-time access to complex information and high-definition images.

Ubiquitous computing provides mobile professionals with the tools and services that allow them to access and interact with information (and services) and to work more efficiently with their customers, colleagues, and enterprises. The main objective is to provide mobile users with access to the data and applications they need seamlessly, independent of a user's location or access device. One goal is to unify the multiple devices used to store various versions of critical information. No longer will users be forced to manually synchronize the contents of computers at home and work, laptops, pagers, cellular phones, and personal digital assistants.

To achieve this goal of universal connectivity, several unique characteristics of the mobile environment must be addressed:

• Most wireless links that serve mobile devices are high latency, low bandwidth, unreliable connections, especially when compared to office networks.
• Disconnected operation is still a necessary (sometimes desirable) mode of operation, even with a goal of ubiquitous network connectivity.
• Power is a limited and precious resource requiring careful management in an application-specific manner.
• The diversity of devices, networks, network connections, and operating systems negates a "one size fits all" solution.

The research challenges and opportunities that this vision of ubiquitous computing raises involve the Computer Science and Computer Engineering areas aimed at providing ezffective mobile devices, wireless protocols and infrastructures, systems support, and applications. It also invites serious investigations in the social science disciplines, economics, and business toward understanding and directing the influence of ubiquitous computing on workforce structures, management strategies, and societal relationships. Ubiquitous computing applications can have a profound impact in the medical and environmental fields where the needs for these applications should inform the systems development just as the emergence of new technical capabilities should inspire new applications.

Intelligent Systems: Intelligent systems embodies the development of sophisticated tools to augment the user in any intellectual task, the label "intelligence amplification" perhaps well capturing the intent and effect. The area of artificial intelligence (AI) within computer science is often thought of in this regard, but many areas and disciplines contribute to this general field, such as statistics, pattern recognition, image analysis, genomics, robotics, expert systems, speech recognition, character recognition, neural networks (and machine learning in general), and data mining. Areas where advanced tools using "intelligence" are in development or actually deployed are too numerous to mention. A small sample includes information retrieval, medical diagnosis, military planning and logistics, task scheduling, chemical analysis for drug development, models for protein folding, and many, many others.

One aspect of intelligent systems is to create performance normally associated with human intelligence. The methodologies that underlay these systems are twofold in nature. On one side we have methods based on now-classic AI paradigms, such as (for example) controlled search, rule systems defined using ad hoc, but studied, mechanisms, and use of highly nonhomogeneous data structures that treat procedures as subparts of data. On the other side are traditional methods such as logic, probability and statistics, stochastic systems, operations research methods (e.g., utility theory), and even differential equations used in creative ways. More and more, intelligent systems use the tools of a variety of disciplines. Another useful synergy is cognitive science, where considerations of how intelligence is realized in humans and other animals are a rich source of ideas, and in turn computational paradigms can shed light on human behavior.

The area of functional genomics will benefit greatly from intelligent systems that extract information from the huge quantity of genome sequencing data. A major problem is to infer which genes in a population of over 100,000 genes cause particular diseases or conditions. The machine learning field, which is a subfield of computer science and is closely related to statistics, has developed a number of computational paradigms that should be investigated for use in this endeavor.

Realistic models now are constrained by computational limitations, leading to dynamic research at the interface between statistics and computer science directed at solving larger problems, and providing real-time solutions. This is one computer science/statistics interface that offers immediate potential for collaborative growth and impact at Duke, and that has major longer-term potential. A related interface between statistics and computer science--and one of similar potential--is that between statistical data exploration and machine learning. Machine learning involves the development of deterministic algorithms addressing the generic "search for structure" problem--the very essence of statistical science--in high dimensional data configurations, and many of the tools are close cousins of statistical methods.

The effect of intelligent systems, perhaps first felt en masse with the appearance of expert systems, will continue to grow and probably accelerate in the coming decades. We must structure ourselves to take advantage of and accelerate this exciting and important area.

High-Performance Computing & Algorithms: In the next century, advances in high-performance computing will enable scientists and engineers to do their research in virtual laboratories, where they can perform their work without regard to physical location. All scientific and technical journals will be available on-line in digital libraries, allowing readers to download equations and databases and manipulate variables to explore the published research interactively. High quality teleconferencing will permit personalized interactions across continents. High-speed computers and networks are enabling a wide variety of scientific discoveries, ranging from mapping the human brain to modeling environmental climate change.

Modern computer systems and architectures are increasingly confronted with huge amounts of data and correspondingly must have huge amounts of memory and storage. Storage sizes through the years have been measured in kilobytes, megabytes, gigabytes, and terabytes. NASA's Earth Observing System produces petabytes (10¹⁵ bytes) of data each year. The immense volume of data that scientists are now gathering overwhelms our ability to efficiently process this data into meaningful information. This presents the Computer Science research community with fundamental challenges at all levels, ranging from algorithm design to computer systems and architecture.

Efficient use of such systems when dealing with massive data sets is a challenging task that requires understanding of memory storage systems, especially hierarchical memories and input/output communication. Further areas of opportunity include digital libraries, data warehousing, data mining, multimedia, and networking, which can bring together the memory-related aspects of databases, knowledge and data discovery, operating systems, and algorithm design. The applications and potential collaborations extend across many departments in the University. Statistics as a discipline has been radically inventive in developing new computational methods, such as stochastic simulation, but less adventurous in seeking new directions based on advances in hardware, distributed computing and operating systems tools. The time to address this is upon us, and this interface arena--that between computational statistics and computing science--holds great promise.

Imaging & Scientific Visualization: Significant advances are needed before we can meet the challenges of visualizing increasingly complex data sets. Future research must explore ways through which to model, represent, visualize, and interpret the information associated with complex structures and dynamic processes. The emphasis will be placed on physical systems and, particularly, on structures and processes that emerge in biomedical systems. The information of interest may reside in discrete, acquired data sets (such as the information contained in medical imagery) or may be created through mathematical models or symbolic representations (such as geometric models of structures or simulations of dynamic processes). Specific needs span several areas:

• Modeling: Creating graphical and/or mathematical models of the systems being studied.
• Vision: Exploring methods and algorithms for image processing, analysis, and understanding.
• Visualization: Investigating and developing representations that are visually meaningful, highly interactive, and as well as quantitative.
• Knowledge Discovery and Processing: Typically interdisciplinary efforts to extract, analyze, and present meaningful information from massive data sets.

Computer Security/Privacy: Research inprivacy and security are critical to ensure consumer confidence in electronic commerce. We need a better scientific base to build the links and nodes of the expanding information infrastructure that represent radically different scales of usage and traffic. There are many technical obstacles--issues of integration, privacy of data, security, reliability, ease of use, and ease of management--that stand between the current state of the art and the sort of global, ubiquitous, heterogeneous infrastructure required to grow to one billion users on the Internet.

Many of the barriers to realizing the benefits of IT are not technical, but rather occur when social, political, or legal issues from deployment and adoption of IT and its applications. For example, deployment of telemedicine is being slowed because of the need for state-by-state licensing of doctors. The use of geographical information systems (GIS) is hampered by the difficulty of sharing this data across organizational boundaries. Widespread use of electronic commerce will require mechanisms for creating trust between parties in an online environment. Social science research could shed light on these barriers and suggest possible solutions.

Future security/privacy issues will require research in the following areas:

• Authorization: How do you ensure that users are who they claim to be? How do different elements in the system locate and determine whether to trust one another? How do you enable new customers or business partners to access existing systems without major changes to existing security infrastructure? Whose identity should be used to determine authorization: the end user, the server, or some other entity?
• Asset Protection: Can you keep data confidential and private when stored and sent across untrusted networks? How can you be sure that the data doesn't change while stored or in transit?
• Accountability: How can you can tell who did what when? How can you ensure, and prove, that requests and results are not altered, inadvertently or maliciously?
• Administration: Can you define the security policy? Can you ensure that policies are consistent across all elements of applications, systems, platforms, and networks?
• Assurance: Can you convince yourself that the system keeps its security promises? How can you ensure that the infrastructure and application resources -- including systems, networks, and data -- are not presently under attack?
• Availability: How do you prevent attacks on elements of the system that cause disruptions in service? How do you design for fault tolerance and ensure that applications and data are restored in the event of a serious failure? How can you keep the system up and running continuously and also make needed modifications to the application, the systems, and the enterprise network?
• Privilege: How do you effectively determine who should have permissions of transfer and confinement?
• Intellectual Property: How do you make protected intellectual property widely available without violating the rights of the owners?

Letter from the Department of Computer Science

I enclose a white paper planning document on Computation and Information Sciences for your planning initiative. The document has been endorsed unanimously by the faculty in the Department of Computer Science and is in full synergy with our Departmental 10-year plan that we're submitting to Berndt.

We are at the dawn of an "information age" where the complex challenges of how to harness, manage, and transform information for the good of humankind will be a driving force throughout society. With coordination and collaboration, Duke can play a guiding role with its educational and research programs.

We've engaged in a series of discussions on computing and information technology with several faculty and staff across campus, as well as with the Industrial Advisory Board of our Industry Partners Program.  Their inputs have provided excellent guidance in the evolution of the document. We recommend that a broad-based University study be undertaken--with representation from all parts of campus as well as from industry and government--to address the challenges and opportunities.

Sincerely,

Jeffrey S. Vitter
Gilbert, Louis, & Edward Lehrman Professor
and Chair of the Department of Computer Science

ISDS endorses the intellectual vision laid out in the planning document on Computation and Information Sciences. The scientific perspectives presented are well-founded, well-balanced and responsive to the rapidly evolving technological landscape, and highlight a panoply of challenges and opportunities that will be central to international scientific and educational agendas and priorities for decades to come. Statistical science -- as a mathematical science, a computational science and an information science -- is cross-cutting and deeply implicated in the interdisciplinary intellectual frontiers that lie at the substantive heart of the proposal. Much of what is developed here resonates with themes explored in ISDS's current strategic plan, and ISDS is wholly supportive of the goals of broad and deep integration of research and educational programs across an intellectual consortium of information sciences.

The proposal spotlights a major arena in which Duke has the potential, and the opportunity, to move into a significant leadership position internationally, and should be explored further in campus-wide discussions.

Best Wishes,

Michael West
Arts & Sciences Professor of Statistics
and Director of ISDS

The Industry Advisory Board (IAB) recognizes the need for a flexible structure that addresses, from an academic perspective, the fast paced changes brought into all aspects of our life by the development of Information Technologies.  These changes underlined the importance and the need for interdisciplinary education and research. Modern Universities will have to create appropriate structures that will allow them to leverage interdisciplinary education in order to better prepare students for the real world, and interdisciplinary research in order to maintain a leading role amongst research institutions. The initiative started by the Computer Science Department represents a first, very important, step

The proposed Computer and Information Sciences project is extremely well received by all members of the Industry Advisory Board. The structure and the goals of the document are impressive, indicating that Duke University is committed to find the best ways to secure a leading position in todayís academic world. The members of the Board expressed their support of the project and desire to actively assist Duke University in the process. This commitment is reflected in the comments and suggestions submitted in reference to the proposal.

1) CISís management of funding is an important factor in ensuring the success of the project. It was pointed out that CIS would have to properly handle the two major funding sources: National Science Foundation and Industry.
2) CIS should provide junior professors with tenure tracks in interdisciplinary research. This option is largely unavailable at the present time.
3) The project will have to clearly identify the ways in which the synergy between researchers will be achieved and the ways in which the knowledge transfer through teaching will be encouraged. Moreover the various levels of knowledge to be transferred should be identified: general public, core curriculum and specialized knowledge.
4) CIS should create an environment where academic processes are accelerated in order to match the time scales used for industry development
5) CIS has to take an active role in promoting the latest developments in IT and this should be done through a specialized office.
6) Concerns were raised with respect to the benefits provided by the technology transfer aspects of CIStech. We caution that this activity not distance itself from the set of values promoted by CIS and actually become an obstacle in the interaction with the industry.

The Industry Advisory Board recognizes this project as a tremendous opportunity for Duke University to consolidate its position amongst the elite academic institutions worldwide.

Members of the Board:

Alan Blatecky (MCNC)
George Bourianoff (Intel)
Fred Brooks (UNC Dept. of CS)
Chi Chan (IBM)
John Dallen (Research Triangle Institute)
Jon Fjeld (Raindrop Geomagic)
Amede Hungerford (HP)
Clinton W. Kelly III (SAIC)
John Meckley (Data General)
Nikos Pitsianis (BOPS)
Ciprian Popoviciu (CISCO Systems)
John Spencer (Microsoft)
Francis Sullivan (Center for Computing Sciences)
William Wulf (University of Virginia; President, National Academy of Engineering)