CISE - Advisory Committee
DIRECTORATE FOR COMPUTER AND INFORMATION SCIENCE AND ENGINEERING ADVISORY COMMITTEE (CISE AC)
Meeting Summary December 2-3, 1998
Dr. Edward Lazowska, Chair Dept. of Computer Science & Engineering, University of Washington, Seattle
Dr. W. Richards Adrion, Computer Science, University of Massachusetts
Dr. Frances Allen, IBM Yorktown Heights, NY
Dr. Forest Baskett, Silicon Graphics, Mountainview, CA
Dr. Christine Borgman, Information Studies, University of California, Los Angeles
Dr. Frederick Brooks, University of North Carolina, Chapel Hill
Dr. Hector Garcia-Molina, Department of Computer Science, Stanford University, Stanford, CA
Dr. Louis Gomez, School of Education and Social Policy, Northwestern University, Evanston, IL
Dr. Mary Jane Irwin, Dept. of Computer Sciences & Engineering, Penn State University, University Park, PA
Dr. Leah Jamieson, School of Electrical & Computer Engineering, Purdue University, West Lafayette, IN
Dr. Sid Karin, University of California-San Diego, La Jolla, CA
Dr. David Messerschmitt, University of California—Berkeley, Berkeley, CA
Dr. Cherri Pancake, Department of Computer Science, Oregon State University, Corvallis, OR
Dr. Richard Rashid, Advanced Technical Research, Microsoft, Redmond, WA
Dr. Daniel Reed, Department of Computer Science, University of Illinois, Urbana, IL
Dr. Herbert Schorr, USC Information Sciences Institute, Marina del Rey, CA
Dr. Andries van Dam, Brown University, Providence, RI
Dr. Douglas Can Houweling, University of Michigan, Ann Arbor, MI
Dr. Jeannette Wing, Carnegie-Mellon University, Pittsburgh, PA
Dr. Paul Woodward, Astronomy Department, University of Minnesota, Minneapolis, MN
Ex Officio Member:
Dr. Paul Hale, Jr., Center for Rehabilitation Science and Biomedical Engineering, Louisiana Tech University
Dr. James Flanagan, Rutgers University, Piscataway, NJ
Dr. Richard Gitlin, Research Vice President, Bell Laboratories, Holmdel, NJ
Dr. S. Thomson Leighton, MIT Lab for Computer Science, Cambridge, MA
A meeting of the Directorate for Computer and Information Science and Engineering’s Advisory Committee (CISE AC) was held at the National Science Foundation, Arlington, Virginia on December 2-3, 1998.
Wednesday, December 2, 1998
WELCOME AND INTRODUCTIONS
Dr. Ed Lazowska, Chair, called the meeting to order at 8:30 a.m. and welcomed member Paul Hale. Dr. Ruzena Bajcsy, Assistant Director for CISE, noted the many changes happening in the Directorate, including her new appointment. She presented an overview of several staffing changes and requested support from the CISE AC in the recruitment of people for the open positions.
REPORT OF THE CISE ASSISTANT DIRECTOR
Dr. Jerome Daen, CISE Sr. Advisor for Budget, Management and Planning, provided an overview of the NSF and CISE budgets over the last three years. In FY 1998 there was litigation about the domain name escrow fund and NSF was authorized to spend $23M late in FY 1998. Approximately $39M is left in escrow and NSF hopes to have the funds available in FY 1999 for networking activities. The FY 1999 budget was reviewed.The FY 2000 budget process will begin soon and will include funding requests for priority areas identified in the President’s Information Technology Advisory Committee (PITAC) Interim Report to the President. Dr. Bajcsy thanked Dr. Daen for the fantastic job he did as interim Assistant Director for CISE.
Dr. Juris Hartmanis, former CISE Assistant Director, summarized the PITAC report planning process. Dr. Merrell Patrick, CISE Chief Science and Technology Officer, spearheaded the plan which began with guidance from Dr. Neal Lane, Office of Science Technology and Policy (and former Director of NSF). Working groups have been formed representing various federal agencies with NSF chairing several of the groups. Dr. Rita Colwell, NSF Director, participated in a meeting with other agency heads on November 23, 1998. The group is identifying important projects that require “over the envelope” funding for research areas identified in the PITAC report. Dr. Lazowska thanked Dr. Hartmanis for the work he has done as Assistant Director for CISE.
Dr. Ruzena Bajcsy, newly appointed Assistant Director, shared her vision for CISE. She emphasized the importance of strong ties to the scientific community. She would like to see the focus of computer science shift from a specific application-driven discipline to one that extracts generalities and abstractions applicable to a host of applications. She requested feedback from the CISE AC to help build the case, in response to the PITAC report, that the NSF should be the lead agency for implementing the recommendations. Both individual investigators and Centers should be supported. The problem to be addressed should determine how the research is organized. This includes support of research that may be considered risky. In general, mission-oriented agencies have shorter-range projects. There should be a marriage between NSF and other mission agencies because NSF can support long-range projects and mission agencies can support shorter-term testbeds.
Communication between the CISE AC and the CISE Directorate is important as the advisory members represent top scientists in the community and help provide guidance on priorities for research and rationale for requesting funding.
DISCUSSION OF CISE PLANNING ISSUES
Dr. Ed Lazowska shared his October 30, 1998 letter of response to Congressman Steven Howell, Committee on Science, concerning Dr. Lazowska’s October 6, 1998 testimony at a hearing on high-performance computing. He responded to questions including the feasibility of CISE and NSF in taking a lead in the implementation of the PITAC report recommendations. Dr. Lazowska has met with Dr. Neal Lane, Vice President Al Gore, ARPA, and NIST. There is no money yet for implementing the recommendations in the PITAC report. To get funding, the community needs a clear, compelling statement about their vision and they need to help NSF build the case for the core CISE disciplines.
DISCUSSIONS WITH THE NSF DIRECTOR AND ACTING DEPUTY DIRECTOR
Dr. Rita Colwell, NSF Director, and Dr. Joe Bordogna, Acting Deputy Director, met with the CISE AC. Since Dr. Colwell’s appointment to the NSF on August 4, 1998, she has been focusing on information technology issues. The FY 2000 budget is looking quite promising for information technology, but it is still in the planning process. Another important area for Dr. Coldwell is biocomplexity. She talked about the meeting she conveyed with other agency heads and noted that the findings in the PITAC report acknowledge an interagency approach is needed to meet the information technology needs of the future. It is appropriate for NSF to be the lead agency in this effort.
Basic research in computer science and engineering is important and will benefit all science and engineering disciplines. To accomplish this, super-computer access is important for researchers. NSF has requested money for basic research and made a smaller request for other research. The under pinning means that other sciences can advance through the kinds of science related to managing databases. Dr. Lazowska asked how the members of the advisory committee could help. Dr. Coldwell responded that, in her view, there has been an under investment in the computer science area and that more investment is needed. To get more funding, Congress needs to be persuaded as to the importance and benefits of this investment. As representatives of the scientific community, the AC members should be promoting their research and "spreading the message". When NSF promotes the research, it is sometimes seen as self-serving. To help NSF see first hand what is happening in the research community, some of the increased S&E funds in the FY 1999 budget will be used for travel for the Program Directors. She also proposed the idea of running panel review meetings interactively, provided NSF was wired to do so.
There is a National imperative to launch an information technology effort and NSF is fully committed.
Ms. Judith Sunley, Office of The Director, NSF, provided an update on NSF’s response to the Government Performance and Results Act (GPRA). NSF has developed a strategic plan and will be going through a mock performance assessment this year before the required assessment is due in January 2000. Ms. Sunley reviewed criteria that will be used to assess the performance of NSF programs and management. One of the major challenges for the NSF in responding to GPRA is to identify outcome goals that are qualitative and reflected the nature of scientific discovery. How do you set fundamental goals for computer science research? The CISE AC was requested to provide input on this as it relates to identifying goals for the PITAC report.
Dr. Lazowska noted that the Committee of Visitors (COV) report for the ACIR division seemed to captured the GPRA goals for output. Feedback from the COV reviews will be used in developing the CISE Performance Assessment. He noted a concern in the community is whether the Advisory Committees and COVs can be objective about NSF’s performance in this review. Other advisory committees have suggested interchanging members so that another science is addressing a different community. It was suggested that a small subcommittee of the CISE AC be formed to address GPRA and the CISE Performance Report that will be presented at the fall meeting. Issues that should be addressed include how to handle multi-agency projects and results that are seen after NSF funded the research. NSF’s new project reporting system will also help to capture some of this information.
Ms. Darleen Fisher, CISE Special Projects in Networking Research, proposed a plan for forming a CISE AC working group on GPRA. This group would provide advice on the process, approve core questions, and review a draft performance assessment report before it goes into the full committee. The Committee would be provided the report at the same time it is provided to NSF (December 15th). Ms. Fisher noted that getting information back from the PIs is important in assessing performance. The Program Managers need to help ensure that this information is received by not approving funding unless reports have been submitted. The FY 1998 performance report is a internal NSF working document and comments from the AC would be valuable in developing the FY 2000 budget plan. The FY 1998 mock report could also be used as a draft for FY 1999.
CISE RESEARCH OPPORTUNITIES
Members of the CISE AC were asked to summarize research activities they are involved in and note exciting research opportunities within the computer science community. Their comments are briefly summarized below:
- Dr. Ruzena Bajcsy was involved in visual perception research and mobility robotics which also includes development of swimming robots. Use of different optical sensors is being researched.
- Dr. Doug Van Houweling’s research is focusing on the second generation Internet and Gigaflops. Sociology issues associated with broadband networking are being assessed. Digital video applications are underway and network storage initiatives have been established with industry involvement. They are also looking at the challenges identified in the PITAC report such as middleware.
- Dr. Louis Gomez’s lab is working to bring today’s state-of-the-art technology into the urban classrooms. This involves collaboration and coordination technology through the world wide web to breakdown the barriers between schools.
- Dr. Cherri Pancake is focusing on an area called "usability engineering". Results from cognitive studies are being applied to how technology works with human users. For example, they might be studying how people make errors in interacting with software systems. They have developed middleware solutions using this approach of usability engineering.
- Dr. Hector Garcia-Molina research interests are in information management and digital libraries. Database technology and information retrieval technologies are being integrated. With the success of e-commerce businesses such as Amazon.com, there is a concern that the real science will not get recognition.
- Dr. Paul Hale runs a center for rehabilitation science and biomedical engineering that is involved with persons with disabilities. Their focus is on making devices and tools available to students.
- Dr. W. Richards Adrion’s university has been addressing issues related to interdisciplinary research. Testbeds are difficult to establish across several disciplines. Software research is an area where more research is needed including research in scale for software engineering.
- Dr. Herbert Schorr noted that the Information Sciences Institute at USC focuses on systems and the establishment of an application base. They are working on protocols to put real-time signal on the Internet and distance learning using the most advanced technology. Other activities include the invention of a MIMS processing device and work on digital government.
- Dr. Dave Messerschmitt’s group is addressing issues related to new wireless networks.
- Dr. Mary Jane Irwin’s research focuses on computer architecture. Research is looking at the design of the clock MHz and logic transistors and the next generation architecture
- Dr. Richard Rashid’s research group has developed a new way to support vector machines and is working in the data mining area. Other exciting areas is graphics learning and distance education and technology learning. One of the fundamental limitations they are faced with is the limited qualified people they can hire.
- Dr. Sid Karin works at a Center involved in large collections of data and problems associated with this. They are working on areas of networking. Specifically, they are looking at developing technology applications for people that are not computer scientists. Dr. Fran Allen noted research is needed in the area of compilers and tools. The basic assumptions about these areas need to be rethought.
- Dr. Jeannette Wing shared three successes: JAVA, the role of symbolic model checking in hardware verification and electronic commerce. She proposed that in the future software engineering will need to utilize open source models.
- Dr. Christine Borgman is working with a group of individuals from communications, computer science and library science. Digital preservation is a concern. While it is providing access to content that was not otherwise available, we don’t know if this content will still be available 5-10 years from now. Partnered with a strong school, they are working on other issues like how to select, collect and organize materials for distance learning and children and information retrieval. They are working with the film industry in archives and with NSF in primary archival materials in a textbook. The definition of a document is changing with electronic documents and archiving means maintaining links to that document as well.
- Dr. Forrest Baskett’s research is in four areas: 3-D graphic vision, parallel computing, real-time image processing and robust computation for non-robust systems.
- Dr. Leah Jamieson mentioned several accomplishments in the signal-processing field to include CD technology, Solomon codes, digital images and video, and medical imaging. Other accomplishments include wireless/mobile technology and work in multi-media and smart antennas.
- Dr. Fred Brooks is doing research in computer graphics. An interdisciplinary team is building a virtual reality interface for scanning tunnel microscope to increase the capability of imaging using the tips of the microscopes. The use of virtual reality technology has created stresses for computation and display technology.
- Dr. Paul Woodward is working with an interdisciplinary team on integrated laboratory fluid dynamics for high-end computing, system software generation. There is a need to develop software approaches for high end computing as well as others.
- Dr. Daniel Reed is working on issues related to high-end computing and performance engineering design. Networking issues include dynamic adaptation, adaptive control, just-in-time computation and storage. There is a need for high-end architecture research.
- Dr. Andries van Dam is focusing on graphics and human-computer interactions (HCI). There are new form factors, such as palm computers and wearable computers. How does this change the human interface? Post-Windows user interfaces that are point and click are being explored. Interactive illustrations, non-photo realistic rendering and office virtual reality are other research areas they are engaged in.
Dr. Bajcsy thanked everyone for their input and noted that this was a wonderful opportunity to learn about what the community is doing. She hoped to use the material and recommendations from the PITAC report to make plans. She added that the NSF has a Major Research and Instrumentation (MRI) fund of approximately $25M over a ten-year period. She encouraged the community to "think big" and identify the "grand challenges" for the next century as more interdisciplinary projects will be needed.
NSF INITIATIVES: LIFE AND EARTH'S ENVIRONMENT (LEE) AND EDUCATION FOR THE FUTURE (EFF)
Life and Earth’s Environment (LEE)
Maria Zemankova, IIS, CISE, provided an overview of the NSF initiative on Life and Earth’s Environment (LEE) LEE expands on existing research at NSF with a goal to identify major interdisciplinary scientific challenges related to the environment. The CISE AC was asked to provide input to Dr. Colwell about this initiative. Dr. Sid Karin offered to draft a response on behalf of the AC to be circulated by email. Dr. Zemankova reviewed existing research areas for LEE with include Global Change, Environmental Observations, Life in Extreme Environments, Urban communities, Engineered Systems in natural Environments, Sustainable Ecosystems, and Integrated Research Challenges. The CISE directorate funded approximately $1.5M in FY 1998 for LEE activities. NSF has set aside a total of $507.4M for LEE research.Members were encouraged to try to identify opportunities in this area. The chair of the LEE working group is Michael Cavanaugh (MPS Directorate) and input from the AC’s was welcome.
Educating for the Future (EFF)
Dr. Mike Foster, EIA, CISE provided an overview of the issues related to educating for the future in response to Dr. Colwell’s request for input on the SMET Workforce. The initiative focuses on undergraduate education, integration of research and education, diversity and global orientation and research on learning and education. Current CISE Activities include Educational Innovations (EI), Combined Research-Curriculum Developed (CRCD), Minority Institutional Infrastructure (MII), Collaborative Research on Learning Technologies (CRLT), Research Experiences for Undergraduates (REU), PACI Educational Outreach, and other special projects.Dr. Foster gave specific examples of projects in each area. CISE can play a future role in this area through relevant research and technology transfer such as digital libraries, networking, ubiquitous computing, and virtual environments which can be applied to distance, self-directed and lifelong learning. They can also establish partnerships within and beyond NSF using the Education Research Initiative with EHR and through the Teaching Experiences for CISE Students (TECS).
John Cherniavsky, EHR, expanded on the Educational Research Initiative. ERI was established in response to the March 1997 President’s report on the Use of Technology to Strengthen K-12 Education in the United States. Current plans are for a $22-$36M initiative in FY 1999. CISE is a major player because the report is referring to information technology when it talks about technology. There is a public web site that talks about ERI activities such as K-3 reading and math and K-12 teacher education. Dr. Borgman and Dr. Gomez volunteered to help Dr. Foster draft a response to Dr. Colwell on the EFF initiative.
The discussion that followed focused on the challenges CISE should be pursuing and how to reorganize to get more money for Computer Science and Engineering community research in relation to the PITAC report. More ambitious projects are needed.
Thursday, December 3, 1998
The CISE AC reconvened at 8:30 a.m. and broke into working groups.
BREAKOUT OF CISE AC WORKING GROUPS
The CISE AC met in Working Groups for the following:
- Advanced Computational Infrastructure and Research (ACIR)
- Advanced Networking Infrastructure and Research (ANIR)
- Computer-Communications Research (C-CR)
- Experimental and Integrative Activities (EIA)
- Information and Intelligent Systems (IIS)
Following the working group session, the entire CISE AC reconvened.
PRESIDENT'S INFORMATION TECHNOLOGY ADIVSORY COMMITTEE (PITAC) INTERIM REPORT DISCUSSION
Dr. Lazowska noted a change in the agenda. Each of the five divisions in CISE, as it related to the research areas in the PITAC, would be reviewed to identify grand challenges for CISE.
Michael Evangelist: Software Workshop, Wireless Workshop
Dr. Michael Evangelists, Computer-Communications Research (C-CR), provided an overview of activities within CISE C-CR.CISE C-CR has requested wireless and software initiatives for the FY 1999 budget in a joint effort with other CISE divisions and ENG. The FY 2000 budget planning is in progress with other agencies. Preliminary and in-depth discussions have taken place with EIA about other research areas in addition to software, such as an initiative in hardware-software co-design.
Dr. Evangelist provided a summary of the software workshop held in mid-October 1998 with widespread representation from industry and universities. The workshop was a forum to gather input from the research community. Results from the workshop were consistent with the PITAC report:
- Extract principles of construction from past successes and failures
- Test those and theoretical principles on real projects
- Continue developing rigorous formal basis – emphasizing scalability
- Build important advanced applications – further testbed for the new ideas and a solution to PITAC problem of desperately needed software not being developed. (want to be testbeds, not ad hoc solutions)
Conclusions from the workshop were that the computer science industry has invested in expanding the economy, but has neglected expanding the scientific underpinnings. More research is needed on basic empirical investigations to balance other research on applications and theory.
Dr. Evangelist introduced Dr. Jamieson who provided a summary of the workshop on wireless communication (Tetherless T3) which was held November 18-20, 1998 in Denver, Colorado. Dr. Jamieson provided an overview on what tetherless T3 includes (wireless, mobile, cellular/terrestrial) and explained that T3 is different from other forms of wireless communication because it has increased bandwidth. T3 can be applied to scalable information infrastructures, universal connectivity for education and telemedicine. Basic research is needed at individual layers (adaptive everything) and the next paradigm of interactions need to be researched across layers such as channel, network, signal processing, applications, data, and architecture. T3 will not be achieved using conventional designs but it will lay the groundwork for scalable high-data rate mobile communications.
Dr. Evangelist mentioned there is a joint program announcement between CISE and Engineering (ENG) in the wireless network area.Wireless communication is critical – it embraces the mobile network and is important in several areas. Dr. Pancake added that IIS is also involved. The focus is on queries and data mining and how you get information across this network. Dr. Jamieson added that this would be included in a report from the workshop. Dr. Evangelist noted that C-CR is interested in hearing ideas related to this area.
Dr. Lazowska requested the AC to identify grand challenges in this area:
- Computer imbedded sensors
- Low-power requirements
- Need for models to help look at systems as a whole from a software point of view
- The issue with wireless with respect to applications is twofold for any application that uses the network. First, does the network adapt to what the application needs or does the application adapt to the network? We have to think more about the application adapting to what the network can provide and this varies with time. Second, latency is still an issue even though bandwidth is becoming less of an issue. Wireless will be bit-rate limited, rather than latency limited.
- Need design methodology and modeling for the management and control of these systems.
- When suggested that there is a need for an overarching theme for a visionary program for software, it was cautioned to not limit it to an identified application or theme. Software is the enabler underlying everything.
- The environment in which our software runs is changing rapidly.In the future nimble adaptive software is needed.
We want to increase dramatically the success rate of large software projects and dramatically improve productivity of software development with a focus on software embedded in large systems (e.g. banking, and transportation).
The CISE AC further discussed the "grand challenges" that face the computer science community. The CISE AC felt that the PITAC report failed to mention artificial intelligence. It was agreed that is should be in the NSF plan. Dr. Gomez added that the PITAC report is an interim one and the final version due out in February 1999 should have some issues filled in.
Michael Lesk: HCI and Information Infrastructure
Dr. Michael Lesk, Division Director, IIS, noted that the human-computer interactions (HCI) and information infrastructure research areas fell under the subcategory of software in the PITAC report. The major challenges faced in the Division of Information and Intelligent Systems are:
1) Dealing with terabyte databases and information overload in areas such as:
- information management and organization
- search and retrieval
- analysis and transformation
- visualization and presentation
- translation and interpretation
- speech, language
2) Human-computer interaction, which involves a wide range of issues related to graphics/sound, interface fundamentals, perceptions, "interfaces", and evaluations.
3) Smart spaces that use independent devices, coordination and systems work and sensors.
Many PITAC themes inter-relate:
- visualization extends across software and high-end computing
- information management covers both software and scalability
- devices and spaces involve networks and software
Dr. Lesk noted that is it hard to isolate particular themes, and that it is not clear we would benefit from excessively narrow programs.
The CISE AC was asked to identify grand challenges in the area of HIC and information infrastructure:
Although bandwidth will be plentiful, when it is untethered it won’t be. Processing will be high-speed, but it will be limited by the speed of light.
In the future function is going to follow form– bandwidth is an example, as the C-factors will change in the future. The desktop computing that has been done for the past 20 years will change as will the time factor for retrieving and processing information.
What needs to be highlighted, given the data, is "What do I do with it?" The future is hands off Data will be input through vision, speech, language, and touch. To process information in this way, more powerful machines with longer battery life will be needed.
In many computer sciences programs people do not learn about humans or cognitive perceptions yet students are creating applications that humans need to interact with. A partial solution is for people in other professions to learn more about computing too. The multi-disciplinary partnerships will help. In the future, we cannot assume that computer scientists will build computer systems. Tools need to be developed that people in other domains can use to build systems using their expertise.
George Strawn: Scalable Information Infrastructure
Dr. George Strawn, Division Director, ANIR, began by saying that the science supported by the Advanced Networking Infrastructure and Research (ANIR) Division is entering a period of extraordinary opportunity at the network level. Fundamental assumptions are being reconsidered. They are moving from a reality of a network of thousands of nodes to millions or trillions of nodes. Data networking will have an opportunity to be on top of the telephone industry.
Dr. Strawn summarized the findings in the PITAC Interim Report:
- Conclusions: Enhance federal investment in IT R&D; focus on longer-term problems.
- Research Priorities: Software, scalable information infrastructure (SII), high-end computing, socio-economic and workforce implications
- Modes: diversify, including Vital Centers
Management: lead agency, advisory committee reviews
Dr. Strawn shared the interagency writing on the issues related to scalable and extensible information infrastructures (SII/EII):
1) Deeply networked systems need to be developed that have embedded devices to go from millions to billions of nodes, provide a greater degree of autonomous operations and have a flexible/programmable substrate.
2)Ubiquitous connectivity is needed for wireless networking, increased heterogeneous access and mobile IPs.
3)Scalability of core capacity and services through simulation and analysis tools (to simulate a million or billion node network), scalable middleware, and optical packet routing.
Dr. Strawn reviewed issues addressed at the NSF SII workshop. Research themes included:
Stay ahead of the growth curves
Distinguish between networking research testbeds and high performance networks for science research
Develop new ideas and new theory to cope with the new reality
Scaling in size, speed, and diversity
Research goals were identified which were to maintain the NSF tradition of encouraging researchers to bring NSF their visions. Research areas where fundamental research is required include: basic theory, understanding global behavior, design methods and tools, physics of networking, system design, software mechanism, models of (in) security and authentication. Modes of research include a desire to increase support for multi-investigator and multi-institution proposals and for theoreticians and practitioners. Educational issues include shortage of faculty and retention of students.
A letter was sent to PITAC in response to the draft report that noted that CISE would help with the networking section and to encourage the continued coordination between PITAC and the networking research community. Comments to agencies were also made.
Discussion followed. It was suggested to make sure that the management of networks was included in talking about networks. This will be even more important as we move into networks that provide transactions. With a stronger "Internet" network, the question was raised if industry would still rely on intranetworks. There was some disagreement as to whether or not industry would no longer use Intranets if the Internet was more secure and robust. Some members felt that pirvate industry would not want to separate their network form the rest of the world while others thought they would not be willing to give up the “control” they have over their own Intranets. The grand challenge is to work on issues of authentication, security, etc. for the public network. Several noted that firewalls are not successful –something better is needed that provides a level of reliability and security that cannot be achieved by current Internet technologies. Well-defined goals and benchmarks are necessary.
Robert Borchers: High End Computing
Dr. Robert Borchers, Division Director, ACIR, reviewed the time-line of activities of the subcommittee in high-end computing within the Advanced Computational Infrastructure and Research Division (ACIR). Three workshops were held in Science Applications, Algorithms and Software. There was a joint workshop with a PITAC interim report sent to the President.
The Science workshop helped to identify high payoff applications requiring alternative increases in computing power. Scientists and engineers made presentations from a very broad range of subject areas. Important applications need large-scale computing, large memories and high bandwidth networking and input/outputs. The Algorithms workshop findings concluded that systems were "underachievers" and can be aided by algorithms. New metrics and support will be needed in evaluating them with teams representing applications and computer science. Future simulations will favor multi-discipline applications and more physical reality per discipline, rather than higher resolution. The Software Workshop findings concluded that the HPCC potential of large parallel systems can only be partially fulfilled due to a number of problems and limitations such as:
limited programmability and effective scalability,
limited understanding of end to end behavior,
poorly understood application and system models,
inadequate data management, and
mismatched human and computing capabilities.
Software research is necessary in all of these areas. Dr. Borchers reviewed the current situation within NSF and the next steps.Details were provided on outcomes from each of the workshops.
The CISE AC commented that support for high-end computing is also in support of other science disciplines that need the ability to do high-end computing. There was some concern that this need might drain some of the computer science resources while aiding other applications. It was also noted that there is not sufficient human resources to address short-term application need and focus on long-term application needs. A similar problem exists in software. It was mentioned that corporate researchers are bringing processes into the realm of using computational power. It these corporations are also talking to Congress and providing financial support, the community might have a tremendous advantage in getting additional funding for this research area. High-end computing should be viewed as an opportunity for CISE and an opportunity for new funding. Cooperation with other scientists is very important in getting support.
Michael Foster: Socioeconomic and Information Technology Workforce Issues
Michael Foster, Acting Division Director, Experimental and Integrative Activities (EIA), presented a summary of the PITAC recommendations related to socioeconomic and workforce issues. The three areas where the recommendations best fit NSF's mission of education were to:
- Ensure an adequate supply of highly trained scientists and engineers.
- Develop new educational programs to retain information technology workers with outdated skills.
- Encourage increased participation by women and minorities.
To support these recommendations, Dr. Foster provided suggested approaches that could be taken by NSF:
- Direct a portion of IT funding to SEW research and IT workforce development.
- Invest in SEW research for both centers and individual investigators.
- Invest in traineeship programs and studies directed toward increasing participation.
The CISE AC noted the importance of teaching Computer Science courses in other discipline programs. They also wanted to know if there was a clearinghouse of any sort that bought together the efforts of the programs in the K-12 area. The group talked about other programs, like PACI, that require proposers to include an education and outreach program in their proposal. With little funding in information technology education, the vast community needs to be leveraged to reach the most people. Several success stories of programs that involved the community in information technology education were shared. The CISE AC discussed the fact that NSF used to offer more summer camps/programs that were very successful. Many programs used to require a dollar to dollar match – this is no longer as feasible for many universities.
In addition to education of the general public, there is also a pro-social layer in which computer scientists should be brining back links to other science programs. There are also policy issues related to privacy issues, trackable systems, etc. that require very different models. Emphasis is needed on the social impacts of computer science design and computer science impacts on society. There is a problem in attracting domestic graduate students. Another concern is that technology centers, such as MOSES, which were great, are offering technology that is a year behind, rather than a year ahead. There needs to be other centers like CADRE that help focus on that.MOSES really accelerated research.
ADVISORY INPUT ON PITAC GRAND CHALLENGES
Steve Mahaney requested help in summarizing the grand challenges that came out in the discussion:
- Software that now suits an environment that is adaptive and changing and more reliable.
- A radical change in the form factor will make as big a difference as the augmentation of batch computing with desktop PCs.: Function follows Form. MOSES excited the research community and improved the workforce, etc. What are the Moses' of the future?
- Suggest providing specific set of middleware that people can use for extraction and building applications.
- Global memory—having one place where you can go to get answers.
- Electronic Commerce
- Dealing with file systems with billions of files
- One of the grand challenges for software is reusable software to demonstrate that it can be done – development of the environment becomes part of the research activities. Looking more at IP and IP reuse. Industry is moving that way, but universities can’t
- afford the IPs – they are too expensive. We need to assure that the academic community has access to industrial-scale processes and resources (hardware and software). There needs some way for academia to be there.
- “Sustainable Global Information Infrastructure”. The challenge is to integrate computer science knowledge with social, behavioral and economic knowledge to bring together information that is usable and sustainable.
- Multi-modal human-computer interfaces.
- How do we shift from individual proposals to grand-scale large activities? Is there any process assistance for NSF? Perhaps there is a way to look at the quality of proposals before a final decision about what is allocated is made. If you are looking at two areas you could see what proposals are of higher quality. Hope funding would be in a program that is not too highly defined so there is some flexibility.
- We need more effort about separating out facilities and infrastructure from research – treat them differently. Also problem with KDI was it was an announcement. Why can’t some of the funding be set aside for unsolicited proposals for bright ideas?
- Peer review has a lot of problems associated with it with one of the most difficult aspects being that it discourages truly inter-disciplinary work. There needs to be more interdisciplinary work and work on the review process to make sure that the review is adequate.
- I am trying to think how one can connect industry to some of needs of researchers. Industry has a need for the research and involvement. Feel this interface isn’t working well. Don’t know how to go about doing it. Would like to broaden this and involve researchers long-term on projects. Certainly IP issues which are always a problem – but it is not built into the structure. If there were more programs with matching required from industry – forces researchers to talk to industry and industry to have vested interest. One problem- when industry commits, funds are there right away, with NSF, still stuck for nine months without funding.
- Allen: On the right programs, there is not a problem with funding. There is the PI and university model. There is another model where universities set up a center where they involve a lot of companies to provide support for research and in return, the companies get the benefit of the result and help set the agenda. There isn’t another way of doing it that might involve a number of universities and number of universities.
- Borgman: I would add a caveat about working with industry bottom up. Industry partnerships are great for certain kinds of programs and research – but may cut out whole areas of research.
- Karin: You may find someone interested in technology that you have knowledge about, you get funding and then when they learn about it, they cut off funding and do it in-house. In order to get a level funding field, you have to have staff that keep looking for funding.
- Messerschmitt: Need long-term research plan – where corporation will fund things that will not be seen in the near future. (There was some disagreement – industry has too short of a focus.)
- Messerschmitt: There are good examples – it can happen.
- Mahaney: Another possibility is the DARPA model where they build research areas through workshops, etc. Are there mechanisms you like to see?
- Speaker: Workshops are not enough. DARPA can fund industry and researchers, NSF cannot.
- Mahaney: Think Alan is looking more for an intellectual exchange.
- the notion of software layers of advancing the knowledge of process
- assessing past successes and failures
- ubiquitous connectivity and computing
Dr. Lazowska noted the next CISE AC meetings will be Thursday and Friday, April 29 and 30, 1999 and Wednesday and Thursday October 13 and 14, 1999. Dr. Bajcsy thanked the Committee members rotating off for their participation. She also thanked Merrell Patrick and Mel Climent who are leaving the CISE front office as of January and noted that John Raymond is also retiring.
Dr. Lazowska moved for the formal approval of the minutes from the last CISE AC meeting. A motion was made by Dr. Irwin and seconded, and the minutes were unanimously approved. Dr. Lazowska encouraged the CISE AC members to communicate with the Division directors and Dr. Ruzena. He asked for someone on the CISE AC to volunteer for the AI subcommittee on databases, electronic libraries and networking, adding it was an opportunity to provide input to the NSF to shape programs and the field. With no further agenda items, the CISE AC meeting was adjourned at 2:05 p.m.