July 14, 1999
Good afternoon, Mr. Chairman, Ms. Johnson, and Members of the Subcommittee. Thank you very much for inviting me to testify today at this hearing on the Networking and Information Technology Research and Development (NITR&D) Act, HR 2086. I am Edward Lazowska, Chair of the Department of Computer Science and Engineering at the University of Washington. I am here representing the Computing Research Association (CRA), of which I am the Chair. CRA is an alliance of more than 180 academic, industrial, independent, and membership organizations working to strengthen research and advanced education in computing and allied fields.
I also chair the NSF Computer and Information Science and Engineering (CISE) Advisory Committee, serve on the National Research Council's Computer Science and Telecommunications Board, and serve on the Technical Advisory Board for Microsoft Research. So I believe I can provide you with a broad and balanced perspective on this legislation.
Let me start by offering CRA's enthusiastic support for the NITR&D Act and expressing our gratitude to Mr. Sensenbrenner for introducing the bill; to you, Chairman Smith, and you, Ms. Johnson, for your efforts in putting the bill forward and convening this hearing; and to the many other members of the Science Committee who have helped shape information technology research policy in recent years.
We believe HR 2086 exemplifies a sound approach to making research policy: it responds to clear national needs with recognizable objectives and sets forth a well defined program for meeting them. The computing research community is eager to take on these expanded responsibilities and is devoting considerable effort to careful planning. While my prepared remarks do not address the expanded scientific agenda that the bill would enable us to undertake, I have attached a white paper that describes some of the challenges in software research, my own field, that we could tackle more vigorously with the increased investment — challenges, in fact, that must be met if we are to realize the full potential of information technology to transform our economy and our society.
In my remarks today I'd like to make three comments about the legislation and three points about the current environment:
Of course you already know the background of this legislation as well as of the Administration's Information Technology for the Twenty-First Century initiative (IT2), which we believe is quite consistent with HR 2086. The President's Information Technology Advisory Committee (PITAC) was chartered by Congress in 1991 in legislation that originated in your Committee, and was directed in subsequent legislation to "assess the extent to which Federal support of fundamental research in computing is sufficient to maintain the Nation's critical leadership in this field...." In essence, this is what PITAC found:
We have an essential national interest in ensuring a continued flow of good new ideas in IT. After careful review of the Federal programs, however, this Committee has concluded that Federal support for research in information technology is dangerously inadequate.
The PITAC report, Information Technology Research: Investing in Our Future, demonstrated a clear need for expanded investment in long-term, broad-based computing and communications research that generates new capabilities and innovations — the fuel of the information technology industry and of the information technology revolution. PITAC identified four areas that need priority attention, and we are exceedingly pleased that the priorities outlined in HR 2086 closely track those identified by PITAC.
The bill's establishment of a new NITR&D program would expand support for "long-term basic research on networking and information technology, with priority given to research that helps address issues related to high end computing and software and network stability, fragility, reliability, security (including privacy), and scalability." This is just the right specification, and the bill is a sound mechanism for redressing the underfunding of fundamental computing and communications research that the PITAC emphasized as cause for alarm.
The bill encompasses this program within the broader context of the High Performance Computing and Communications Program (HPCC) and the Next Generation Internet initiative (NGI). We are particularly pleased that it would provide a five-year authorization for HPCC and a two-year authorization for completion of NGI. These multiagency programs have been tremendously successful.
Federal support has been and will continue to be crucial to progress in information technology. Many if not most IT technologies and the strength of U.S. IT industries stem from an extraordinarily fruitful long-term partnership among government, universities, and industry in research and development that has evolved over the last 50 years. The success as well as the virtual explosion in the number of IT companies that you hear about in the financial news every night are a direct result of investments in fundamental research made by federal agencies over the past decades.
This point is addressed in the Congressional Budget Office report commissioned by the committee, Current Investments in Innovation in the Information Technology Sector, which also noted the growing investments in R&D made by the private sector. CRA is pleased that the NITR&D legislation includes a provision to permanently extend the R&D Tax Credit to encourage further growth. The R&D Tax Credit is complementary to an increased federal investment in long-term university-based research. Industrial R&D, entirely appropriately, focuses on developing innovative products from the fruits of long-term research; industry cannot be expected to assume a significant proportion of the long-term research role.
At the same time, the bill strikes the right balance between the complementary federal and industrial roles in IT research. It appropriately limits federal support to pre-competitive long-term research the kind that generates new knowledge and capabilities, the bank of ideas from which the private sector draws. Such long-term research is a public good with diffuse benefits, and the IT industry cannot be expected to conduct it at sufficient level. Given the intense pace of the IT marketplace, firms must devote the bulk of their R&D resources to shorter-term applied research and product development. Nearly all available human and capital resources must be focused on bringing the next product to market if a firm is to be successful in the current environment. The industrial representatives on PITAC went to considerable lengths to make this clear:
The information technology industry expends the bulk of its resources, both financial and human, on rapidly bringing products to market. Nearly every available person and dollar in this industry is focused on bringing the next version or the next product to market. Delivery product cycles are as short as every three to six months. The company that fails here misses the next short-term cycle and will not be successful.... For information technology companies, the fraction of the total budget devoted to R&D is roughly twice the U.S. industry average. Over 90 percent of information technology R&D expenditures are allocated to product development, with the major portion of the remaining expenditures going toward near term, applied research.
The legislation concentrates support for the new NITR&D program and for terascale computing activities within the National Science Foundation. As the NSF is the only agency that focuses almost exclusively on fundamental, university-based research and instrumentation, with projects selected via merit review, we believe this is a wise policy choice.
We are particularly pleased to see the new terascale program concentrated at NSF and associated with the existing PACI program. The PACI centers are one of NSF's best success stories. They address a wide range of scientific and engineering problems through an open allocation process; they build the bridges between computing researchers and researchers in other fields that are critical to solving today's problems and to laying the foundation for solving tomorrow's problems; and they make full and effective use of university-industry partnerships. Moreover, we commend the NITR&D Act for requiring the terascale funds to be awarded through an open competition to ensure the best possible use of scarce federal funds.
Of the agencies under the Science Committee's jurisdiction, the NSF is also in the best position to properly and thoroughly support the long-term, fundamental research called for in the NITR&D program, research that will lead to tomorrow's computing technologies and prepare us to tackle tomorrow's problems - ones that we can't even imagine today. Other agencies are certainly in a position to contribute to NITR&D, and we hope that the Department of Energy and NASA, for instance, will take advantage of the discretionary authority provided in the bill to do so, with appropriate interagency coordination.
The NSF has an impressive track record in supporting high-risk computing research that leads to high-payoff returns. NSF's far-sighted investment strategies have enabled researchers to undertake high-potential explorations before there was a market or a demand and invent the forerunners of many of the technologies we take for granted today. The many innovations that underlie the burgeoning electronic commerce industry, ranging from the Internet to web browsers to encryption, are wonderful examples. This proven formula of long-term, high-risk research underlies today's microprocessors and operating systems, word-processing programs and web browsers, databases and search engines, computer graphics, high-speed networking, and high performance computers.
With the new funds provided through the NITR&D Act, the NSF could further apply the formula toward technological revolutions in software development and testing; human-computer interfaces; computers that communicate in human languages; the distilling of knowledge and insight from complex and massive data; scalable networked systems for anytime, anywhere connectivity; and improved performance and efficiency of high end computers. Progress in these areas requires sustained and vigorous attacks on challenging problems, as the attached document on the challenges in software alone indicates. The bill's provisions for a variety of grant sizes and centers are especially important to the success of this enterprise.
In the course of developing the NSF's FY 2000 budget, which includes as you know $146 million for participation in the Administration's IT2 initiative, the NSF is undertaking a thorough planning process to guide the proposed expansion of support for long-term IT research and the development of new computational resources. The NSF Director, Dr. Rita Colwell, is overseeing this herself, and her Assistant Director for Computer and Information Science and Engineering (CISE), Dr. Ruzena Bajcsy, has worked extensively with NSF senior management, including all of the NSF's assistant directors, to lay the groundwork for appropriate collaboration.
Many people from throughout the NSF have been involved in creating a management plan and proposal solicitation for the program, and this important cross-disciplinary interaction will continue through an NSF-wide advisory committee and task forces. The proposal solicitation will include material specifically aimed at encouraging submissions from principal investigators in non-CISE disciplines who are doing IT-related work. Partnerships will be a key element in the success of the IT research program.
The collaborative nature of the expansion of the IT research program is "business as usual" for the CISE directorate, which already takes responsibility for several programs intended to accelerate progress across the fields of science and engineering. CISE's Advanced Computational Infrastructure program, which at $71 million represents more than 20 percent of the CISE budget, primarily provides resources to investigators from other fields, especially the physical and biological sciences. Similarly, the $44-million Advanced Networking Infrastructure program, an additional 15 percent of the CISE budget, provides networking infrastructure for the broad national and international scientific community. CISE's outreach is even more broad than these infrastructure programs, though: in FY 1998, nearly 25 percent of awards made by CISE were to principal investigators in academic departments that are not usually associated with the field (compared to, for example, roughly 15 percent of awards made by the Engineering and Biosciences Directorates, and roughly 7 percent of Mathematical and Physical Sciences Directorate awards).
The terascale computing initiative will be another significant investment that directly and primarily benefits the broad scientific community. But it is important to recognize that information technologies encompass far more than networks and high performance systems, and computer science and engineering have much more to offer the broader research community, and, indeed, the nation. Examples include datamining, the web, sophisticated algorithms for computational astrophysics, telecollaboration tools, and the deep intellectual partnerships that characterize the confluence of biology and computation. Computer scientists and engineers are proud to be successful tool-builders for the other fields of science and engineering, and we need to continue to apply the fruits of yesterday's computing research to solve problems in other fields of science and engineering. But we also need to generate the breakthroughs that come from investing in fundamental, long-term research at the core of computing, breakthroughs that will create tomorrow's tools to tackle tomorrow's problems and power the research enterprise forward in ways we cannot envision today.
Just as are other areas of science and engineering, computer science and engineering should be regarded as fields of discovery — there are questions to be pursued and avenues to be explored for which the end results and applications are unforeseen. The NSF needs to speculate on some of these questions and venture down some of these avenues, for the sake of the future; the potential payoff is enormous — and not just for science and engineering but for our society as well.
It is also important to recognize that information technology is far more than an enabling technology for science and engineering; the true power of information technology is as a human enabler. Almost overnight, information technologies have become ubiquitous, transforming all aspects of our lives: commerce, education, employment, health care, manufacturing, government, national security, communications, and entertainment, as well as science and engineering.
Across the board, tomorrow's "killer applications" will be ones that we cannot even envision today, made possible by extraordinary advances at a fundamental level of the information technology field. A sound federal research investment strategy must look beyond extrapolations of today's technologies and beyond the needs that we currently envision; only in this way will the full potential of information technology for revolutionary and transformational change be realized. HR 2086 reflects such a strategy by balancing support between meeting the needs of today, and investing in fundamental information technology research.
Strengthening our investment in computing research has attracted widespread support from the research and higher education community. The Council of Scientific Society Presidents (CSSP), which represents more than 60 societies in the full spectrum of scientific disciplines, is on record in support of the NITR&D bill, and I've attached its recent letter to the Chairman of the House Appropriations Committee. CSSP attests to the profound impact that advances in computing and communications have on the nature and conduct of research:
The combination of basic computing research and applications in scientific and engineering problems, incubated in the environment of our system of higher education and university research, has proven to be extraordinarily productive.... These innovations are particularly crucial to advancing the scientific research enterprise. New computational tools and methods, and innovations in scientific and technical communications radically increase the capability and productivity of all fields of scientific research.
Similarly, the Association of American Universities and the National Association of State Universities and Land-Grant Colleges recently wrote to Congressional appropriators "...to support increased funding for information technology (IT) research," noting that "...increasing investments in IT will accelerate the pace of discovery in all science and engineering disciplines..."
In a joint statement, the American Association of Community Colleges, the Association of College and University Telecommunications Administrators, the Association of Research Libraries, the Coalition for Networked Information, EDUCAUSE, the Institute of Electrical and Electronics Engineers — USA, the Southeastern Universities Research Association, and the University Corporation for Advanced Internet Development urged bipartisan support and funding for the programs embodied in NITR&D, stressing their importance to the future stability and robustness of the network.
Federal investments in computing, information, and communications R&D over the past several decades have yielded spectacular returns, leveraging the development of new products and services, even whole new industries, for the benefit of Americans as individuals, workers, citizens, and consumers. These investments have given rise to an impressive list of billion-dollar industries: the Internet, for instance, and high performance computers, multiprocessors, database systems, local area networks, and graphic displays.
The payoff has been huge. It is well documented in a 1995 report of the National Research Council, Evolving the High Performance Computing and Communications Initiative to Support the Nation's Information Infrastructure, and in CRA's 1997 publication, Computing Research: A National Investment for Leadership in the 21st Century. With the increased investment provided through the NITR&D Act, the nation can expect to continue to reap the benefits of IT research well into the next century.
Mr. Chairman, let me thank you again for moving this bill and for your time and attention today. I would be pleased to answer any questions you might have.
Attachments:
1. Software Challenges, 7-11-99
2. CSSP letter to Chairman Young, 6-21-99
Copyright © 2004 Computing Research Association. All Rights Reserved. Questions? E-mail: webmaster@cra.org.