Computing and the National Science Foundation, 1950-2016. William Aspray

Чтение книги онлайн.

СКАЧАТЬ Directorate.

      I returned to NSF in January 2000 as Division Director for Experimental and Integrative Activities (EIA). Chapter 8 includes a description of the President’s Information Technology Advisory Committee (PITAC) report that led to the government wide initiative on Information Technology for the 21st Century (IT²), the designation of NSF as lead agency, and the planning and experiences that led to the NSF implementation of IT², the Information Technology Research (ITR) program.

      In addition, I served on a number of advisory committees and was involved in three more reorganization efforts: chairing the NSF/CISE Committee on CISE Organization in 1995–1997 for Paul Young, chairing the divisional NSF/CISE/EIA Reorganization Working Group in 1997–1998 for Juris Hartmanis, and—as a part-time CISE senior advisor—chairing a committee that advised Peter Freeman on his 2003 reorganization.

2.2Making the Case for NSF’s Computing Research Programs

      NSF provided funding for computing, communications and information infrastructure, applications, and fundamental research from its beginning. The physical scientists who ran the NSF were not quite sure there was a “discipline” of computer science, but they clearly appreciated the growing importance of computing, communications and information infrastructure, and applications. Scientific and engineering disciplines typically turned to related professional societies or the National Academies to describe the field, its accomplishments, and its future promise. An influential report was needed to define computer science, its value to the nation, and the need for investment and support.

      The professional societies—ACM, IEEE-CS, AFIPS, SIAM, and AAAI—established the conferences and journals in this new field. None of them adequately represented academic computer science research in Washington, DC. This gap led to the creation of the Computer Science Board in 1972, later renamed the Computing Research Association (CRA), which created a Washington presence in 1988. Ever since, CRA has played an important role in advocating for the computing research community.

      From 1978 to 1986, the National Academy Board on Telecommunications and Computer Applications primarily published reviews of information technology issues and challenges experienced by federal agencies such as the Social Security Administration, the Internal Revenue Service, NASA, and the Departments of Defense and Commerce. One exception was a 1982 report from an ad hoc committee on the roles of industry and the university in computer research and development.³ The National Research Council created the Computer Science and Telecommunications Board (CSTB) in 1986 to replace the Board on Telecommunications and Computer Applications.

      Earlier in the 1960s, a number of individuals attempted to define computer science as a discipline. In addition to Louis Fein’s⁴ efforts described in Chapter 1, Saul Gorn of the University of Pennsylvania wrote in 1963 that “a new basic discipline is emerging which might be called ‘The Computer and Information Sciences’ [that] makes application of concepts from the traditional fields of mathematics, philosophy, linguistics, psychology, engineering, management science, library science, etc.”⁵ George Forsythe,⁶ the founder of Stanford’s computer science department and ACM President, commented on Gorn’s analysis, suggesting that computer scientists are concerned with the pragmatics of the applications of mathematics. In 1967 Allen Newell, Alan Perlis, and Herbert Simon⁷ defined computer science as the study of phenomena related to computers. Donald Knuth’s definition⁸ of computer science as the study of algorithms appeared in 1968. Curriculum 68⁹ defined computer science as the study of information structures. Edsger Dijkstra defined computer science as the study and management of complexity.¹⁰ Historian Janet Abbate observed that computer scientists, in arguing for scientific status of their field, drew on “three distinct meanings of science (sometimes in combination)”¹¹: (1) science as the study of natural phenomena (information in this instance),¹² (2) science as the derivation of abstract ideas from concrete phenomena,¹³ and (3) the experimental method as the defining characteristic of science.¹⁴

      These assertions about computer science as a science did not persuade NSF management that computing was or was beginning to be a mature scientific discipline. Abbate notes: “. . . organizational control wielded by the established disciplines, as well as NSF’s emphasis on basic research, put the emerging field of computer science at a disadvantage. In this context, the notion of computing as a ‘science’ and the appropriateness of NSF funding for computing researchers were both contested.”¹⁵

      After NSF moved the Office of Computing Activities into the Research Directorate, renaming it the Division of Computing Research (DCR) in 1974, the weak support for computer science as a discipline resulted in DCR programs being placed in a section (CSS) within a Mathematical and Computer Sciences Division in the Mathematical and Physical Sciences, and Engineering Directorate in 1976. When DCR was created, Gordon Bell, then with Carnegie Mellon University and Digital Equipment Corporation, was “concerned about funding for computer science within the National Science Foundation and that we [the computer science community] lack representation on the National Science Board.”¹⁶ Saunders MacLane, a Chicago algebraist on the National Science Board (NSB), was a good supporter of computer science but not a true representative of the discipline. NSF provided a 12.2% increase for Computer Science research for FY 1976, while MPE overall was increased 6.3%. The $13.22 Computer Science research budget, however, was only 6.6% of the total MPE budget.

      To offset the perception that computing research was well-served by industry, Bell argued that funding for basic research in computing should be directed to universities and not industry. Bell added that while mission agencies, such as ARPA, played a significant role, NSF had the role of supporting basic computer science research. Bell also suggested that NSF funding of basic computer science research introduce a “question of scale” and that NSF consider investments of an ARPA-like magnitude in several non-ARPA-funded, leading computer science programs.¹⁷

      Facing skepticism from NSF leadership about the emerging field of computer science and its core research questions, John Pasta and Kent Curtis mobilized influential scientists. In 1974, they funded the Computer Science and Engineering Research Study (COSERS) under the direction of Bruce W. Arden of Princeton University. “For the first time in its quarter century of activity . . . this discipline will be given a comprehensive examination by researchers in the field. . . . The report will define what computer science and engineering is, describe major research problems now under investigation, and point out future educational and research opportunities.”¹⁸ Apart from a brief progress report¹⁹ in 1976, the massive 1000+ page report, What Can Be Automated?: Computer Science and Engineering Research Study,²⁰ unfortunately did not appear until 1980. By that time, other influential reports had appeared and diminished its impact.

      By the late 1970s, Curtis and Pasta were working with leading members of the computer science community to address a serious concern about the health of academic computer science. Academic salaries were falling behind industry salaries, there was a significant lack of computing equipment except at the ARPA-funded departments, undergraduate enrollments were rising, and many scientists, engineers, and policymakers still viewed computer science as consisting only of programming, computing applications, and hardware development. Faculty, new PhDs, and promising graduate students were leaving academia for industry in large numbers.

      The NSF sponsored a workshop in Washington, DC, on November 2, 1978, led by Jerome Feldman (Rochester) and including Gordon Bell (DEC), Bernard Galler (Michigan), Patricia Goldberg (IBM), John Hamlin (Missouri), Eliot Pinson (Bell Labs), Ivan Sutherland (CalTech), and William Robert “Bert” Sutherland (Xerox PARC). The “Feldman Report,”²¹ also published in the Communications of the ACM, called for universities to recognize the special resource needs of experimental computer science, use appropriate criteria in evaluating experimental computer science programs СКАЧАТЬ