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

Чтение книги онлайн.

СКАЧАТЬ believe the process I have described led to thorough and thoughtful reviews and recommendations, which corrected the perception that computing research proposals were of comparatively lower quality. The number, breadth, and quality of the research the Computer Science Section supported under its constraints and with limited funds demonstrates an effective stewardship of NSF investments in a growing field.

2.5Funding the Innovators in Computer Science

      It is not easy to measure the impact of individual funding decisions on the health and growth of computer science. One indicator might be the role NSF played in the careers of Turing awardees. The A. M. Turing Award is the oldest and most prestigious award⁵⁵ in computing. It is presented annually by the Association for Computing Machinery (ACM)⁵⁶ “to an individual who has made lasting contributions of a technical nature to the computing community.”

      Many of the Turing Award winners from the 1960s and 70s were in industry (Maurice Wilkes, Richard Hamming, Charles Bachman, John Backus, and Kenneth Iverson), Europe/Israel (Wilkes, Jim Wilkinson, Edsger Dijkstra, Michael Rabin), or the (D)ARPA-funded universities (Alan Perlis, Marvin Minsky, John McCarthy, Don Knuth, Allen Newell, Herb Simon, Dana Scott, and Bob Floyd). However, Don Knuth received significant NSF funding for the work that went into his The Art of Computer Programming⁵⁷ series and the development of TEX.⁵⁸ Both Alan Perlis and John McCarthy were involved with NSF facilities grants in the 1960s, which provided them with an environment for their early work on programming languages and operating systems. John McCarthy was funded by multiple NSF programs during the later 1970s.

      During the period from 1976 to 1978, the Computer Science Section launched the research careers of many future Turing Award winners. Of the winners from the 1980s through 2017, again some spent most or all of their careers in industry/government or outside the United States.⁵⁹ From 1976 to 1978, the Theoretical Computer Science (TCS) program made grants to Richard Karp, John Hopcroft, Robert Tarjan, Juris Hartmanis, Manuel Blum, Amir Pnueli (as a visitor at University of Pennsylvania), Andrew Yao, Leonard Adelman, Ronald Rivest, Adi Shamir, Judea Pearl (with Intelligent Systems), Martin Hellman, and Whitfield Diffie (Hellman and Diffie with Engineering Systems). Michael Stonebreaker was funded from the Special Projects program in 1980. At a later time, the TCS program funded Leslie Valliant and Shafi Goldwasser. Edward Clarke, Alan Emerson, Barbara Liskov, and Leslie Lamport all received NSF grants from the Software Systems Science program.

      Many of the Turing Award winners, including those in industry, played significant roles in advising and advocating for computer science within NSF, including John Hopcroft and Fredrick Brooks, who both served on the National Science Board (NSB). Vinton Cerf and Robert Kahn were important to the development of CSNET and NSF, and Cerf recently served on the NSB.

      In addition to the Turing awardees, there were other important grants made in the 1976 to 1980 period. I discuss cryptography and security below, including the work of Hellman and Diffie (Stanford); Rivest, Adelman, and Shamir (MIT); George Davida (Wisconsin Milwaukee); and Dorothy Denning (then at Purdue and SRI). Lawrence Landweber’s Theorynet project and the concurrent analysis of collaboration over networks by Starr Roxanne Hiltz were important in building support for CSNET and later NSFNET. The Coordinated Experiment Research (CER) program addressed the national issue described by the Feldman and Snowbird reports, but the successful grantees would not have succeeded without the equipment grants (typically VAXes and PDP-11s), which initiated experiment work in the grantee departments and established their credibility as potential centers of experimental research.

      In Chapter 7, William Aspray describes research done by some of the most respected, NSF-supported scientists. During 1976–1978, most of these people were funded by the Computer Science Section. Mary Shaw and Barbara Liskov were among many influential women researchers funded by CSS in the late 70s—a group that included Sue Graham, Sherry Turkle, Irene Grief, Lori Clarke, Anita Jones, Mary Jane Irwin, Ruzena Bajcsy, Nancy Lynch, Diane O’Leary, Shari Pfleeger, Elaine Cohen, Sheila Griebach, Dorothy Denning, and Naomi Sager.

      Additional grants from this period illustrate the impact of the Computer Science Section. The work of Arthur Burks and John Holland became the basis of classifiers used in machine learning. The Stanford AI lab (with John McCarthy, Edward Feigenbaum, and Cordell Green) moved artificial intelligence ahead. The Ingres Relational Database developed by Michael Stonebreaker, Lawrence Rowe, and Eugene Wong was arguably the first practical research relational database. Concurrently, the Division of Information Science and Technology funded early work in Information Retrieval by Gerald Salton, Naomi Sager, Michael McGill, and several others.

2.6Facilities

      In his book on applications of case study research, Robert Yin took David Gries’s abstract from the final report on the first Coordinated Experimental Research (CER) grant to Cornell and analyzed it as a case study. Yin quotes Gries’s final report and identifies the outcomes:

      From 1980 to 1986, the Computer Science Department at Cornell was radically transformed from a theoretical, pencil-and-paper research operation to one with a high degree of experimental computing. The departmental computing facility grew from a VAX780 and a PDP11/60 to an integrated complex of almost 100 workstations and UNIX mainframes. All faculty and graduate students now use computing daily, and much research that was hitherto impossible for us is now being performed.

      The CER grant enabled the department to attract bright young faculty who would not have joined a department with inadequate facilities. As a result, the department has been able to branch out into new areas, such as VLSI, parallel architectures and code optimization, functional programming, and artificial intelligence. The CER program did what it set out to do: It made it possible for the department to expand its research activity, making it far more experimental and computing intensive while still maintaining strong theoretical foundations.⁶⁰

      The CER program was transformative in the ways that Gries describes. Earlier support for the VAX780 and a PDP11/60 likely came from the Computer Science Research Equipment program. The program made seven grants in FY 1977, totaling $753,200, to departments that later received CER grants: Cornell, Arizona, Illinois, Pennsylvania, Utah, Washington, and Wisconsin. It also made grants to other departments, totaling $817,700, some of which competed unsuccessfully in the CER program. In FY78 the program made eight more grants, totaling $790,249, to departments that later received CER grants: Brown, Stony Brook, Berkeley, Illinois, UMass, Utah, and Wisconsin; and additional grants totaling $755,403. In many ways, the equipment program was as important to the computer science community as CER, moving departments from “pencil and paper” to a point where they had facilities for experimental research. The program continued, under various titles, from the 1980s until 2001, when the CISE Research Instrumentation program was incorporated into a revised CISE Research Infrastructure program along with the successors to the CER program. I will return to the CER infrastructure programs below.

      There were attempts to develop a national research network, or regional ones, prior to the Office of Computing activities move to the Research Directorate. Don Aufenkamp, then head of the OCA Applications in Computing Section, announced at the 1972 EDUCOM meeting that NSF was going to sponsor research that might lead to a network linking universities and other institutions.⁶¹ He and Ed Weiss delivered a paper⁶² at the International Conference on Computer Communication in October 1972 discussing further details. In Science in October 1973, Greenberger et al. noted that NSF had funded

      . . . EDUCOM to bring together interested users and administrators with those possessing shareable resources and relevant experience in a series of three 2-day working seminars. . . . The seminars . . . were designed to help identify the central organizational, political, and economic issues in building and operating networks on a national basis.⁶³

      What happened to this effort is not at all clear. It may have been inspired by the success СКАЧАТЬ