Computing and the National Science Foundation, 1950-2016. William Aspray
Читать онлайн книгу.
supplies supercomputer resources and network access to programs in all directorates. CISE will initiate programs to facilitate more effective use of supercomputers, including: understanding vector multiprocessors, improved algorithms, software development faster networks and high speed graphics workstations for more effective and enhanced use.157
Bell decided to argue that CISE encompasses all fields in which the major fraction of the intellectual discipline is computer science or engineering (e.g., robotics, VLSI, signal processing), and that other disciplines would have a “non-trivial” portion of their budget devoted to computing as an “experimental apparatus” and would be responsible for their own applications and for utilizing the computer as a simple component. CISE would “provide the scientific and engineering knowledge for these fields.” Bell wondered if educational activities, including supercomputing training and multidisciplinary projects, should be included.
In a memorandum dated February 27, 1986, to Bloch, Clutter, and Engineering AD Nam Suh, Bell proposed that CISE should include the Computer Research Division (from Mathematical and Physical Sciences), the Information Sciences and Technology Division (from Biological and Behavioral Sciences), the Office of Advanced Scientific Computing, including NSFNET, programs in real time computing applied to signals and communications systems, image understanding, and systems theory (from the Engineering Science in Electrical Communications and Systems Engineering Division), Computer Engineering and Manufacturing Engineering for Computers and Semiconductors (from the Design, Manufacturing and Computer Engineering Division), the Columbia University Engineering Research Center, the Advanced Technology program (from Science and Engineering Education), and the EXPRES Project.158
Alarmed by Bell’s wide-ranging vision, Nam Suh responded in a memorandum to Bell (copied to Bloch and Clutter) dated February 28, that “the only thing that really deals with the essence of Computer and Information Sciences and Engineering that you ought to take into your new directorate is Computer Engineering. In the rest of the programs, the computer is a peripheral tool, but not the intellectual driving force behind them. You will find that this view is widely supported in engineering schools throughout the country.” Suh added, “Sometimes we have the feeling that this world evolves around computers [but the] role of computers in our society has got to be looked at in the proper context.”159
Bloch issued a memorandum,160 dated March 3, 1986, to all NSF staff indicating that he was officially appointing Bell as a consultant to assist him in reorganizing the NSF computing activities with the intention of naming him AD/CISE. Bloch intended to “consolidate into a new directorate several computer-related divisions and programs [including] the Division of Computer Research (MPS); the Division of Information Science and Technology (BBS); the Office of Advanced Scientific Computing (O/D); and certain engineering programs from ENG.” In an attachment, Bloch stated the following rationale—that creating CISE:
(1) Brings together ongoing activities now spread among several NSF units; (2) Simplifies formulation and coordination of new policy directions; (3) Makes it possible to deal easily with full span of functions, from basic research through systems engineering in an area critical to national well-being; (4) Facilitates internal management; takes program activities out of Director’s office and puts them into a technical area; and (5) Will [create a] small disciplinary research directorate—in the range of $110–130 million, [with an] approximately 50 person staff drawn largely from other parts of NSF.
The allocations of budget and personnel attached to Bloch’s memorandum are shown in Table 2.1. The budget increase from the FY 1986 current plan to the FY 1987 Estimate is $21.89 million (19.5%), but with almost half of the increase ($10.35 million) going to OASC.
Table 2.1Bloch’s initial allocation to CISE (in $ millions)
a. Staffing did not include the approximate 6 for the AD Office.
b. Portions of DMCE and ECSE yet to be determined, with the transfer amounts estimated in the $10–30 million range. Amounts shown are mid-range estimates.
During March 1986, many people in the Engineering Directorate became alarmed about the potential scope and definition of CISE. Nam Suh was not happy with Bloch’s initial decision and mobilized161 members of his Advisory Committee (Frederick Garry, Sheila Widnall, Lester A. Gerhardt, Paul C. Jennings, and Herbert H. Richardson) at NSF on March 18, 1986. Meeting attendees also included Suh’s Task Group on Computing (Herbert Voelcker chair, with program directors Alan de Pennington, John Mayer, Howard Moraff, Michael Gaus, Michael Polis, Elias Schutzman, and Donald Silversmith) and Frank C. Huband, Division Director of Electrical, Communications, and Systems Engineering (ECSE). Voelcker was also the Deputy Division Director for Design, Manufacturing, and Computer Engineering (DMCE). ECSE and DMCE were the Engineering divisions most likely to be impacted.
The Engineering Advisory Committee had access to Voelcker’s Task Group report,162 which considered a “broad” and a “narrow” construct for CISE; but the draft report failed to get the full support of the Task Group. The Task Group also analyzed the DMCE Computer Engineering program in a report163 concluding that many elements of the Computer Engineering program had stronger connections to the Engineering Directorate programs than to the CISE programs. The committee members were particularly concerned about a home for the joint DARPA-NSF MOSIS VLSI fabrication facility, which became a key activity of the MIPS Division in CISE. Norman Caplan, the Deputy Division Director of ECSE, forwarded to the attendees his memorandum164 to Nam Suh concerning robotics, which raised concerns about the definition of the field of robotics being assigned to CISE.
Following the March 18th meeting of the Engineering Advisory Committee, Frank Garry, its chair, wrote to Bloch that the Committee concurred with “the consolidation of the following into the new CISE Directorate: DCR/MPS, IST/BBS, OASC/OD and the Program in Computer Engineering from the Engineering Directorate.” Garry went on to say that “the remaining Engineering programs outlined in Gordon Bell’s memorandum of February 27 have their intellectual base in the Engineering Directorate” and we “fear that their transfer to CISE would narrow their focus and eventually erode their disciplinary strength.” The recent reorganization of Engineering had been carefully constructed and based on broad input from the Advisory Committee, the National Academy of Engineering, and other members of the engineering community. It “would be precipitous to alter the [d]irectorate’s programs in a major way without a similar review.”165
Other Engineering Directorate managers also pushed back against Bell. Frank Huband stated, “The creation of a computer-related directorate is an exciting event, and has the potential to create new opportunities for development in this important discipline” but research funded in CISE “must pass muster as computer-related” . . . the practitioners in non-computer disciplines “want—and I believe deserve—an independent home for their research proposals.”166 “Assignment of program elements to ENG or CISE should be governed by considerations of their fit to the respective missions of the directorates, and of their relative contributions to strengthening U.S. research in computing or in engineering,”167 suggested program director Howard Moraff. He also raised concerns about possible disruptions to the new programs created in Nam Suh’s recent reorganization of Engineering and urged collaboration between CISE and Engineering.
Bell wrote to Bloch defending his approach to organizing the directorate, suggesting:
The rationale for Engineering disciplines in CISE is that contemporary engineering naturally divides into two parts: those research areas based on the physical and biological sciences and those which deal with information. This provides a working boundary: on one side are research areas governed by the physical transformations, and on the other side those research areas concerned with the transformation of information from one form to another.168
In Bell’s personal notes dated March 20, 1986, he characterized the Nam Suh position as being that CISE should be “pure computer,” while Bell himself thought it should be “pure information processing (and storage,