Click for next page ( 140


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 139
5 Recommendations This report emphasizes that the computing technology that un- derpins so much of modern society is in large measure the result of past advances in CS&E. Where CS&E goes in the future will do much to shave and influence future developments in computing practice ~ ~ ~ . . . ~ . - . . . 1 1 ~ ~ and therefore to equip tne nation to meet tale social ana economic challenges that dominate public concern and policy. Unfortunately, the resources available to support CS&E are not nearly as bountiful as the potential applications of computing to eco- nomic and social needs. Various constraints always force policy makers to make decisions about how to allocate resources, and thus the com- mittee believes it is important to articulate a set of overall priorities for the field that describe a philosophy within which its subsequent recommendations are framed. OVERALL PRIORITIES Priority 1: Sustain the CS&E Core The first priority is to sustain the core effort in CS&E, i.e., the effort that creates the theoretical and experimental science base on which computing applications build, bearing in mind that the core effort is highly dynamic as the result of rapid changes in the field. This core effort has been deep, rich, and intellectually productive 139

OCR for page 139
140 COMPUTING THE FUTURE and has been indispensable for its impact on practice in the last cou- ple of decades. But this track record of success has a down side, in the sense that any field with a long history of successes risks being taken for granted. Only by continuing support for the core effort (support whose importance to the nation may well grow if industrial CS&E research is cut back substantially in the future) will the field continue to make progress that is broadly applicable over many fields of inquiry and areas of human endeavor. While tantalizing successes have been achieved with promising technologies such as distributed and parallel computing, object-oriented programming, and graphical user interfaces, the full practical exploitation of these and other com- puting technologies will require considerable research in the future. The committee notes with approval that federal funding agencies ap- pear to recognize the importance of continued support for core CS&E activities, and it wishes to encourage this trend in every way possible. The committee calls attention to its use of the word "sustain." Many in the CS&E community (and some on the committee itselfl have been concerned about the increasing tightness in the availabili- ty of research funding for core topics in CS&E and have argued with some cogency that the first priority should be to strengthen rather than merely sustain the core. Advocates of this position would say that the track record of CS&E in research and education has been so positive and successful that it speaks for itself, that there is not enough support for computer scientists and engineers to perform the "core" research in CS&E that will be necessary in the future, that computing technology will improve as the result of advances in CS&E, and that the information revolution promises to develop as it has in the past. Why, these individuals would argue, should a winning research agenda be changed? The committee is sympathetic to this perspective and would have liked to recommend a substantial increase in such funding, especially in light of the growing numbers of academic CS&E researchers rela- tive to available research funding. However, it concluded that such a recommendation amounting in essence to "we should continue to be supported in the style to which we have been accustomed"-would have been seen unfavorably by policy makers as an entitlement argu- ment, particularly in view of the substantial increases in research funding that will be made available to the CS&E community by the HPCC Program. In the committee's overall judgment, more benefit is likely to accrue to the field and the nation if the broadening course is taken rather than if efforts at the core are redoubled. The reason- ing is clear: relatively few CS&E researchers are devoted to the pur- suit of interdisciplinary and applications-oriented work, while rela

OCR for page 139
RECOMMENDATIONS 141 lively many are devoted to investigating problems at the core, and human and fiscal resources devoted to the former are likely to have a more significant impact. Accordingly, the committee was led to its second priority. Priority 2: Broaden the Field The second priority is to broaden the field. Given the many intellectual opportunities available at the intersection of CS&12 and other problem domains and a solid and vigorous core effort in CS&E, the committee believes that academic CS&E is well positioned to broaden its self-concept. Given the pressing economic and social needs of the nation and the changing environment for industry and academia, the committee believes that academic CS&E must broaden its self-con- cept or risk becoming increasingly irrelevant to computing practice. More specifically, academic CS&E must: Increase its contact and intellectual interchange with other dis- ciplines (e.g., other science and engineering fields). Increase the number of applications of computing and the quality of existing applications in areas of economic, commercial, and social significance, and understand that from such applications substantive CS&E problems often emerge. Embrace the creation of significant new knowledge and de- monstrable intellectual achievement as the relevant standards of mean- ingful scholarship in CS&E, rather than focusing on artificial distinc- tions among basic research, applied research, and development (as discussed in Chapter 2~. Increase traffic in CS&E-related knowledge and problems among academia, industry, and society at large, and enhance the cross-fertil- ization of ideas in CS&E between theoretical underpinnings and ex- perimental experience. Such broadening would serve the interests of society at large by coupling the formidable intellectual resources of academic CS&E more directly to the practice of computing, thereby increasing the likeli- hood that the full potential of computing can be realized. It would also serve the field by increasing intellectual opportunities and di- versifying the sources of funding. Priority 3: Improve Undergraduate Education The third priority is to improve undergraduate education in CS&E. As discussed in Chapter 4, undergraduate CS&E education is highly

OCR for page 139
142 COMPUTING THE FUTURE variable in quality and outlook from institution to institution. Given the importance of CS&E to computing practice and the large flow of those with undergraduate CS&E degrees to business and industry, the quality of undergraduate CS&E education is inextricably tied to the state of computing practice in all sectors of society. Moreover, better undergraduate education is necessary for better research, since it is necessary for transmitting recently developed core knowledge to the next generation and for providing the intellectual basis in CS&E for individuals pursuing a broader research agenda. Thus, improv- ing undergraduate education is a necessary component of both prior . . sties. The natural evolution of undergraduate CS&E education will ul- timately result in the synthesis and dissemination of modern approaches to CS&E, enhancing the present skills and future adaptability of CS&E graduates as well as providing a good foundation on which to build knowledge in other fields. But natural evolution occurs on the time scale of several decades. A major program aimed at accelerating the process could reduce the time to a decade or less. More importantly, the nation is likely to reap considerable benefits from such a pro- gram, since undergraduate CS&E programs from non-Ph.D.-granting institutions supply a considerable fraction of the computer specialists responsible for implementing and maintaining the software systems in all areas of application that underlie the information age. To suggest more specifically how these priorities translate into an action plan, the committee grouped its recommendations into two categories: research) and education. Each category contains action items for universities and federal funding agencies. Taken together, these action items constitute a coherent plan that will improve the state of the CS&E discipline on a much shorter time scale than would otherwise be possible, to the benefit of the discipline and the nation as a whole in a rapidly changing world. All of the action items described below will demand considerable leadership from the academic CS&E community. If the community is to adapt to changing circumstances in a proactive and constructive manner, senior researchers in the academic CS&E community the ones whose words and actions shape the values of the community- must take the lead in promoting the cultural changes necessary for success in the new environment. Moreover, senior academic researchers in the CS&E community are widely regarded as spokespersons for the discipline, and their continuing presence and participation in policy debates in both the executive and legislative branches of the federal

OCR for page 139
RECOMMENDATIONS 143 government will be necessary for years to come if federal policy and funding are to evolve in the best interests of the field. RECOMMENDATIONS REGARDING RESEARCH To Federal Policy Makers As noted in Chapter 1, federal policy toward computing and CS&E has had an enormous impact on the field's shape and development. As the scale of computing activities increases, the importance of a strong federal role can only grow. Recommendation 1. The High Performance Computing and Com- munications (HPCC) Program should be fully supported through- out the planned five-year program. Full support for the HPCC Pro- gram will entail about $3.7 billion dollars over the next four years, or about 1.2 percent of the entire federal research budget.2 The HPCC Program is of utmost importance for three reasons. The first is that high-performance computing and communications are essential to the nation's future economic strength and competi- tiveness, especially in light of the growing need and demand for ever more advanced computing tools in all sectors of society. The second reason is that the program is framed in the context of scientific and engineering grand challenges. Thus, although the program will sup- port research and development in a variety of fields, the program is a strong signal to the CS&E community that good CS&E research can flourish in an applications context and that the demand for interdis- ciplinary and applications-oriented CS&E research is on the rise. And finally, a fully funded HPCC Program will have a major impact on relieving the funding stress affecting the academic CS&E community. Consistent with Priority 1, the committee believes that the basic re- search and human resources component of the HPCC Program is critical, because it is the component most likely to support the re- search that will allow us to exploit anticipated technologies as well as those yet to be discovered through such research. The committee is concerned about the future of the HPCC Pro- gram after FY 1996 (the outer limit on current plans). If the effort is not sustained after FY 1996 at a level much closer to its planned FY 1996 level than to its FY 1991 level of $489 million, efforts to exploit fully the advances made in the preceding five years will almost cer- tainly be crippled. In view of the long lead times needed for the administration's planning of major initiatives, the committee recom

OCR for page 139
144 COMPUTING THE FUTURE mends that funding necessary for exploitation of recently performed research and the investigation of new research topics be fully as- sessed sometime during FY 1994 with an eye toward a follow-on HPCC Program. Recommendation 2. The federal government should initiate an effort to support interdisciplinary and applications-oriented CS&E research in academia that is related to the missions of the mission- oriented federal agencies and departments that are now not major participants in the HPCC Program. Collectively, this effort would cost an additional $100 million per fiscal year in steady state above amounts currently planned.3 For the participating agencies, the HPCC Program is a good mod- el for how to encourage interdisciplinary and applications-oriented research in CS&E. But many federal agencies are not currently par- ticipating in the HPCC Program, despite the utility of computing to their missions, and they should be brought into it. (As noted in Chapter 2, 12 federal agencies controlling over $10 billion in FY 1991 obligations for scientific research each allocated less than 1 percent to computer science research.) Such agencies can be divided into two groups: those that sup- port substantial research efforts, though not in CS&E, and those that do not support substantial research efforts of any kind. The commit- tee believes that both groups would benefit from supporting interdis- ciplinary or applications-oriented CS&E research, but for differen reasons. Support of interdisciplinary CS&E research, i.e., CS&E research undertaken jointly with research in other fields, should be taken on by mission agencies with responsibilities for those other fields. That research will often involve an important computational component whose effectiveness could be enhanced substantially by the active involvement of researchers working at the cutting edge of CS&E re- search. Examples of interdisciplinary CS&E research were discussed in the Chapter 2 section "A Broader Research Agenda." The case for support for applications-oriented CS&E research from agencies that do not now support research is less obvious, but to the committee nevertheless cogent. While these agencies are generally focused on operational matters (e.g., processing tax forms or income support payments) and thus are expected to make the best use of available technology, it may be that in many cases the efficiency of their operations would be substantially improved by some research advance that could deliver a better technology for their purposes. A

OCR for page 139
RECOMMENDATIONS 145 case in point is the research in the application of image-recognition technology to the processing of government forms that is being per- formed by the National Institute of Standards and Technology in support of the Bureau of the Census and the Internal Revenue Ser vlce.~ Moreover, the federal government's computer operations are of- ten conducted at scales of size and complexity whose ramifications are poorly understood. Without an adequate understanding of these ramifications, it will be difficult to prevent computer-related disaster or reduce the likelihood of computer-related inefficiency or fraud. Given its operational responsibilities, the federal government must do the best it can with what it has, but CS&E research undertaken to better understand these problems could have substantial payoff later with respect to reliability, security, efficiency of operation, lower cost, and so on. An additional benefit is that applications explored and developed in such a context may have considerable "spin-off" benefit to the private sector, since many government information processing needs (e.g., for security) are similar to those found in the private sector.5 How can the talents of the academic CS&E research community be tapped to provide maximum benefit to the nation in these inter- disciplinary and applications-oriented areas? A first step would be to establish a research program within mission agencies that would tap the talents of CS&E researchers in the service of each agency's own needs. This may be easier said than done, since CS&E research- ers interact primarily with only the four federal agencies that con- tribute the bulk of CS&E research supports group of agencies that places a high value on research and provides many opportunities for interaction between agency staff and researchers. The very existence of such a program would prompt strong interest on the part of aca- demic computer scientists and engineers in pursuing interdiscipli- nary and applications-oriented research, but some care must be taken to ensure that the "bridging of cultures" between CS&E and others is successful. Agencies might jointly sponsor research on problems of collec- tive importance. For example, several agencies process vast amounts of Eager and might benefit from advanced imaging and database technologies; another group of agencies might have a special interest in using computers and communications to facilitate service for the disabled. When the work is specified and undertaken, it is essential that such work be done by investigators from CS&E and other disci- plines and areas who regard each other as intellectual equals; only in

OCR for page 139
146 COMPUTING THE FUTURE this way will it be possible to maintain both an understanding of the future state of the art in computing and an appreciation of the real problems in the application domain. One way of ensuring true col- laborative work is to consider only proposals whose principal inves- tigators are drawn both from both CS&E and some other discipline or area. The location of such a program within the federal government is a sensitive issue. On balance, the committee believes that the exist- ing HPCC Program provides the most reasonable home for this pro- gram, subject to one crucial provision to be discussed below. (Other organizations have developed similar positions; for an example, see Box 5.1.) The HPCC program has strong support from Congress, the White House, and the Office of Management and Budget; thus indi- vidual agencies have strong incentives to participate. Most impor- tantly, the Federal Coordinating Council for Science, Engineering, and Technology (FCCSET) is already in place and can facilitate inter- agency collaborations. In performing the coordinating role for this new program, FCCSET would approach agencies not already participating in the HPCC Pro- gram, such as the Department of the Treasury and the Department of Transportation. It would also be appropriate for FCCSET to ask large commercial users of computers to indicate what CS&E research might be relevant to their needs. Such users (and the computer industry) might be willing to support applications-oriented research to a cer- tain extent, especially if such support could be leveraged (or matched) by federal dollars. The committee recognizes a certain danger in recommending that the HPCC Program be augmented to provide for this new program. In particular, it is concerned that planned HPCC budgets would sim- ply be reprogrammed to accommodate this program. Such repro- gramming would be inconsistent with the framework of priorities laid out above. It is the intent of this recommendation that agencies that have not traditionally supported CS&E should also participate in the HPCC Program; along with such participation should come additional re- sources from those agencies. These resources would support research that would contribute directly to the goals of those agencies by im- proving the efficiency of computing practice in support of those goals. Box 5.2 describes some additional implementation issues that this effort could entail.

OCR for page 139
RECOMMENDATIONS . ... ....... .... .. ....... .................... . .......... "' ' .''.'.'' 8~\'' "'P05 TION''''OP''THE'''COM'PC'TE'R"""'SYSTE'' S"""" .......................................... ........... .............................. ....... .... ... ....... .................... ..... . ~ a , ~' ~l air ~ ~ r~ . ~., ~. ~ ~ .. ~. ~.. I ..... .. ~ - ...... .~ ~., , ............................................................................... . ~, ~, t t , ~.~.r . ~ ~.t te ~.~h, ~.~. est .~. ~, ~.~ . ..... . .... . ........... H ~ ~1 the m - ~m ha~ ~e and network1 ~h ..... . . . . .... ........ .... ........ ..... ... - : I. of what blUh ner~:~e comOutIn~ and ~ .. ::: ~ ~ :. . .. ke A.. A.... ret ThIS..~q {~s ex .. . . . ... .. . .. ... ~ . .. . . . .... . ~n ~ the Hw~ {~m .b .. I.. and...~C .~s In amas of .~t ~ the. ~.~t . .. . . . . . . . . . . . . .. ~ .. . . . . . : :: : ::::::: ::::::: :: :: ::: ::: :::::: ::::: :::::: :::::::: :. :: ::::: :: ::: ::::: :- :::: ::: ::: ::: ::::: ::: :: ::::: .. am ~ ~a . pU ~ tc ... ~ as aovan=s In . ne ~ .~ : ~ : . and sew Ces ~t sen ot (~t 7605: Impm - ~s tn. ~t On and. o. . pO~U ~ [.~] es . m(. . ~ he tong. . [~a rn {.n . . . Ilsent .~r ng ..................... ... . . . . . .. .. . . . ... . f .. .... .. .2 p . . -.- - - ........ -.- - - -.- .-.: -.- - -.- - .... teverag~s ~ ~] tn~s ttn compu~r re~ch) the ~ . ..................... ........................................................................................................................... ............................ ~. l. b~} ~ 1t b} ~ ttc ~ tell smch , ............... ..................... .......................... a,06 ~.,t,~.~.~m,.,., ~e, , ~.~.~.~.~. ~.~. ~ '.'. '.' 2'''s' ' '' " "'' ' ''' '"' ''"' '"'""" ................................................. ... ..... . . ................... ~. ............................ ..... . ~- ................ ................. ... ...... I~ ~ - ~ ~ ~ ~ ~ S~ ~ I- .................................................................... ' ,'.'.' ~'.~.''2~'~m''' ' ' ' work1'ng''m86~G the'~n ent 'tnous '' ....................... ~. ~and ~b ~. ~. ............................................................ ,.,, ~.~.~., ,3.n HPcc pmmm tnat-~t contr1~e ~ ou ............................ ...... .... .. .................. m~. many of the w16~: engrn : : :~ :: ::: :: : :::: ::: :::::: .: :: ::: : ::: ::: :: :::: ::: : :: : ~ ~ rE ~ dd ~ ~: :: :::::: ::: : ~ ::::: :::: :: :: : :::: . . : . ~ . . . . . . . . . . .... . ... . . :. t~w ~ ~: :::. ~ tIe wompu.~r >~.~.~....~.~..~L ..~.F r' . . : :: : :: :. .~e 12 m~r ~ sy comp~r ~ ~m . .. . . . .. . ~ ... . . . . . cray~...~.. ~ ......... ~ .. .~ ... 118t EnuI~nt cOr=~n HewI~t Pac~d I DM Sun M1~m ........................................... ? ~ .......................... ~a .~.~ ~.~. ~, ~. ~. ~. ~.~.~. ~.~.~.( ~.~ ............. . .. . . ......... ....... - ........................ ~. ~m and. the ~. ~.~.~. ~ ..... ..... . . .. .. . . ...~e IO~ tan~ po! ~ es - h [n .~e d - ~.~.~. ~ ~: :::: : :::::: ~ : :! ::: :::: . :::::: :: :::: . :: :: .::::::: : ::: :::::::: ::: ::: :: : :~:::: : t::: ::::::: ~ ::::::::::::: ::::::: ~::::: :::::: ::: ::::::::::::::::::::::::::::::: ~ ::::::: :~ ^: ::: ::: . ~ tn tne .~.~. m. ~ 1e ~ ~ [~e posttIon gIoDa ~ .y :: ~ ~: ::: :: : :: ::: : ::: :::: :: :::: :::: :::: :: :: ::: ::::::::::~:~:~:~:::::: ~::::::: :::::::::: :: :: :::::: :: : :::: ::: ::: ::: : .: .............:.... pumr Systems Poftcy Pm'ect ~ ~ W$~ ~f H~k ~ . ............ ........... . . . i... ~~ ~ - ~ ~ I=~o : :: :: ~ :: : ::: .::. Was.~.~::. D:~: .::: .1 991: ~: :. ~ : ~ ~ .: ~ : *: . ~ ~ :::: 147

OCR for page 139
To Universities COMPUTING THE FUTURE University policy will play a key role in broadening academic CS&E. Any one of the recommendations below may suggest a specif- ic action that has been taken in the past, but their collective strength is that they are part of a coherent strategy to broaden the scope of academic CS&E. Their implementation will define a leadership role for many senior CS&E faculty, who together have (or should have) very influential roles in defining the tone and character of unixTersi- ties and academic CS&E departments with respect to promotion pol- icies and the boundaries of "acceptable" research and education. Recommendation 3. Academic CS&E should broaden its research horizons, embracing as legitimate and cogent not just research in core areas (where it has been and continues to be strong) but also research in problem domains that derive from nonroutine comput- er applications in other fields and areas or from technology-trans- fer activities. "Nonroutine" applications are those that pose sub- stantive and intellectually challenging problems and may be best solved

OCR for page 139
RECOMMENDATIONS 149 by research collaborations with experts in the application area; some examples are provided in the section "A Broader Research Agenda" in Chapter 2. Current and future CS&E faculty should be encour- aged to undertake collaborative research both with faculty in other disciplines and with appropriate parties in industry and commerce,6 and in government laboratories. These collaborations will benefit both computer scientists and engineers (as a result of new intellectu- al challenges posed) and those from other problem domains (as a result of the more effective use of their computational resources).7 As argued in Chapter 2, the central focus of scholarship in CS&E should be activity that results in significant new knowledge and de- monstrable intellectual achievement, without regard for whether that activity is related to a particular application or whether it falls into the traditional categories of basic research, applied research, or de- velopment. To promote broadening, action should be taken to make the uni- versity environment more accommodating to interdisciplinary and applications-oriented research and to stimulate the interpersonal in- teractions needed for the successful conduct of such research. Uni- versity administrations and CS&E departments should: 3a. Develop and promulgate explicit policies that assure and inform all faculty members that research in interdiscipli- nary or applications-oriented areas or work oriented toward technology transfers will be competitive in the tenure and promotion evaluation process with work that is more tradi- tionally oriented, assuming that necessary standards of quali- ty and achievement are met. Such policies will require mech- anisms by which interdisciplinary and applications-oriented work can be evaluated, possibly including: (i) evaluation committees with members familiar with the intellectual requirements of the other (non-CS&E) problem domains represented in the work being evaluated. Such com- mittees will have to address the very problematic issue of how to interpret the traditional criterion of "demonstrable intellectual achievement" in an interdisciplinary or applica- tions-oriented context. (ii) ways to take into account the fact that meaningful evi- dence of intellectually substantive work in CS&E often takes the form of system demonstrations as well as the publication of journal articles, and thus that many CS&E experimentalists up for promotion or tenure may submit portfolios with fewer published papers than their peers.9

OCR for page 139
150 COMPUTING THE FUTURE 3b. Support CS&E faculty who wish to gain expertise in other fields so that they may more effectively pursue interdisci- plinary or applications-oriented research. Possible mecha- nisms for support could include: (i) establishment of short-term academic appointments (one to three years) that academic computer scientists and engi- neers could use to develop familiarity with and expertise in other areas. Such appointments would typically involve re- duced teaching responsibilities and could be held by new Ph.D.s and senior faculty alike. (ii) sponsorship of seminar series that describe challenging CS&E problems that arise in other disciplines. 3c. Invite qualified individuals from industry and commerce to serve on university and academic departmental advisory and review committees for CS&E programs. This is a common practice among some leading research universities, but the practice should be more widespread. 3d. Eliminate or reduce practices that impede intellectual con- tacts with industry. In particular, universities should con- sider greater use of more open arrangements with respect to the protection of intellectual property, such as cross-licensing for university-developed technology, rather than insisting on exclusive rights for themselves. Such practices conflict with norms in the computer industry and set up roadblocks to collaboration. 3e. Encourage CS&E research faculty to seek out nontraditional sources of funding to pursue interdisciplinary or applica- tions-oriented research. Nontraditional sources would in- clude the program described under Recommendation 2; fed- eral agencies other than DARPA, NSF, NASA, and the Department of Energy; large commercial users of computers; and state governments. As noted in Chapter 2, federal agencies with- out a tradition of supporting CS&E research may still control substantial research budgets. Recommendation 4. Universities should support CS&E as a labo- ratory discipline (i.e., one with both theoretical and experimental components). With respect to its need for equipment, many parts of CS&E are more like physics or engineering than like mathematics. CS&E departments need adequate research and teaching laboratory space; staff support (e.g., technicians, programmers, staff scientists); funding for hardware and software acquisition, maintenance, and

OCR for page 139
RECOMMENDATIONS 151 upgrade (especially important on systems that retain their cutting edge for just a few years); and network connections. New faculty should be capitalized at levels comparable to those in other scientific . . . . .. Or englneermg c .lsclpl~nes. RECOMMENDATIONS REGARDING EDUCATION To Federal Policy Makers The federal government has a history, dating to the days of Sput- nik, of taking strong and decisive action to improve science and math- ematics education in times of great national need. The committee believes that undergraduate CS&E education would benefit tremen- dously from such action today and that the benefits of such action will echo throughout all sectors of society. Recommendation 5. The basic research and human resources com- ponent of the High Performance Computing and Communications Program should be expanded to address educational needs of cer- tain faculty. In particular, college and university CS&E faculty who are not themselves involved in CS&E research and researchers from other scientific and engineering disciplines that depend on computa- tion need to become familiar with recent developments in CS&E. The program described below to address these needs is estimated to cost $40 million over a four-year period. As argued in Chapter 4, the lack of current knowledge about the approaches and intellectual themes of modern CS&E is an impedi- ment to the full exploitation of computing. This is true for those who teach undergraduate CS&E without the benefit of sustained contact with cutting-edge research as well as for many scientists and engi- neers whose education in computing was received many years ago. For these individuals, programs of continuing education to bring them up to date on recent developments in CS&E would have significant value. Such programs would enable them to develop their own ap- proaches to the subject material, informed on the one hand by expo- sure to the current state of the art and on the other by knowledge of local institutional needs, and they could have a major impact on the quality of undergraduate CS&E education in the United States, as well as on its ability to use computing in support of other science . . . anct engineering. As major players in pushing back the frontiers of CS&E through their research and in educating students through their teaching, aca- demic researchers are best equipped to take responsibility for dis

OCR for page 139
152 COMPUTING THE FUTURE seminating their knowledge to parties that could benefit from it. More specifically, it is their broad knowledge about important develop- ments in the field in the last ten years that is most important to disseminate to these parties, rather than their detailed knowledge about their own particular research specialties generated in the last couple of years. A continuing education program to meet the needs described above could first sponsor intensive month-long workshops to promote dis- cussion among the top researchers from academia. These workshops would focus on the problems of undergraduate CS&E education (e.g., content, scope, style, broadening, recruitment and retention of wom- en and minorities). Neither course development nor consensus among the individuals participating would be necessary outcomes of these workshops; instead, the object would be for participants to become acquainted with the various approaches to teaching undergraduate CS&E in order to provide a basic platform of understanding from which would emerge different ways to integrate various paradigms. Following these workshops, the participants would give a series of short courses for individuals who are not current with recent de- velopments in the field, including CS&E faculty at non-Ph.D.-grant- ing institutions, scientists and engineers from other disciplines, and appropriately qualified high school teachers. (The program would provide course leaders with financial support for the development of materials text materials, exercises, software, and so on. It would also provide some financial assistance for the workshop attendees.) The active participation of senior academic CS&E researchers is critical to the success of this program; indeed, participation could be seen as an active demonstration by these individuals of leadership for the field as a whole. Since senior academic researchers have, by definition, made their careers by performing research of extraordi- nary quality, it will take more than mere exhortation to persuade them to become substantially involved in educational matters. One mechanism to encourage their attention to such matters would be to couple research funding to participation in these workshops. For example, an augmentation fund for research grants could be set aside, for which only researchers taking part in these workshops would be eligible. Research proposals would be submitted and awarded through the ordinary review process; researchers whose proposals were suc- cessful and who had participated in these education workshops would be eligible to receive an additional amount from the augmentation fund to support their research as they saw fit. Alternatively, grant- awarding agencies might give some degree of preferential treatment to proposals received from participants in this program.

OCR for page 139
RECOMMENDATIONS 153 The estimate of $40 million for the total cost of this program is based on an assumed 100 researchers leading 400 short courses for 6000 other individuals.~ These funds should be an addition to the very important elements already covered by the basic research and human resources (BRHR) component of the HPCC Program. Like the original BRHR component, it is appropriate that the proposed con- tinuing education program be funded by most if not all agencies participating in the HPCC Program, although such a program could be administered within the NSF. To Universities Universities are the front line of educational delivery. If CS&E is to broaden, university policy and departmental programs must sup- port and encourage such change. Graduate CS&E education should reflect and be supportive of a broader research agenda. Only in this way can CS&E graduates understand the applicability of current and rapidly emerging future CS&E developments to the increasing num- ber of business, commercial, scientific, and engineering problems that have (or ought to haired a significant computing component. Recommendation 6. So that their educational programs will re- flect a broader concept of the field, CS&E departments should take the following actions: 6a. Require Ph.D. students either to take a graduate minor in a non-CS&E field or to enter the Ph.D. program with an under- graduate degree in some other science or engineering or math- ematical field. Those in the latter category may lack some of the skills and knowledge possessed by incoming graduate stu- dents with undergraduate CS&E degrees, but they can use the time that others would use for a non-CS&E minor to strengthen their CS&E background. The choice of graduate minor should be broad enough to allow the student a high degree of discre- tion to select the minor, but constrained enough that the stu- dent cannot evade the spirit of the requirement by selecting a minor in a field that is too closely related to his or her major interest. The committee recognizes that a recommendation of this scope may well generate considerable resistance in the affected departments, but it nevertheless believes that attempts to overcome this resistance will ultimately benefit the field. 6b. Encourage Ph.D. students in CS&E to perform dissertation research in nontraditional areas, as described in Chapter 2.

OCR for page 139
154 COMPUTING THE FUTURE In addition, expose Ph.D. students to a variety of projects and intellectual issues in their predissertation work. 6c. Offer undergraduate students not majoring in CS&E a wide range of CS&E courses and programs. By teaching other courses less frequently, CS&E departments might: (i) offer undergraduate minors in CS&E and/or general ed- ucation courses in computing. (ii) collaborate with other departments in teaching courses that familiarize non-CS&E undergraduate students with ad- vanced computational tools in the context of their own fields of interest. Such courses might have appeal to CS&E majors, thereby contributing to their broadening as well. 6d. Provide mechanisms to recognize and reward faculty for de- veloping innovative and challenging new curricula that keep up with technological change and make substantive contact with applications in other domains. In particular, find ways to give credit for the professional effort involved in develop- ing the following: (i) Laboratories. In both software and hardware engineering education, laboratories are essential if students are to obtain first-hand experience with the nontheoretical side of CS&E. In the fast-changing CS&E environment, laboratories must be completely revised frequently, i.e., every several years. (ii) Textbooks. Textbooks that are both good and current are important to CS&E and are difficult to produce as well. The commitment of effort and time needed to write a quality text- book is far greater than that needed to produce multiple re- search papers, and in the case of a fast-changing field such as CS&E the amount of professional competence and talent re- quired is often at least as great. (iii) Interdisciplinary courses. Given the requirements for a minimal level of applications-specific competence in teaching applications-oriented CS&E, the development of interdisci- plinary courses should be expected to take longer and be more difficult than teaching core CS&E courses, even if faculty from other disciplines are involved. Recommendation 7. The academic CS&E community must reach out to women and to minorities that are underrepresented in the field (particularly as incoming undergraduates) to broaden and en- rich the talent pool. As noted in Chapter 8, CS&E attracts women and minorities at all levels at about the same rate as the physical sciences. However,

OCR for page 139
RECOMMENDATIONS 155 CS&E is also significantly younger than the physical sciences, and to the extent that a younger field should be expected to be more inclu- sive of women and minorities, the field has an opportunity for out- reach that it is not fully exploiting. Moreover, the underrepresenta- tion of women and minorities in CS&E is particularly unfortunate given the impact of CS&E on society; their exclusion from CS&E will mean that their voices and values will not be heard as society is transformed by the information revolution. A secondary benefit of outreach is that such outreach might well contribute to achievement of a broader agenda. This report has ar- gued that the field will be enriched by interactions with those from other disciplines and fields. Recommendation 6c recognizes the need for these individuals to learn about CS&E. To the extent that women and minorities constitute a larger fraction of these fields than they do of CS&E, outreach programs for these groups should focus their at- tention on CS&E, thereby increasing the likelihood of coupling be- tween their "home disciplines" and CS&E. Computing practice as well as CS&E can only benefit from the greater inclusion of individu- als with a more varied set of perspectives and experiences. Outreach programs need to take into account the special needs and backgrounds of individuals from underrepresented groups so that more are retained within and attracted to the field. Although these programs are useful at all levels of CS&E education, under- graduate CS&E education is the point of highest leverage for the academic computer scientist or engineer. Thus outreach is an essen- tial element of improving undergraduate CS&E education. Additional Studies In the course of its deliberations, the committee identified several areas of special concern that should be addressed in future reports. These areas include: The computer infrastructure for undergraduate CS&E educa- tion in all CS&E departments. While important parts of the under- graduate CS&E curriculum are technology-independent, other aspects are strongly dependent on the technological state of the art. Without suitable, up-to-date equipment and software, it is impossible to ex- pose students to concepts and environments that will affect all as- pects of future practice. For example, very fast computers with large amounts of storage are necessary to support three-dimensional real- time graphics and certain new and important programming languages and systems. Keeping the educational computer infrastructure ap- proximately current with the cutting edge of technology will be an

OCR for page 139
156 COMPUTING THE FUTURE ongoing enterprise. The committee hoped to be able to make recom- mendations about the cost of a program that would keep educational institutions current technologically, but was unable to locate firm and relevant data on the subject. A study on this subject would address issues such as the mag- nitude of the need for new machines, current university policies re- garding replacement of educational computer equipment and soft- ware, community views on how current the educational computer infrastructure must be to support a good undergraduate CS&E edu- cation, and ways to fulfill the need in the most inexpensive manner possible. Continuing education for CS&E. In considering the views of the computer industry and large commercial users of computers, the committee concluded that the needs for continuing education in CS&E, especially among those responsible for designing, programming, test- ing, and maintaining the software systems on which the information age depends, are enormous, especially given the speed with which the field changes. These needs are often recognized by all potential participants in the continuing education endeavor, but for various reasons not fully understood by the committee, continuing education is often relegated to the backwaters of universities and neglected by industry and commerce. A study on this subject would document the magnitude of the need for continuing education and explore mechanisms to encourage industry and academia to pay more attention to continuing educa- tion. Such a study would also focus on the needs of industry and commerce for the continuing education of those already in their work forces, and would speak to a continuing education issue different from the one underlying Recommendation 5. Precollege CS&E education. In considering the state of un- dergraduate CS&E education, the committee was struck by the large extent to which incoming students have some computer experience. Acquired in high school classes or in avocational pursuits, such fa- miliarity has both a positive and a negative impact. The positive impact is that these individuals arrive with some of the basic vocabu- lary for and a certain intimacy with computer hardware. The nega- tive impact is that these individuals often have misconceptions about the nature of the intellectual discipline, imagining, for example, that programming (all too often, even bad programming) is identical to computer science. Moreover, these individuals tend to be overwhelm- ingly white and male, a fact that works against the recruitment of women and minorities into the field. A study on the subject of precollege CS&E education would ad

OCR for page 139
RECOMMENDATIONS 157 dress both pedagogical issues (e.g., what are the essentials of CS&E that should be presented at the precollege level?), teacher training issues (e.g., how are those who teach CS&E at the precollege level to be prepared to make appropriate presentations?), and recruitment issues (e.g., how can more interest in computing be generated among women and minorities at the precollege level?. Meshing precollege education with undergraduate CS&E education would be an impor- tant task of such a study. CONCLUSIONS Over the past 50 years, CS&E has blossomed into a new intellec- tual discipline with broad principles and substantial technical depth. By embracing the computing challenges that arise in many specific problem domains, computer scientists and engineers can build on this legacy, guiding arid shaping the course of the information resro- lution. This expansive view of CS&E will require a commensurately broader educational agenda for academic CS&E, as well as under- graduate education of higher quality. Adequate funding from the federal government and greater interactions between academia and industry arid commerce will help immeasurably to promote the broad- enir~g and strengthening of the discipline. (Table 5.1 recapitulates TABLE 5.1 Relating Recommendations to Priorities Established in This Report Priority 1. Sustain 2. Broaden 3. Improve the CS&E the Field Undergraduate Recommendation Core Education 1. Support full funding for HPCC Program 2. Augment HPCC to include more interdisciplinary and applications- oriented CS&E research 3. Academia to embrace interdisciplinary and applications-oriented CS&E research 4. Support CS&E as laboratory discipline 5. Expand HPCC to provide continuing CS&E education for certain faculty and thus improve undergraduate education 6. Academic CS&E to change graduate education to accommodate interdisciplinary and applications-oriented CS&E research 7. Reach out to groups underrepresented in CS&E education X X X X X ~X X X X

OCR for page 139
158 COMPUTING THE FUTURE the relationship of these recommendations to the overall priorities discussed at the beginning of this section.) If the major thrusts of this report sustaining the CS&E core at currently planned levels, broadening the CS&E discipline, and upgrading undergraduate CS&E education to reflect the best of current knowledge- are widely ac- cepted in the academic CS&E community, the community-as well as government, industry, and commerce will be well positioned to meet the intellectual challenges of the future and to make substantial and identifiable contributions to the national well-being and interest. NOTES 1 The reader will notice that the committee has not laid out a set of topical re- search priorities. As noted in Chapters 1 and 6, CS&E changes with extraordinary speed; moreover, its subdisciplines are highly synergistic, as progress in one subdisci- pline may have profound effects on another. Thus recommendations that favor one subdiscipline over another could divert the field and funding agencies from opportu- nities that could well emerge in the future. The recommendations of the committee are structured to emphasize flexibility and to hedge against developments that are not now foreseen. A historical precedent is worth some mention. In 1966, the Automatic Language Processing Advisory Committee of the National Research Council issued a report, Languages and Machines: Computers in Translations and Linguistics, that was widely viewed as a highly influential study in the machine translation of foreign languages. By concluding that the basic technology for machine translation had not been devel- oped at that time (and by implication that work on machine translation was not likely to be immediately fruitful), the report contributed to a subsequent and substantial decline in funding for such research. Supporters of machine translation argue that such a decline was inappropriate and that the current state of the art would otherwise be much more advanced, as it is in Japan, where support in the last two decades has been greater. For more discussion, see Office of Japan Affairs and Computer Science and Technology Board, National Research Council, Japanese to English Machine Transla- tion: Report of a Symposium, National Academy Press, Washington, D.C., 1990, pp. 3-4. 2. This percentage is calculated on the basis of $14.3 billion in basic research and $59.3 billion in applied research proposed in the president's budget request for FY 1993. These levels in the budget request represent a growth in basic research of 8 percent and in applied research of 3 percent over the FY 1992 levels. If these growth rates are maintained, a total of some $312 billion will be allocated to research. (For this calculation, then-year dollars were used. The $3.7 billion is the total projected "new moneys' for FY 1993 to FY 1996 from Table 1.2 in Chapter 1 plus the baseline funding from FY 1991 of $489 million in each of these four years.) 3. The size of the proposed effort ($100 million) was estimated on the following basis. According to data provided by the NRC's Office of Scientific and Engineering Personnel, there were 3860 academic CS&E researchers in 1989. This figure suggests that in 1991, there would have been about 4500 researchers (assuming that one-third of the Ph.D. production since 1989 went into academic research). The CISE Directorate at the National Science Foundation received about 1200 proposals in FY 1990, of which about 300 received funding. NSF program officials state informally that about half of all proposals deemed scientifically meritorious do not receive funding due to budget

OCR for page 139
RECOMMENDATIONS 159 limitations. Thus it appears reasonable to suggest that perhaps 500 CS&E researchers might be available as co-principal investigators for interdisciplinary or applications- oriented work. Of these, the committee assumed that about half would be both will- ing and able to pursue such work. Thus 250 awards per year could be made to teams consisting of two principal investigators, one from CS&E and the other from outside CS&E. Assuming $200,000 per award per year in direct costs, and a total cost (includ- ing overhead) of perhaps $400,000, an additional $100 million would be needed. This sum might pay for a modest equipment purchase, summer research salaries for- the principal investigators, a couple of graduate students, and a computer scientist or engineer investigating some applications-oriented problem or a researcher from the applications domain investigating potentially relevant CS&E. 4. U.S. Department of Commerce, National Computer Systems Laboratory: Annual Report 1990, NISTR 4492, National Institute of Standards and Technology, Washington, D.C., 1990, p. 21. 5. In this context, note that the federal government explicitly acknowledges a re- sponsibility to "participate with the private sector in pre-competitive research on ge- neric, enabling technologies that have the potential to contribute to a broad range of government and commercial applications. In many cases, . . . technical uncertainties are not sufficiently reduced to permit assessment of full commercial potential." (Pre- competitive research is defined as research that "occurs prior to the development of application-specific commercial prototypes.") See Office of Science and Technology Policy, U.S. Technology Policy, Executive Office of the President, Washington, D.C., September 26, 1990, p. 5. 6. For purposes of this discussion, the term "industry" refers to the computer industry. The term "commerce" refers to commercial (or nonspecialized governmen- tal) users of computers, especially those with information-processing needs in large volume. 7. This recommendation is consistent with an ACM recommendation that "institu- tions should encourage more faculty in the discipline of computing to engage with business people in the design of commercial applications, especially those that will give contact with industry thinking on long-term issues. Institutions should encour- age more computing researchers to embrace computational science by joining in projects with physical scientists, bringing their expertise in algorithms and architectures." See Association for Computing Machinery, "The Scope and Directions of Computer Sci- ence: Computing, Applications, and Computational Science," Communications of the ACM, Volume 34(10), October 1991, p. 131. (This paper uses the term "computing" as this report uses "computer science and engineering.") 8. Work on technology transfer is not envisioned as consulting activities that con- sist primarily of giving advice. Rather, this work should usually involve sustained and intimate interaction between academic computer scientists and engineers and those in working in nonresearch activities in industry and commerce. While it would be most desirable if the computing aspects of the problem were novel, such activity would in any event enhance the social and economic impact of CS&E research. 9. A forthcoming CSTB report will address in detail the issues faced by experimen- tal computer scientists and engineers in academia. Among these issues are the long time that system building requires relative to publishing papers (and thus the lower volume of papers), the tendency for many system builders to present their work in conferences rather than in archival journals, and the evaluation of systems in an aca- demic context. 10. Course leaders (i.e., participating researchers) are assumed to spend one full- time month in intensive discussion workshops on undergraduate education, one month

OCR for page 139
160 COMPUTING THE FUTURE preparing a short course, and four months teaching four short courses over a four-year period. Assuming that about $100,000 would be needed per person per full-time year in summer salary, travel, lodging, and so on, the efforts of course leaders would cost about $5 million in direct costs over four years, or about $8 million including a 66 percent overhead rate. Grant augmentation is calculated on the basis of 25 researchers every year receiving an additional $70,000 in research funding (not including over- head), or $12 million over four years. Over a four-year period, a program for 6000 individuals (including a large fraction of the 5000 or so CS&E faculty in the nation) attending a four-week short course might cost $2000 per attendee (likely not to cover the entire cost of the course), for direct costs of $12 million and an additional $8 million in overhead. 11. This sentiment was expressed at the recent CSTB Workshop on Human Resourc- es in CS&E, on which a report will be forthcoming in the summer of 1992. 12. Inadequate laboratory infrastructure for CS&E was noted as a problem in 1989 by the National Science Foundation. See National Science Foundation, Report on the NSF Disciplinary Workshops in Undergraduate Education, NSF, Washington, D.C., April 1989, p. 39.