Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
i] Panel Findings and Recommendations The vision of a more productive research enterprise, and the recognition of difficulties in reaching that vision, prompted this report. In preceding sections, we reviewed many aspects of the use of information technology n research. In this section we present our findings and recommendations. FINDINGS The Panel finds that 1. Information technology has already had a significant and widespread impact on the conduct of research. For the future, that impact amounts to a revolution. Computer and communication technologies are valuable to every scientific discipline and essential to a growing number of them. The Panel examined the uses of information technology in ten science and engineering disciplines. Although these fields use information technology in different ways, the Panel also found many similarities. These can be summarized as follows: · As the power and speed of computers have increased, numerical computa- tions of increasing complexity have become practical. The result has been more realistic simulation of systems either physically impossible or too costly to study directly. · Another consequence of increased computational power has been the capability to collect, store, retrieve, manipulate, and analyze enormous quantities of information. Electronic storage of information means that huge information archives for example, the Human Genome Project- are not only feasible but also accessible to more researchers from different disciplines. · An additional consequence of increased computational power has been the capability to present the results of numerical computations as visual images. The 47
48 [NFORAiATION remarkable ability of the human brain to recognize patterns in pictures allows TECHNOLOGY AND faster understanding of computational results and speedier, more efficient THE CONDUCT interaction with models when numbers are turned into visual images. OF RESEAnCH ~ Information technology allows computers to take over monitoring and control of scientific instruments. This in turn makes scientific observation more convenient, more reliable, and often lower in cost; in some cases, it has led to new computer-based instruments that extend the bounds of observation. · Information technology has greatly expanded the capabilities for communi cation among researchers. Communications systems mediated by computers have led to the rapid and relatively inexpensive exchange of everything from memoranda to massive data files. The Panel further finds that: 2. Significant impediments to the widespread use of information technology in research require careful attention. Some impediments are technical and finan cial. Other impediments, which up to now have received the least attention, are behavioral and institutional. · Technical impediments are serious in a few fields. Fields doing large-scale experiments, using satellites to gather data, and using graphics to analyze large amounts of data will for the foreseeable future need computers, software, and networks that are bigger, faster, more capable, and more efficient. The needs of these fields deserve, and will continue to receive, attention from computer scientists and engineers, from information technology manufacturers, and from institutions such as the federal government and universities. · Financial impediments are chronic. Despite the decreasing cost of hardware, no sources will ever supply enough money to provide every researcher the best information technology environment. The institutions that fund researchers will continue to do their best, and information technology will continue to need more funding. While the Panel cannot suggest detailed means for increasing the total resources devoted to information technology, it does feel strongly that the provision of such resources is critical to progress in American clinical, engineer ing, and scientific research. In the resource-constrained environment likely to confront research in the future, difficult decisions will be made on reallocating necessary resources. With a declining population pool from which to draw new scientists, clinicians, and engineers, and with the increasing complexity of research, the Panel believes that increased support of information technology in research deserves high priority. If the nation's best researchers are not given the appropriate tools, our role in the international scientific community will con tinue to diminish in importance. The limited resources of federal agencies must continue to be allocated on the basis of scientific merit and significance of the proposed research activities.
49 · The Panel believes that serious impediments to increasing the use of information technology in research are behavioral and institutional. These can be sorted into four major categories: problems of access, problems with learning and use, attitudes of individual researchers, and problems of management. · Problems of access. Computer networks, hardware, and software are not necessarily accessible to the researchers who want them, nor can everyone who wants them afford them. Network access is still limited and inconvenient. The limitation is especially important since network access permits both collaboration with distant colleagues and access to computing resources. For many researchers in a large number of fields, the hardware to which they have access is adequate. Some researchers, however, require large amounts of time on specialized, expensive hardware; supercomputers are a prime example. Access to the national supercom- puting centers will increasingly be made possible through computer networks. As the availability of supercomputers leads researchers to reconsider scientific prob- lems, one can anticipate increased demands for access and network use; researchers in the future may find time on supercomputers limited and networks congested. Finding the right software is a more serious problem. Software that is commercially available is often unsuited to the specialized needs of the re- searcher, and professionals who create software for research groups are rare. Consequently, many researchers, who are usually not skilled software creators, develop their own. Too often, members of the research community waste time, effort, and money duplicating one another's efforts. What appears to be lacking are institutional means for providing the services of skilled professionals to create and maintain appropriate software. Funding sources do not consider support for such professionals essential; institutions give them neither recognition nor career status. Also needed is an institutional system to collect, review, document, and disseminate scientific and engineering application software. Prototypes exist but need broader evaluation. Such a system could evaluate software's effectiveness and eventually lead to standardization of software for wider use. · Problems with learning and use. Learning to use information technology requires a large investment in time and effort before the investment pays off, and help is hard to find. Researchers who need information technology face difficult choices: they must either learn to use whatever hardware and software the market offers, create their own, or do without. When researchers learn to use existing information technology, they receive haphazard help in learning; in- struction or specialists in information technology are often unavailable. Neither the researchers' disciplines nor their institutions provide incentives for learning. Furthermore, researchers must invest time at the expense of their research productivity during the learning period. Once researchers learn the necessary information technology, they face problems in using it. Networks pose a significant problem. Most of the approxi- mately 100 research-oriented computer networks in the United States were established to serve the needs of small and widely scattered communities of PANEL FINDINGS AND RECOMMENDATIONS
50 INFORMATION researchers. As a result, a researcher on one network wanting to communicate TECHNOLOGY AND with a researcher on another faces problems of compatibility. The technical THE CONDUCT aspects of these problems are tractable; the institutional and behavioral aspects OF RESEARCH are less so. No one agrees on how procedures for using the networks might be standardized. The networks are not well coordinated with one another, and users have limited opportunity to suggest improvements. · Problems of the attitudes of individual researchers are twofold. For reasons enumerated above, researchers often approach new information technology applications cautiously. When senior researchers, who are involved in peer review and decisions about publications and research proposals, are resistant, their attitudes can lead to the rejection of innovative applications of information technology to research. Another problem is that of proprietary attitudes toward research data in many disciplines. So long as primary data are viewed as the exclusive property of the researcher who collects them, they will not be available to other researchers. Even if data are made available, they will be left in idiosyncratic formats with insufficient explanatory documentation; and the effort required to make these data usable to the research community will not receive high priority. Increased access to data does, however, raise issues of how large volumes of primary data should be stored, and of the need for validating stored information. · Problems of managing information technology. Some basic questions need to be addressed: who is to manage, maintain, and update information services? Who will create standards? How will costs be charged, and who will pay for them? No current institutional framework provides the ideal answer to these questions. Federal agencies, professional societies and scientific associations, and private profit-making groups need to consider how to address the needs of research users of information technology. RECOMMENDATIONS RECOMMENDATION I The institutions supporting the nation's researchers must recognize and meet their responsibilities to develop and support policies, services, and standards that help researchers use information technology more widely and productively. Specifically, we recommend that · Universities provide accessible, expert help in learning and using informa tion technology. · University departments, and scientific and professional groups, establish career ladders for scientific programming positions. · Funding agencies provide support for scientific programming and for help services in learning and using information technology systems for research.
51 · Scientific associations establish disciplinary standards for the storage and indexing of scientific data. · University departments, and scientific and professional groups, implement mechanisms for the evaluation, merit (peer) review, and dissemination of software useful in the conduct of research. · Vendors, in collaboration with scientific groups, establish standards for simplified and consistent user-machine interfaces. · Network administrators provide simple user interfaces and addressing schemes, add gateways to other networks, improve system reliability and capacity, and provide online help, such as guides to services and mail addresses of individuals who can answer questions. · Information service providers create simplified common standards for accessing and querying information sources, and eventually provide unified access to information. · Software vendors, and scientific and professional groups, create program libraries and make them accessible through the networks. Rationale. Information technology is now becoming an essential component of the research environment. The services needed by research users include · Access to computers; · Access to networks, both local and wide-area; Long-term storage of and access to data; Hardware maintenance and augmentation; Help in learning to use existing software and services; · Production of new software and customization of existing software; and · Collection, review, documentation, and dissemination of software. In some instances these services may be efficiently provided by a central organization; in others, by decentralized groups. The services may be provided either by augmenting the responsibilities of existing groups or by creating new groups. Currently such services are being provided in a variety of ways, with highly variable degrees of success and efficiency, across many laboratories, professional societies, universities, and federal agencies. In the most successful models, the researchers feel their needs are paramount. These models need to be publicized, evaluated, and disseminated so that policy and allocation decisions are well informed by the views of research users of information technology. With regard to policies, the Panel recognized a number of issues, cited earlier, that need further discussion; the appropriate groups may wish to consider the specific recommendations that follow. · The federal government, through policies of its research-supporting agen- cies, should ensure proper support for software development for scientific research. Software developed should meet minimal standards of compatibility, reliability, and documentation, and should be made available to other research ers. PANEL FINDINGS AND RECOMMENDATIONS
52 INFORMATION · As a general principle, data collected with government support rightfully TECHNOLOGY AND belongs in the public domain, although the right of researchers to reasonable THE CONDUCT time for first publication must be respected. Federal agencies, scientific societies OF RESEARCH and professional associations, university consortia, and other private groups may wish to make specific recommendations based on their reexamination of the implications of the adoption of such a policy. Such recommendations might include the creation and maintenance of data banks and indices to data by research communities, the potential of evaluated databases, and the possibility of including reanalyses of data as part of peer review of publications. · There is a pressing need for new and more compact forms of data storage. One particular area in which these techniques would be useful is that of image compression. One solution seems to lie in optical disk storage. Unfortunately, these new techniques are still immature and lack commonly accepted standards. Federal agencies should encourage engineering research on optical storage techniques for scientific purposes. · Efforts to create toolbuilding and research resources for nondefense software should be encouraged. · The federal government should fund pilot efforts of two kinds. One would implement information storage and software dissemination concepts for selected disciplines. The other would implement the concept of software markets, with emphasis on the development of generic tools useful in several or many disciplines. · In addition, these pilot projects should be coupled with exploration of such policy issues as protection of confidentiality of information about human subjects, protection of intellectual property, and information security concerns in a global electronic information environment. RECOMMENDATION II The institutions supporting the nation's researchers, led by the federal government, should develop an interconnected national information technology network for use by all qualified researchers. Specifically, we recommend that · The Office of Science and Technology Policy (OSTP) in the Executive Office of the President and the federal agencies responsible for supporting and performing research and development plan and fund a nationwide infrastructure for computer-based research communication. · Planning and development of this nationwide infrastructure be guided by users of information technology in research, rather than by technical experts in information technology or hardware or software vendors. The Panel believes strongly that such a national network is too important to the future of research to be left only to the technical experts.
53 · The national research network be founded on the fundamental premise of open access to all qualified researchers/scholars that has nurtured the world's scientific community for centuries. · The national research network be developed in an evolutionary manner, making full use of the existing successful networks for research. Rationale. The Panel views the federal government's role in developing a national network for research as analogous to its role in developing the nation's network of roads, streets, and highways. Here, the federal government has planned and funded the interstate highway system and the national highways, and has imposed or encouraged certain national standards. State and local governments have planned and funded a network of highways, roads, and streets that is fully interconnected and compatible with the federal framework. In the national research network, analogs would be research institutions, scientific and professional associations, and corporate groups operating individually or banded together in consortia. Funding for this undertaking, as outlined in the report of the Federal Coordinating Council on Science, Engineering, and Technology (FCCSET) Committee issued recently (Office of Science and Technology Policy, 1987) should be made available to ensure an advanced national network infrastructure and sentences for the nation's research communities. Appropriate division of responsibilities among federal and state agencies and research institutions warrants careful attention. The Panel notes that the Director of the National Science Foundation has announced the Foundation's intention to solve as lead agency in developing a national network. The Panel strongly supports the concept of a lead agency, believing that leadership in coordination of support for a national research network is an essential element in the nation's science policy. The Panel believes the National Science Foundation would be an appropriate lead agency, given its legislatively mandated responsibility for supporting research and education across the full range of science and engineering disciplines. The creation of a national network will take a considerable amount of time. In the meantime some valuable and much-used networks such as BIONET, OMNET, BITNET, or HEPNET-exist within particular scientific or academic communities. Many of these networks have been improved with advice from their users, and can be an invaluable source of advice to the designers of any national science network. It is particularly important, therefore, that the existing academic and disciplinary networks continue to receive support until such time as they can either be integrated into or supplanted by the national network. The Panel's recommendation for a national research network is similar in substance and spirit to the more detailed recommendations contained in a recent report of the National Research Network Review Committee of the NRC's Computer Science and Technology Board (National Research Council, 1988). Crucial to setting up and running a national network infrastructure is partic- ipation by users. The Panel urges that agencies work with an advisory board PANEL FINDINGS AND RECOMMENDATIONS
54 ABORTION largely composed of users. A research network of national scope must be TECHNOLOGY AND oriented toward the research user of information technology. Its philosophy and THE CONDUCT structure should be such that learning, entering, using, and leaving the network OF RESEARCH are simple. This should involve, among other things, an easily intelprete program for helping the user, instructions for different levels of use, and simple connections between the network and many varieties of terminals, personal computers, and workstations. The network should be capable of transmitting graphics, symbols, and large amounts of data quickly. It should have gateways to networks in other countries. It should be supported by a professional staffwhose main task is to help users. The Panel recognizes that constraints on access to information are sometimes warranted in cases where the privacy of personal information, the protection of human subjects or of intellectual property, or national security concerns are of overriding importance. Nevertheless, the Panel believes that the interests of the global research community are best served by establishing open and unfettered access as a fundamental presumption in the operation of a national research network. RECOMMENDATION III To facilitate implementation of Recommendations I and II, and to focus continuing attention on the opportunities and impediments associated with research uses of information technology, the Panel recommends the establish ment at a national level of a user's group to oversee and advise on the evolution and use of information technology in support of scientific, engineering, and clinical research. Specifically, the National Research Council (NRC) should charge a standing committee or board (whether existing or newly created) with the mandate to oversee and advise on research use of information technology. The membership of this board should include a majority of users from a variety of research disciplines. Rationale. The problems and needed changes addressed by Recommenda tions I and II are diverse and do not have short-te~ solutions, and, therefore, require some institutional setting for ongoing examination and discussion. Leadership and coordination in the application of information technology in research would be provided best by a single organization. Many organizations currently promote developments in information technol ogy at a national level. Among these are: the National Science Foundation and its Computer and Information Science and Engineering (CISE) Directorate and Division of Advanced Scientific Computing, especially through the NSFNET initiative, EXPRES program, and the Panel on Graphics, Image Processing, and Workstations; the National Research Council, especially its Computer Science and Technology Board, Numerical Data Advisory Board, Committee on National
55 Statistics, and Board on Telecommunications and Computer Applications; a variety of initiatives within the National Library of Medicine; and the Federal Coordinating Council on Science, Engineering, and Technology (FCCSET) of the Office of Science and Technology Policy (especially its committees on Networking and on High Performance Computing). These activities, however, are fragmented, specialized, and tend to give little attention to the behavioral aspects of research uses of information technology. The Panel recommends location of the user's group within the National Research Council (NRC) rather than a federal agency because it believes that the group ought to be free to focus on the interests and concerns of users, unconstrained by immediate concerns for the distribution of funds among disciplines or agencies. Several boards or committees now existing at the NRC have titles related to the charge to be given this new board. However, it is the Panel's view that none of the existing groups represents adequately the needs of users. Whether one of the existing groups might be reconstituted to create a body of the required nature or whether a new body might be founded is, in the view of the Panel, far less important than the nature of the resulting body. What is needed is a group of researchers who use information technology, supplemented by a few expert providers of information technology. The National Research Council unit Whether existing or new) would have several specific functions. One function would be to advise policymakers on a broad range of issues related to research uses of information technology. Some of these issues are: international implications of networking; national security concerns associated with scientific databases; and issues of cost for network communications. Ad hoc study panels would be established as appropriate. Another important function of the unit would be to collect, analyze, and disseminate data on how researchers use information technology. In the course of its deliberations the Panel became painfully aware that these data presently do not exist. Such data would inform not only the actions of those responsible for the support of research but would also apprise researchers themselves of new opportunities offered by information technology. A model for such activities among present National Research Council operating groups might be the Office of Scientific and Engineering Personnel. This mechanism would also provide a forum to facilitate the transfer of technology: by sponsoring workshops for scientists on newly developed infor- mation technology and on coordination of approaches to simplified standards; by exchanging information with technology developers; and by coordinating interaction among scientific organizations and professional associations. The board would disseminate information on current uses of information technology and new developments. The unit would also convene meetings between researchers and those respon- sible for supporting research. By providing a forum for discussion, it would ensure that the needs of the research community are brought to the attention of appropriate officials and administrators. PANEL FINDINGS AND RECOMMENDATIONS
