Here's what we have on this site

Online Publications
Publications List
JD's Background
Feedback to JD
Search this site

Copyright 1997, JD Eveland. All rights reserved.



JD Eveland

Working Paper Prepared for the National Science Foundation, Division of Industrial Science and Technological Innovation, July 1986


The purpose of this paper is to sketch some dimensions of what is currently known and not known about the transfer of scientific and technical information between academic institutions and industries, with particular emphasis on the implications of this transfer for U.S. industrial competitiveness. In recent years there have been a number of studies bearing on these issues, none comprehensive but each contributing to the picture. Generalizing from this body of work to either practice or policy conclusions is difficult, however, because of the sensitivity of the analysis to context, setting, and technology variations; conclusions valid at one time or in one sector may be only marginally valid in others.

We begin this paper with a description of the context of scientific and technical information (STI) transfer, and some of the underlying issues that need to be considered in all applications. An overview of STI systems and methods serves as an introduction to a review of the effectiveness of different mechanisms and venues for STI exchange. A brief discussion of the international dimensions of the STI exchange and utilization problem and of disciplinary and professional differences rounds out the discussion. We conclude with some general observations on where STI appears to fit into the general issue of industrial competitiveness, and some suggestions for implementing some changes in existing patterns and systems..




Information, most generically defined, is anything that reduces uncertainty about cause-and-effect relationships. Scientific and technical information is that body of information bearing on real world relationships; it should be understood as encompassing social as well as physical science knowledge, although physical science and engineering referents are probably more commonly intended. "Use" simply refers to someone's actually using an item of information to reduce uncertainty in some practical or theoretical context of interest. These contexts cover a wide range of phenomena, which often have very different meanings for participants depending on their needs. It is helpful to distinguish (following Yin and Moore, 1985) four general (and nonexclusive) types of use of STI:

bulletResearch stimulation: the use of STI to provoke further research, to define new research questions, to reconcile previously unexplainable phenomena, or to suggest a new range of applications for particular tools and techniques;
bulletPractical application: use of STI to accomplish some particular outcome state, such as creating a new product or process;
bulletDecision making: use of STI to set general policies for practice, either in industry, government, or (more rarely) academia;
bulletEnlightenment: use of STI to raise the general awareness of a set of opportunities or problems posed by science and technology.

Information exchange can take place within many different settings. The simplest and most classical is that of the "user" (defined in terms of any of the above purposes) perusing the open professional literature in search of new ideas. The user may also undertake a deliberate search for information in the public or semipublic record relating to a particular more or less defined need. The effectiveness of the search process is highly dependent on the efficiency of the search technology employed, on the ability of the searcher to understand the dimensions of what s/he is searching for, and on the quality of the underlying information base. In recent years, there has been considerable advance in making STI available in machine processible forms such as online data retrieval formats, but deliberate search is still very difficult and constrained.

In addition to consultation of the literature base in a given field, there are also a number of channels for STI exchange that involve interpersonal interactions. These range from consultancies, or one-to-one contractual relationships aimed at specific results, to retainers (long term consultancies), to professional meetings at which ideas circulate for the taking, to internships and other forms of bringing in entry level people. A relatively new and interesting interpersonal approach is the university/industry cooperative center (discussed in more detail below), within which a wide range of formal and informal exchanges can take place.

Several different models have been proposed to describe the dynamics of STI exchange that is, the reasons for it and its consequences (see Yin and Moore (1985) for a good background on this issue). One model is described as the knowledge-driven or technology-push approach. In this formulation, knowledge originates at some defined location and is gradually disseminated among a population of potential users, some of whom may discover a need for it as a result of the exposure. Most classical diffusion models, including agricultural extension, are based on this premise. Another approach is the problem solving or demand-pull model; in this approach, needs have priority, and stimulate search behavior. Under this model, the knowledge base itself forms the critical limit on applications, rather than the imagination of the users.

Both these formulations are described in Schoen's (1971) terms as center-periphery systems. Another formulation is the social interaction approach, which posits a sort of evolutionary convergence of knowledge and problems through interactions among people. Problems and solutions encounter each other, and it does not matter much whether a need precedes ideas about solutions or vice versa, since they will both be changed over time anyway. The properties of these "network-based" or "peer diffusion" systems are still unclear. It does appear that in certain of the higher-technology fields where knowledge is being generated rapidly in many different settings, they are more efficient than center-periphery designs (Rogers and Leonard Barton, 1980). It is clear that they more accurately map the dynamics of technology implementation than does either kind of unidirectional model (Eveland, 1986).


Numerous possible dimensions for analytically disaggregating STI exchange systems in terms of what they exchange and how they exchange it have been proposed in the literature; no one schema is universally useful, but most highlight some useful distinctions. Here, we will discuss three dimensions for categorizing exchange systems: active/passive, formal/informal, direct/indirect (see Figure 1). Note that these categorizations for the ways in which transfer systems operate are more or less independent of the underlying models of utilization discussed above.

Active/Passive: This dimension of the structure of exchange systems refers to the modes of communication involved. Active systems have some kinds of "transfer agents" whose job it is to take information from one place and move it to another, often face-to face; the classic example is the agricultural extension agent. Passive systems, by contrast, simply array information for the taking, relying on the initiative of the "user" to search out the portion of the information that s/he may need. Online data bases such as NTIS or Dialog are good examples of passive systems.

Formal/informal: This dimension reflects the channels of communication. Formal systems are those established explicitly to transfer information; informal systems are those that transfer information while serving some other formal purpose. The distinction is mirrored in the roles assumed by the parties involved. In formal systems (such as agricultural extension), the roles of transferor and recipient are clearly specified and understood; in informal systems (such as a cocktail party at a professional society meeting) roles are not clearly defined, and may shift rapidly.

Direct/Indirect: This dimension reflects the relationships of the participants and the distance between them. Direct systems share information more or less immediately from producer to user; indirect systems generally transfer the information to some intermediate point or points, perhaps written, perhaps interpersonal, often with intervening analytical stages. This dimension is reflected, for example, in the difference between reading a journal article written directly by a researcher and reading a "state of the art" review article summarizing and commenting on a body of research.

It is important to remember that these functional differentiations are not mirrored explicitly in institutional distinctions. That is, a given institution such as a University/ Industry Cooperative Research Center or a professional society meeting can and does play both active and passive, both formal and informal, roles at various times and through various people and substructures. Some structures are not so flexible; it is hard to make an online data base behave as an active or informal medium. In general, institutions that can behave flexibly have a definite advantage over those that have only one mode of operation.

It is also worth noting that each kind of system is valuable and useful at particular times and in particular ways. Formal systems, for example, are particularly good at exchanging highly compressed information in a short period of time; informal systems, by contrast, are very inefficient at this, but are very good at providing contextual and interpretive information needed to make sense of formal findings. Active systems move a lot of technology, but are extremely expensive to operate; passive systems have lower "hit rates", but are generally relatively cheap. An overall "STI Exchange System" must incorporate elements of each dimension to be maximally effective.

Cutting across all of these dimensions is the question of the technology of transfer that is, the tools, methods, and techniques employed in any of these kinds of systems. To the traditional channels of print media, oral presentation, face-to face contact, telephone, and the like, have been added an enormous variety of electronically augmented information storage, retrieval, analysis, and communication technologies: online data bases, computer conferencing, electronic mail, teleconferencing, expert systems to expedite search processes, etc. (Rice and Associates, 1984). Clearly the systems available for information exchange have become vastly more facilitative with these technological advances.

But the critical point remains what it always has been namely, the quality of the information and its utility in the particular situation. Online data retrieval is paced by the ability of the searcher to bound his/her problem; computer conferencing is no better than the quality of the participants and the amount of time they can afford to devote to the exchange. The "new media" perhaps can be seen as "multipliers", changing the power and magnitude of the exchanges they facilitate rather than their basic nature. Over time, advances in quantity tend to become advances in kind, and we may eventually see entirely new dimensions of exchange introduced. At present, however, the technology of transfer remains largely an adjunct to, not a determinant of, the exchange relationship.


A number of key concerns will underlie the remainder of this discussion of STI exchange relationships. As a basis for our subsequent discussion, let us sketch each briefly.

The "Producer-User" Distinction: In much of the STI utilization literature, there is an implicit or explicit distinction between those who generate scientific and technical knowledge and those who put it into practice. While the distinction is useful in some respects, it is particularly important to remember that it is a distinction between roles in a conceptual system and only incidentally a distinction between people. That is, most participants in STI exchange systems function part of the time as "producers" and part of the time as "users", in a reciprocal and recursive sequence of knowledge transfer (Havelock and Eveland, 1984).

The rhetoric of STI can often leave the impression that universities generate knowledge that is then "applied" in industry, and that the sequence from basic research, to applied research, to development, to production, is linear and monotonic. In fact, there are substantial feedback loops, large numbers of dead ends and backtracking, and a great deal of evolutionary redefining of "earlier" steps based on later discoveries. In our subsequent treatment of STI exchange, we will frequently use the terms "producer" and "user", but always within this context of shifting and reciprocally defined roles.

Disciplinary focus and the role of vocabulary: Despite the recent expansion of interdisciplinary centers, it remains a fact that most academic research takes place within the bounds of more or less traditional academic disciplines. Disciplines are most essentially sets of shared vocabularies for describing real world phenomena and the conceptual structures that define and link them. The sociologist, the psychologist, the political scientist, and the economist may construe the same social behavior in radically different terms. So too the physical chemist, the biochemist, the chemical engineer, and the structural physicist may all describe essentially the same physical phenomenon but are likely to do so in significantly different words that may be less than wholly mutually comprehensible.

The boundaries of discipline cut across research structures more than is often recognized (Friedman and Friedman, 1985). Chemical engineers in academia and industry tend to have more in common in terms of things to talk about and ways to talk about them than either has with, say, an analytical chemist in their own organization. It is true that research/development problems in industry tend to be more interdisciplinary in focus, requiring more points of view for effective resolution, than do research problems of the kind favored by academic people. This does not mean that industry is necessarily better at dealing with interdisciplinary problems, or that industrial personnel are more at ease crossing disciplinary lines than are academics. Indeed, the disciplinary focus problem is frequently cited as one of the main barriers to effective application of research findings in industry (Fusfeld and Haklisch, 1986). We will consider later some of the implications of this problem for STI exchange and ways it can be and has been addressed; for now, we will simply note that it is a problem that is likely to grow ever more serious as knowledge becomes more specialized and thus more fragmented in terms of any one person's or group's attention capacities.

The "validity" of research results: One of the problems frequently cited as a barrier to STI exchange is the difficulty of establishing the domain of validity of findings. That is, when a particular finding is published about, say a particular chemical reaction, it is by no means established completely over what range of conditions the finding applies, whether there are problems of scale in the reaction, whether there are side effects to consider. and a host of other considerations necessary to effective application of the reaction in an industrial process. Moreover, it is not even established that the finding is "true" just because it has been published. Scientific and technical information, even more than other information, is always tentative, always subject to revision, extension, restriction, or amendment.

Rather than think of this problem as one of "validation" or establishing "truth", it is more helpful to think of it as one of establishing a range of utility for findings. Utility is a slippery concept, dependent as it is on changeable and time dependent criteria and subject to different evaluations by different people against those various criteria. It is also difficult to establish, depending as it does on varying interpretations of the same evidence against different criteria, However, it is essentially what matters. Only through rather extensive (and often expensive) research is the range of utility usually established. Unwillingness to participate in that range establishing research is the most general reason for late participation in technological change.

An interesting sub-dimension to the "validity" problem is that of access to and utility of negative findings the information that a particular process does not work, or a particular expected cause-and-effect relationship does not hold in general or in a particular case. Anyone who has ever tried to publish a research report knows that it is almost impossible to get journals to entertain seriously a report of negative findings. Where such findings get into print, it is usually in the context of something that one did find, even though the positive result may be both theoretically and practically less interesting than the negative finding. Informal contact, usually in person, often at professional meetings, appears to be the only effective channel for transmission of this information. Industrial people are particularly sensitive to this issue, and frequently report that access to negative findings through personal contact is one of the major benefits of UICRC participation; one good negative finding can save a firm considerable amounts of investment in following up what turn out to be deadend applications (Hetzner and Eveland, 1986).

Structural/organizational vs. technical barriers: Up to this point we have been concerned primarily with technical issues those related to the inherent capacity of people to absorb, understand, and apply STI. But it is absolutely vital to remember that all STI exchange takes place within the context of social/organizational systems. That is, the application and use of information is no more effective than the capacity of those human systems to organize, process, and relate the data to their own complex sets of criteria, only some of which relate to the technical rationality of the system. Social, professional, economic, and political criteria are at least as important to most organizational decision processes as technical/scientific criteria. In recent years, there has been considerable attention to the shortcomings of decision making in American industry, in particular to the tendency to overemphasize short range economic criteria such as rate of return. But the allocation of research effort within universities, and within government, is equally affected by political and administrative factors having little to do with the inherent scientific quality of the issues.

It would be misleading to say that the use of social/organizational criteria is wrong, or even that it leads to ineffective utilization of STI. Social organization is the only context people have for generating and using science, particularly complex science. What is important is to be explicit about the role of social institutions in STI exchange, to recognize or at least systematically examine the validity of their criteria and assumptions, and in general to allow for the use of science and technology within carefully attended to sociotechnical systems. We will find this perspective critical to our subsequent discussion of these issues.

The problem of "implementability: It is generally acknowledged that the inherent quality of a scientific/technical finding is less important in terms of accounting for the effect it has on society than the way in which it is implemented or put into practice in a sociotechnical system. Yet this perspective has been slow to be applied to the discussion of STI exchange. One appropriate handle would be to focus on the "implementability" of findings. Implementability is a dimension not precisely specified relating to how much sociotechnical effort is required to incorporate a new thing into an existing system. It is generally related to the traditional dimensions describing "adoption potential" (compatibility, trialability, relative advantage, etc..), but interprets these dimensions as they relate to the broader organizational context rather than just the perspective of an individual decision maker. It plays a role in a wide range of technological changes, including advanced manufacturing (Ettlie, 1986) and office systems (Bikson and Eveland, 1985).

Numerous analyses of implementation processes have outlined clearly that:

bulletputting technology into practice is a process of many decisions, usually made over limited domains with limited criteria;
bulletthat it passes through several defined stages stages which can be skipped or steamrollered only at the cost of ultimate failure;
bulletthat different information both scientific/technical and social/organizational is needed at different times by different people;
bulletthat people have a wide variety of criteria, and respond to organizational situations in terms of complex interplays of criteria at different levels; the key issues of the distribution of costs and benefits within organizations the economics and politics of technological change cannot be ignored or stipulated;
bulletthat "diffusion systems" aimed at transferring STI into use contexts must focus on implementability and implementation dynamics if they are to be successful.

Yet somehow we often think that "science is different", that the imperatives of organizational processes can be overridden by "truth", and that we can safely short-circuit implementation when we know that certain things really work. This is simply not true, and all our subsequent discussion will reflect the underlying critical dimension of the necessity of systematic attention to implementability and implementation processes.

The "property" problem: One of the stickiest issues in STI exchange revolves around the fact that in human systems information is not merely a thing that reduces uncertainty about relationships it is also a crucial resource in the political and economic context within which application must, as we have noted, take place. We have evolved an elaborate structure of intellectual property rules to recognize and manage this function of information. But like all social rules, intellectual property rules are manipulable and subject to interpretation against different criteria. How people use and trade the resource of STI is a constantly shifting problem.

The problem is made more difficult by the complexities of implementability and finding a context for application of the knowledge. For example, university/industry centers almost all have provisions for publication delays where center-generated information can be made available first to participants and only later to the general scientific community. Yet these provisions are rarely invoked, and participants almost universally report that what is important for utilization is not so much having the information before others as it is having it in more complete form that is, with the accompanying context that allows implementability to be readily and effectively diagnosed (Gray and Gidley, 1986; Gray et al., 1987). The open literature is very effective in circulating basic information, but is singularly unrich in terms of factors related to implementability.

The advantage of center-type interactions in conveying implementability information is gained particularly by their multichannel access to research projects and personnel. Interactions between university and industry personnel typically span a wide range of topics and use a variety of information exchange channels, even where primary reliance is on formal meetings (Eveland, 1985). Moreover, industrial members usually have access not only to senior researchers but to the junior faculty, postdocs, and students who actually do a lot of the research and have a great deal of the needed information in their heads but who seldom make formal presentations on it. We noted earlier the importance of informal interactions over time in terms of access to negative findings; the same point applies to the range of qualifications, adjustments, interpretations and other amendments and glosses to the basic finding that are needed by anyone who tries to install something. Part of the problem is that it is not until a fair amount of contact has taken place that people really know what to ask. Only contacts repeated over time and across multiple participants can convey sufficiently rich information.

An example may be found in one center that recently succeeded in developing a product later marketed by one of its member companies. Representatives from the company were on site in the center almost weekly for the last six months of the research process to the point where some of the other companies were concerned about the excessive focus of the research. In the course of that contact, they learned many things about how the product actually behaved under a variety of conditions that they could never have deduced from the spare prose of a typical research report and that would have cost them a great deal of money to find out if they had had to duplicate the process in their own corporate labs.

Another dimension to intellectual property is its time dependency. That is, particularly in high technology areas, the half-life of STI is quite short; information that is helpful today is superseded rapidly and loses its value tomorrow. This is particularly true in areas such as semiconductor technology, where patents are generally not applied for; it is more effective to keep the information as a trade secret for the short period of its useful life than it is to disclose information through the patent process that competitors might be able to use to determine the overall direction of the R&D program (Larsen, 1985). In these rapidly advancing sciences, the information that makes its way into the open literature is useful more for archival purposes than for any real scientific advances.

Another interesting problem is the degree to which cooperative research programs such as centers can and do engage in proprietary research with and for particular industrial clients. The core of most centers is a program of basic research which is of more or less interest to all the sponsoring group. Centers differ considerably, however, in their policies related to proprietary work. Some do not accept such commissions at all (although almost all allow individual faculty to accept them, often using center facilities); some have provisions for particular member firms to "augment" the core research program along lines of their particular interest in ways that may be more or less proprietary.

A certain number of center-type programs (although not generally the NSFsponsored UICRC's) have parallel structures to do work for particular clients (Tornatzky, Solomon, and Eveland, 1987). Thus member firms may contribute a set amount to fund the core program which is accessible to all, while also paying for some proprietary work in a different part of the lab. Often the proprietary program may have a different name, although sharing facilities, personnel, and resources with the core program. This latter pattern is particularly common where state funded centers have economic development as a major governing criterion; the Microelectronics Center of North Carolina is a good example (Nelkin and Nelson, 1985), as are several of the programs funded under Ohio's Edison Centers program and Pennsylvania's Ben Franklin Partnership.

Clearly there is a needed interplay of proprietary and non proprietary research and development effort, and this effort does not allocate simply to different institutions such as university and industrial laboratories; the continuum is more complex than that. Institutional experimentation with ways of creating mixed research/development agendas is needed, and is in fact under way (Tornatzky, 1986). Ways of compensating researchers in such a way as to encourage their participation in proprietary ventures while retaining their commitment to the open basic research process are needed. In general, the lines between basic research, applied research, and development are more clear to the analyst than they are to the practitioner. The institutional structures we use to manage the research process will eventually have to reflect this truth.

Intellectual property rules, therefore, are not as critical a factor in STI exchange and utilization as might be supposed. In particular, they probably do not have much effect on the transmittal of information at the cutting edge of new science and engineering. Where they are important is at the point where production or procurement is initiated. The not infrequent practice of the government's underfunding R&D (on which many firms lose money, in hopes of setting up an inside track to production), and then bidding out the technology so developed on the grounds that it owns the intellectual property involved and can transfer it at will, has created considerable ill feeling not to mention a certain unwillingness to play the game, to the overall detriment of creative R&D. The role of STI as a resource must be carefully considered and factored into any measures proposed to address STI exchange.



It is logical to inquire, given the considerable experience with STI exchange and the body of studies that have been applied to it, what we know and do not know about how to make transfer relationships more productive and efficient. Unfortunately, the results are somewhat disappointing. The relationships appear to be so complex and contingent, and so situation-specific, that general principles are hard to come by. However, some broadly applicable ideas can be found in recent literature.

The diffusion of research results from academia to industry has not changed all that much in over 100 years, since the creation of the applied research system particularly in the American land grant universities (Thackray, 1982). Growing out of the German research tradition but applying many ideas from the agricultural extension model, U.S. universities evolved a system with approximately three tiers. At the top, there were the major basic research universities: Johns Hopkins, Harvard, MIT, Stanford, etc. The second tier, of which many also participated in the first, was made up of major universities with substantial applied as well as basic research interests: Illinois, Michigan, California, North Carolina, etc. The third tier was the less sophisticated institutions where cutting-edge research was relatively rare and the emphasis was on filling in holes and developing applications based on the performance of the more central institutions.

This system was a reasonable approximation of a "center periphery" diffusion system, described earlier. Knowledge tended to flow from the universities out to the other participants and eventually into industry through the traditional channels, particularly published articles and presentations at professional meetings. On a more focused level, the favored channel was the individual consultancy, in which a firm rented a professor to get a more or less specific item of information from him. Knowledge also flowed through hiring of academically trained graduates, although it was expected that the most talented would undoubtedly remain in academia with industry absorbing those who could not get tenured slots.

In many respects, this system remains the core of current practice. The lines between the "tiers" have become less pronounced over time, although there are still tremendous differences between the top research universities and the rest of the pack. New media have been introduced into the dissemination process, but have yet to play a determining role.

Some major changes are apparent. For one, the incredibly accelerating pace of knowledge expansion has made it increasingly difficult for single individuals to command the breadth of effective knowledge to serve as key exchange resources even within disciplines, let alone across disciplinary boundaries; it is increasingly necessary to assemble teams of experts rather than single consultants. Also, the increasing constraints on faculty salaries and tenure slots even at major institutions have tended to make faculty positions less desirable in many fields; a great many of the "best and brightest" students are now expecting to enter industry directly and find there many of the rewards traditionally associated with professional excellence plus money, plus entrepreneurial opportunities. In addition, industrial labs in many fields have become steadily better set up and equipped, as the distance between state-of the-art research and industrial research has decreased to the point where the center-periphery exchange system is not infrequently beginning to approximate a peer network system.

Given this changed context, what works? The answer is, probably, pretty much what has always worked: that is, interpersonal contact, usually in relatively informal settings. The increasing complexity in the body of knowledge is largely offset by the greater efficiency of interpersonal media at attending to context and finding the right subset of information to convey through interactive feedback. There is probably greater emphasis now on the team concept, and less on finding the "right" expert who will have all the answers. Industrial people in charge of bringing in new ideas report that "haunting the halls" at professional society meetings is far more effective than going to sessions. This interpersonal information sharing has been facilitated, of course, by increased exchanges of personnel in many cases usually, but not exclusively, industry people going to universities. We noted earlier the critical limitations posed by failures in shared problem describing vocabulary. The more people can talk to each other, the more they are likely to speak each others' language. This is the picture that has emerged from a number of current studies.

It is also clear that the question of effectively transferring knowledge from universities to industry cannot be effectively separated from the parallel question of the formulation of the research agenda itself. As we will note in the next section, universities have partly voluntarily, partly under compulsion become significantly more responsive to participation by industrial scientists in the formulation and execution of the applied and even the basic research agenda (this differs, of course, by fields). The important aspect to this process is that universities and industry have in many cases been able to create a climate of reciprocity in the research enterprise that has profited the ability of both sides to exchange meaningful information.

The traditional mechanism of trained students "spreading the word" about their fields within companies that hire them appears to remain effective. In fact, companies generally rate "access to students for hiring purposes" as the first or second most important reason for participating in cooperative centers (Gray and Gidley, 1986). This has been taken so seriously that one center entertained a motion from some of its member firms that students be forbidden to take jobs with companies who were not members, lest they take with them technical information paid for by the sponsors' research money. The proposal was rejected, but the concern is real in many sectors. Students really are good STI transferors, particularly since they generally have not yet developed the kind of blinders put on by social and organizational practice during the course of a career they have not yet absorbed that "we don't do things that way."

What does not work is less easy to specify. Any given information channel can supply some of the needed information some of the time. The least effective approach is almost certainly to rely primarily on one or two channels, be it on line search, journal perusal, or consultancies. Multimedia sensitivity is essential. The generally positive experience of some "technology parks" (for example, Stanford and RTI) can be at least in part attributed to this sort of "saturation exposure" bringing together university and industry personnel regularly and in a variety of settings outside those of their formal roles.

Some highly touted exchange mechanisms have been disappointing. "Technology extension" services have been experimented with in a number of states (as well as in several abortive Federal programs), but have largely failed to generate much documentable success (Watkins and Wills, 1986). Attempts to commercialize so-called "dormant technologies" by transferring them from large to small firms have been very limited in accomplishments (Roitman et al., 1984) Online data bases have proliferated, but computer aided search remains the exception rather than the rule. There are more journals than ever, but publication delays are ever-extending.

Another critical set of limitations seems to revolve around the ability of industry itself to absorb and use STI effectively. As we noted earlier, knowledge application must always take place within a social and organizational context, a context where industry's decision criteria come under increasing pressure. Several recent studies have documented the basic unwillingness of U.S. industry in many cases to invest the resources in the interpersonal exchange relationships that bring them the most useful knowledge, for long term stays in research facilities or even for short term visits (Tornatzky, Solomon, and Eveland, 1987). The Japanese, by contrast, have no such problems, and are generally willing and able to let their people go get new ideas without fearing that they will defect to a competitor (even one who might be a university). Thus industry often hobbles its own efforts to obtain the STI it knows it needs.

In sum, the critical determinants of effective academic-to industry information transfer appear to be largely a function of the ability and willingness of the parties to engage in relatively intensive interpersonal interaction over a period of time. The University/Industry Cooperative Research Centers have been one venue within which such cooperation has been achieved, although the successes of Silicon Valley and Route 128 testify to others (Rogers, 1984). Such interaction requires commitment from both academic and more difficult to achieve industry partners. But without interaction over time, no exchange system, no matter how technologically sophisticated, can hope to achieve much of lasting value.



For many years, it was widely assumed that the intellectual agenda was set by the universities and enforced by the proposal peer review system. Industry was viewed as the more-or-less passive recipient of knowledge so generated. When ideas about industrial problems fed back into universities, it was generally through one of two interpersonal exchange mechanisms: consulting relationships, which brought faculty face to face with a specific industrial problem, or hiring of people with industrial backgrounds for faculty slots. For many years, this worked.

Recent years have seen some severe shocks to this pattern. The major changes are the result of major shifts in the funding of research in universities, away from primary reliance on the Federal government and toward greater participation by industry, state governments, and other sources. These latter sources have not been content, as was the Federal government, to let the agenda be set solely by the body of "ideas in good currency" in the profession. Rather, in return for their support, they demand a role in formulating the agenda and monitoring the results. This is particularly true in the cases of cooperative centers with state support as well as industry funding. It is also the case that as industry moves increasingly into higher and higher technology their interest in research moves more toward the basic end of the spectrum and so there is increasing convergence of research priorities on both sides.

When this cooperative agenda setting is done carefully and with real commitments from all sides, it works well. The problem is that it is frequently approached as a sort of perfunctory exercise, done once a year at an "industrial advisory board" meeting with no clear understanding of the ground rules. The problem is compounded if, as is often the case, the industrial participants differ from meeting to meeting while the academic participants provide the only elements of continuity and institutional memory in the situation. Moreover, setting the formal agenda is only the first step in the research process, and formal procedures governing this stage can do only so much.

What seems to assure success is again continuing interpersonal contact the on site involvement of industry people with the academic researchers in a regular pattern of informal exchange. There is literally no substitute for talking. In a large part of U.S. academic science, the evidence indicates that many of the traditional barriers against paying attention to industrial priorities are less severe if not yet entirely evaporated.

What does not seem particularly helpful is approaching the problem at too high a level of generality, particularly at the global or Federal level. Overall task force reports detailing needed research, or high level commissions, or even National Academy or OTA reports, simply do not translate easily back into specific research projects and procedures. The primary focus has to be on the middle range that body of questions interesting to a number of firms and a number of researchers, but which is still small and focused enough to be translated without great effort into particular projects.

Inter-industrial cooperation can play a needed role here. The contributions of groups such as the Chemical Manufacturers Association or the Semiconductor Research Council in focusing industry attention on common problems and channeling that information to research performers (along with money) has been well documented. This cooperation has until relatively recently been seriously hampered by fears justified or otherwise of potential antitrust action. Law changes in the past few years have reduced this problem significantly, but there is still a mind set prevalent in industry that makes it difficult to sit down with one's competitors even when such collaboration is not harmful to the public and of positive benefit to those participating. The lesson of the Centers is that a "united front" of industrial participants is necessary to overcome the inherent tendencies of academic participants to dominate the research agenda setting process. Where industry fails to achieve this, their priorities are likely to remain relatively unaddressed.



One of the major issues that recurs in the STI exchange field is that of source credibility that is, whom do you believe, and why? A related dimension is who speaks your language that is, who gives you information in a form you can understand and use. The enormous success attributed over the years to the agricultural extension program has been attributed in significant measure to the "homophily" that existed traditionally between extension agents and their clients (farmers). Agents were in fact usually farm boys themselves, educated reasonably locally, and practicing in settings close to what they were familiar with. While they had new knowledge to share with farmers, the ways in which they shared it and the language and criteria they applied to it were similar enough to those of the clients themselves that the rapport necessary to effective knowledge transfer was easily established.

In the area of STI exchange either university to industry or industry-to-university homophily is a chancy commodity. Although there have been some changes in recent years, there has traditionally been a significant social and intellectual gap between the professor and the industrial scientist. This has made communication difficult, particularly communication aimed at influencing the professor's research priorities. To the degree that there are status gaps between the communities, information transfer in the interpersonal (i.e.., most effective) mode is impeded.

This pattern may vary significantly by field. Larsen's (1984) study of the semiconductor industry's attitudes toward cooperative research noted that industry researchers not infrequently see themselves as better qualified and more effective than their counterparts in academia, due to their proximity to the cutting edge and significantly better research facilities. In fields such as chemistry and chemical engineering, there has long been a general parity between the sectors, reflected in relatively easy personnel transfers and sharing of facilities (Thackray, 1982; Pimentel et al., 1984). In other fields such as biotechnology, universities continue to enjoy significant advantages in both facilities and status.

There is clearly unequal access to STI on the part of firms, although there are few formal barriers. The firms with the best chance of getting significant new ideas from the academic community and of in turn influencing its research agenda are those that:

bullethave significant numbers of research scientists who look and act much like university personnel;
bulletare involved in high technology with a significant basic research component;
bullethave the resources to participate in several cooperative research ventures simultaneously in different sites;
bullethave established connections with university personnel through cooperative projects, consultancies, personnel exchanges, etc..;
bulletare located in geographic proximity to the academic institution(s) with which they work, thus enabling cheap informal contact both within and outside the working environment;
bulletcan afford to take a long range view, not expecting immediate payoffs from the investment.

It should be immediately apparent from this catalog of characteristics that it is an extremely rare smaller firm that can command access to the STI infrastructure or expect to benefit from it on terms anything like those of large firms. The problem is most acute at the level of the middle sized, semi technological company, in fields where there has been a good deal of progress that has not filtered down through the system, particularly a lot of materials processing firms. Such firms do not generally have much in the way of research funds to distribute, through UICRC's or other mechanisms, and are generally largely out of interpersonal contact with the research infrastructure. There is probably a good deal of knowledge that could be used by these firms if it were somehow made available; this is the core of the "STI crisis" group, if there is one.

The real "small firm" is another issue. Such firms generally are not research dependent, particularly in terms of broad scale access to a prospectively useful body of knowledge. These firms are almost entirely single or dual product operations, which makes it extremely difficult to invest in a wide portfolio of research on the presumption that something of benefit to one or another of one's lines of business will show up sooner or later. Small firms have as much access as anyone else to the formal media online data bases, published journals, professional society meetings, etc. but, as we have noted, these are significantly less effective as sources than are the informal channels that small firms find it difficult to tap into. But for most of their purposes, they may be adequate.

An exception is provided by the small firm that is essentially a spinoff from the research program of a particular university, often the brainchild of an individual professor or even a student who goes off to found an entrepreneurial venture. These operations are likely to retain close informal ties to the university and to the academic environment. They do suffer from the problem noted earlier of limited product focus there is a great deal of the research portfolio they are simply not capable of using effectively. For example, it is extremely rare for small firms to participate in UICRC's, even when special arrangements such as reduced fees are provided.

It should not be assumed from this discussion that the industrial clientele of UICRC's is limited to the Fortune 100. Although the roster of center sponsors does include the IBM's, the Monsantos, and the DuPonts, it also includes a good many middle sized firms with aspirations. Well over 200 firms are involved in UICRC's, distributed widely across the country. On balance, UICRC sponsorship is about democratic as American industry comes.

We have centered a good deal of discussion in this paper around centers, particularly UICRC's, with good reason. The multi-sponsor center is one of the truly original ideas to emerge in the last ten years. Joint R&D ventures, even joint sponsorship of university research, are not new in the American experience. However, the creation of the center model, and its demonstrated success in both capturing the imagination of universities and industry alike and in retaining that interest past the initial "fad" phase, is the beginning, however tentative, of a recognition that fundamentally new organizational forms are needed to meet the research challenges of the next generation. Individual research projects, sponsored either by government or by industry, are simply not broadly enough based to mobilize the resources and attract the continuing involvement needed to sustain a commitment to a line of research. While the long term future of centers remains to be determined, it is clear that this approach is here to stay. Whether the newly emerging NSFsponsored Engineering Research Centers (ERC's) can develop the same momentum as the UICRC's have done remains to be demonstrated. But the basic center model is worth replicating widely.

The essence of the center concept is that of the "research mutual fund" that is, a relatively small investment broadly distributed over a portfolio of projects such that no individual project has to succeed for the investment to show a return. The value returned, while perhaps not as great as that from a single firm investment in a high payoff project, is more assured across the portfolio. Firms invest in centers partly for the return, partly out of good "corporate citizenship" but in the long run it is the quality of the interaction that keeps them involved (Gray and Gidley, 1986).

Centers typically begin as the idea of one or more professors, who mobilize the support of their colleagues, solicit the cooperation of some of their industrial contacts (typically gained through consulting services), and eventually persuade their university to go along with the deal (Eveland and Hetzner, 1984; Hetzner and Eveland, 1986). Interestingly, less than half of the industrial representatives to Center boards reported that they personally had been involved with the university prior to the Center's institution (Gray and Gidley, 1986). What this probably indicates is that creating the Center is done by industry people who do have the contacts, but maintenance of the long term relationship is carried out by functional specialists to whom the job is handed off once things get started. It bodes well for the eventual institutionalization of the centers for them to escape exclusive dependence on prior friendships and preexisting personal relationships for maintenance of the contact.

Firms generally go into centers for reasons rather different from those that impel them to support specific cooperative research projects in universities (Johnson, 1984; Gray and Johnson, 1986). Support for centers generally revolves around general expansion of knowledge and training of students; support for projects is much more likely to involve expectations for specific products or processes. Interestingly, in both types of arrangements faculty and industrial participants tend to share much the same goal structures. These findings from UICRC experience are essentially similar to results reported from a study of non-NSF cooperative centers (Tornatzky, Solomon, and Eveland, 1987).

Firms not infrequently have problems figuring out how to manage the center connection. In UICRC's, according to the latest tabulations (Gray and Gidley, 1986), 75% of the participating firms have two or more groups in close contact with the center (usually central R&D and Engineering), but only 13% have significant involvement by top management. The industrial liaison personnel in firms report an average of 10.5 requests for information about center activities from within their firms over a year, the overwhelming bulk of them being highly technical inquiries.

Centers do have an influence on firms as well as vice versa. 24% of participating firms report significant changes in the topics and issues of their internal R&D program resulting from center input; they also report an average of almost one new R&D project per firm directly stimulated by center activities. On the other hand, there are very few demonstrable new products or processes that have emerged from center research portfolios, although that is perhaps a function of their emphasis at the basic research end of the spectrum. Clearly both sides do take the encounter seriously, and are learning from each other.

In sum, the access of firms to STI sources is a function in large part of the resources they have to make use of the information. Large firms with multiple research dependent lines of business have a better chance than smaller, more limited firms of being able to tap into the research base, and of sustaining the kinds of interpersonal interactions, particularly through cooperative centers, that constitute the most effective kind of transfer mechanism. Academic research sources are still generally regarded as the most "credible" in strictly research terms, although the willingness of academics to give credence to the ideas and opinions of industrial scientists can be expected to increase in direct proportion to the amount of the research bill they pay.



The evidence on U.S. performance in STI transfer and utilization, as opposed to that of other countries, is largely impressionistic and often ideological rather than empirical. The competitiveness figures are real enough the U.S. is lagging badly as is the evidence that countries such as Japan are making applications of American basic research, particularly in fields such as microelectronics and materials sciences, at a rate far exceeding the performance of domestic firms (NSF, 1985). However, the sources of these differentials are subject to considerable debate, as are the long term consequences of this situation.

What is clear is that other countries particularly Japan have some rather different corporate culture attitudes toward what we have described as the sine qua non of effective STI exchange namely, interpersonal interaction (Moritani, 1982). In a recent study of center-type operations in the materials and semiconductor industries, a consistent observation was that Japanese industrial personnel were consistently better prepared to visit industry productively, and by contrast with American engineers, were not merely allowed but actually encouraged to visit academic research facilities for extended periods (Tornatzky, Solomon, and Eveland, 1987). U.S. industry, by several accounts, is simply afraid of its own people afraid that they will go to a competitor or perhaps even to a university, afraid of giving away any corporate secrets even when they get several times as many back, afraid of exposing its people to outside ideas.

There are also structural impediments in many parts of American industry that put them at a significant disadvantage relative to other countries in terms of effectively utilizing the STI resources to which they have access. These include a lack of attention to the sociotechnical system requirements of changing technology (Ettlie, 1986; Taylor, 1986), a confrontive rather than collaborative attitude both within as well as between firms (Moritani, 1982), and a set of organizational design assumptions and principles that simply do not reflect current global realities (Hull, Hage, and Azumi, 1984). These are extremely difficult to change, and will continue to complicate America's attempts to take advantage in the world market of her clear basic STI advantages.

The problem is compounded by lack of access to foreign STI on the part of American firms which is in turn largely a function of our own insularity and illiteracy. Most Japanese electronics engineers read and speak English; virtually no American engineers reciprocate the favor. It is sometimes argued that since we generate most of the useful knowledge anyway, access to foreign information is not particularly important. However, even if it is true that we are not missing many scientific or technical secrets, we are certainly missing potentially critical information concerning applications and the sociotechnical adaptations necessary to engineer those applications.

In sum, we have only aggregate statistics and impressions to document a negative "STI balance of trade"; we know remarkably little in a systematic comparative way about the processes and assumptions that condition this balance. Nevertheless, it is clear that American industry, for all its rhetoric, cares remarkably little about this situation and is doing virtually nothing to respond to it. Perhaps a little empirically supported apocalyptic preaching is indicated.



It is clear that industrial interest in supporting science and technology in industry is very unevenly distributed across fields of science and engineering. The study of university/industry cooperation sponsored by the National Science Board (Peters and Fusfeld, 1982) found that 60% of industrial funding of academia goes into engineering programs, about 10% to agriculture, and the remaining 30% is distributed across all remaining science fields, mostly in chemistry and computer science. Almost no industry money goes to fields such as physics, astronomy, and mathematics and still less to social sciences. Nelkin and Nelson's (1985) survey for the National Research Council noted that perhaps as much as 85% of industry funding for center-type operations is concentrated in the fields of microelectronics and biotechnology. Over two thirds of NSF's UICRC's are in these disciplines.

A recent study of academic research management in three fields (Walton, Eveland, and Tornatzky, 1985) found that electrical engineering departments (in the top 100 research universities) average about $68,000 per faculty FTE annually in research support, chemistry departments about $101,000 per FTE, and economics departments about $7,000. While all three types of departments received the preponderance of their funds from the government, electrical engineering departments received much higher percentages of their funding from industry often as much as 50% while the others received far smaller shares of the private pie.

A number of reasons may be suggested for this concentration of industrial interest in a few fields. The predominant explanation probably lies in the related criteria of the highness of the technology and the degree of value added by an academic research focus. That is, those fields that attract industry investment to universities are those that are seen as (a) rapidly changing, (b) mostly still in the basic research end of the spectrum, with only occasional commercial developments emerging, (c) too speculative for industry to devote its own resources, and (d) too potentially lucrative to stay out of.

It is interesting to note that the bulk of industry funding for a field such as microelectronics comes from firms that have this field as a peripheral rather than a central line of business. As we noted earlier, Larsen's (1984) study of university/industry cooperation in microelectronics found that industry tended to view itself as significantly ahead of universities in many key research areas, and was thus disinclined to support university programs on the grounds of a lack of value added.

It is also clear that industry interest in research and particularly in cooperative research and technology transfer is conditioned heavily by overall economic factors as well as perceptions about the research process itself. Interviews with firms that have dropped out of UICRC's (Gray and Gidley, 1986) generally indicate that overall firm profitability, with its attendant emphasis on the returns from particular activities, is the main cause for disengagement; some firms have in fact returned to the centers when their financial picture improved. Lack of value added from the research was a relatively minor cause for withdrawal; as we noted earlier, firms generally have relatively few expectations about specific payoff from such research commitment, and thus are not inclined to pull out when immediate results do not emerge. Small firms, already a small component of most centers, do have a significantly higher likelihood of withdrawal, and are more likely to get out because of payoff concerns than are larger firms.

In sum, there is relatively little systematic evidence relating to differential industry interest in university research, but what there is indicates that it is largely cutting-edge fields with high potential value added that attract investment. Considerably more inquiry into the reasons why corporations fund universities is indicated.



The generic conclusion of this treatment of STI issues is that industrial competitiveness is a problem that extends far beyond the mechanics of information transfer and that reforming or improving the information transfer system will have only limited effects in the absence of a number of other structural and ideological changes. In particular, industry must recognize that its organizational cultures set crucial limits on its ability to incorporate new knowledge effectively into ongoing processes. Assumptions about criteria particularly economic criteria need to be examined explicitly and perhaps rethought to reflect the new global realities. Assumptions about people and the relative investment in people and the contributions that might be reaped by encouraging rather than discouraging interpersonal contact likewise need addressing.

Explicit attention needs to be given to the set of processes generically dubbed "implementation" all those myriad of things that must transpire after a "big-D Decision" about technology to make it a reality in the organization. Science and technology are not self implementing, and organizations need to think as carefully about the social and organizational criteria they apply to their STI utilization decisions as they do about the technical issues involved. This must be the responsibility of research producers as well as users the concept of "implementability" should become an explicit criterion at every level.

There are few specific general solutions to prescribe. The Centers model works well in many different technical areas, but it is not necessarily the solution for all cooperative research problems. Federal labs have real contributions to make, but they will remain limited by agency mandates and accountability. Industry cannot be expected to shoulder the entire burden of deciding what works out of the total STI burden there is simply too much to sort through, and in the face of such a challenge the tendency is to withdraw. Firms command vastly unequal resources to make use of STI, but this is an issue with ramifications far beyond any STI exchange system to rectify. New information technology will for the immediate future at least make a limited contribution.

We stress again that research utilization generally and STI application in particular is a problem largely of interactivity. What solutions work are largely those that foster talking and working together rather than in separate sectors. It should be the function of and initiatives undertaken in this area to reinforce rather than undermine the potentials for interactivity.

There are some specific suggestions that should be within the power of NSF to implement. While they will not "solve" the STI exchange problem, they will have some useful impacts, particularly in terms of defining what can and cannot be expected. Ideas include:

Expansion of the knowledge base about Centers: While NSF has wisely continued to fund the evaluation effort in current and future UICRC's, it no longer has a central focus for consolidating and comparing the information so obtained and relaying that information in turn to the field. While NSF knows about UICRC's and ERC's, it has only a hazy idea how many other non NSF centers there are, what their experience has been, and how they compare on key dimensions to NSF's operations. It also has no way of documenting what role NSF may have played in the creation of centers; it would be a shame for NSF to abandon the credit where it is due.

Accordingly, NSF should:

bulletconduct a general survey of center-type operations, aimed at assembling a profile of cooperative research models and variants;
bulletconduct some research on the differential potential for cooperative research in various technical fields. As noted, current centers are extremely unevenly distributed across NSF's fields of science and engineering; there is no clear empirical evidence whether this results from chance, the inherent nature of particular academic and industrial fields, or some combination of these and other factors. Having some information on the potential for industrial support for cooperative research and other STI exchange activities in different fields would be a significant advance for NSF's planning and program development efforts.
bulletcreate a central coordinating mechanism for assembling and analyzing Center evaluation materials, either internally or under contract;
bulletcreate an assessment effort for the ERC's that is at least comparable to that for UICRC's. This is NSF's major new initiative, and it is being almost entirely undocumented in any systematic way. The scale would be approximately comparable to that for UICRC's ($5,000$10,000 per ERC) a very small total relative to its potential contribution to knowledge about cooperative research. The effort could be coordinated by the same people who manage the UICRC assessment coordination.
bulletfund periodic conferences on cooperative research models, at which analysts could present data, share observations, and generally sharpen their understanding of the social and organizational dynamics that make cooperative research work as a venue for STI exchange and utilization. Without such an effort, there is no current forum where such discussions can regularly take place. If we really believe that how people manage cooperative research is as important as what that research consists of at least in terms of accounting for its transfer and utilizability than we ought to pay attention to that management process.

Implementing these suggestions would have the additional benefit of providing NSF with a systematic base for planning its future commitment to center-type operations, as well as responding to the kinds of external and internal questions that are sure to be raised if there is any significant expansion of NSF's program. Real data is always better than informed or semi-informed guesses.

Encouraging attention to the "implementability" dimension of the research it funds: While much of what NSF appropriately funds is extremely basic research in areas of limited or no current commercial interest or involvement, it is also true that a good deal of the program, especially in engineering, does have such interest. Particularly if NSF is serious about using "industrial competitiveness" as a rationale for part of its program, it owes the industrial world its own best efforts to attend to what it knows to be the issues involved.

Accordingly, NSF might:

bulletfind some way of encouraging the research community to take a positive view of negative findings. This is a deeply ingrained academic pattern; nevertheless, NSF might be able to at least lend its influence to a professional dialogue. The first step might be a study to document the utility of negative findings; this might then be followed by the Director using his "bully pulpit" to call attention to the (expected) results of greater sharing of negative findings and stimulate discussion in professional societies. It is more a matter for "consciousness raising" than for immediate substantive change; however, a first step should be made.
bulletencourage discussion of the question of implementability in proposals and reports. This certainly could not be mandated, but there are ways of getting people to think in these terms. Certainly Centers could be strongly encouraged to attend to this dimension.

Taking a more active role in consolidating and transferring STI: NSF has flirted over the years with a variety of roles relating to STI exchange; this has usually been an early casualty of any budget restrictions. While it is not necessary for NSF to enter the field in a big way, it might consider some steps that it could take to expedite others' efforts to bring sense to various kinds of STI.

NSF might:

bulletconsider some adaptation of NIH's "consensus development" model to various fields of science and engineering. These periodic conferences are aimed at clarifying "best practice" or "what is known/unknown" in specific areas of biomedicine. An initial step might be an exploratory study of the consensus development process aimed at determining how its context is like and unlike that of NSF's STI problems. Such a short study could suggest some guidelines around which specific efforts might be undertaken.
bulletencourage professional societies to take a stronger role in information transfer beyond sponsoring meetings and journals.
bulletinteract more directly with various state programs (e.g., Ohio, Pennsylvania, Michigan) that are actively involved in experiments to transfer more STI to middle sized firms as we noted, the major "deprived" group at this point. This need not involve giving these programs actual money, but might take the form of sponsoring conferences, colloquia, and other opportunities for practices and experiences to be shared. It might also involve funding some comparative assessments of how well different approaches work in terms of reaching the target community with effective information.
bulletConsider a systematic assessment of the real STI needs of small firms. This might profitably be combined with a good empirical assessment of the existing SBIR programs.

Expansion of initiatives aimed at strengthening interpersonal and intersectoral contact among scientists and engineers: If, as we have suggested, interpersonal contact is really the most significant venue for effective STI exchange, NSF ought to be encouraging it with as many resources as possible.

NSF might:

bulletstrengthen or create new programs aimed at facilitating personnel exchanges between universities and industry. These could be sweetened in various ways in terms of institutional as well as personal incentives. An initial step might be a short study aimed at documenting the current problems with exchange programs and suggesting improvements.
bulletTry to encourage industry to see such personnel transfers as opportunities rather than threats to their personnel systems.
bulletIncrease its sponsorship of conferences aimed at bringing together industrial and academic scientists and engineers; the current annual meetings sponsored by the Division of Industrial Science and Technological Innovation are a step, but could be strengthened with more content that would attract highly reputable academic personnel. Other divisions might also try to structure parallel exercises.

It is true that a number of these suggestions sound suspiciously like the standard refrain at the end of all NSF reports: "Further research is needed..." On the other hand, NSF would be better advised to admit that it really does not have adequate information about the context within which STI exchange takes place in order to formulate effective policies, than to make policy based on anecdote, whim, and special pleading and find itself vulnerable to all the academic political pressures that can be brought to bear. A modest investment in some highly focused studies aimed at resolving particular open issues, as suggested here, would go a long way toward creating an effective response to this complex and multidimensional issue.

This is not to say that there are not immediate steps that can be taken. A number of these suggestions involve simply a recognition that a problem exists, and calling attention to it using the prestige of NSF as a lever. This is both needed and entirely appropriate. Journals such as Science, Scientific American, and disciplinary publications like C&EN are good possible fora to launch such a rhetorical assault. Ideas involving conference sponsorship would involve only modest redirection of current funds. But longer term initiatives should await a better understanding of the real dimensions of the problem.

Scientific and technical information is a complex phenomenon, and cannot be effectively considered apart from its social and organizational contexts. NSF needs to remember constantly that the science it is engaged in is essentially a behavioral process one of putting the face of human reason on the world and only incidentally the kind of epistemological exercise described in books on the scientific method (Kuhn, 1962). To the degree that NSF understands the social dimensions of what it does, it will fulfill its promise.



To be provided