Partha Dasgupta [a] and Paul A. David [b]
Toward a new economics of science
Policy Research, Vol. 23, 1994, 487-521
a - Cambridge University, Cambridge, UK
b - All Souls College, Oxford University, Oxford OX) 4AL, UK
1. Introduction and motivation: Science, economics and politics
2. The old economics of basic research, and the emergence of a new economics of science
3. Knowledge: Codified or tacit? Public or private?
3.1. Knowledge, information, and the endogeneity of tacitness
3.2. Science and Technology: Public and private knowledge
4. Priority and adherence to the norm of disclosure in the reward
system of science
4.1. Priority and the science reward system
4.2. Priority and secrecy in Science: Public virtue and private vices
4.3. Culture: The enforcement of cooperative rivalry and collective regulation in Science
5. Resource allocation within scientific fields and programs
6. The timing of research programs within Science
7. Policy challenges: Maintaining Science and Technology in dynamic
7.1. Capturing the training and screening externalities generated by open science
7.2. Managing competition for scientists between complementary research activities
7.3. Promoting greater ‘industrial transferrability’ of university research findings
Science policy issues have recently joined technology issues in being acknowledged to have strategic importance for national ‘competitiveness’ and ‘economic security’. The economics literature addressed specifically to science and its interdependences with technological progress has been quite narrowly focused, and has lacked an overarching conceptual framework to guide empirical studies and public policy discussions in this area. The emerging ‘new economics of science’, described by this paper, offers a way to remedy these deficiencies. It makes use of insights from the theory of games of incomplete information to synthesize the classic approach of Arrow and Nelson in examining the implications of the characteristics of information for allocative efficiency in research activities, on the one hand, with the functionalist analysis of institutional structures, reward systems and behavioral norms of ‘open science’ communities - associated with the sociology of science in the tradition of Merton - on the other.
An analysis is presented of the gross features of the institutions and norms distinguishing open science from other modes of organizing scientific research, which shows that the collegiate reputation-based reward system functions rather well in satisfying the requirement of social efficiency in increasing the stock of reliable knowledge. At a more fine-grain level of examination, however, the detailed workings of the system based on the pursuit of priority are found to cause numerous inefficiencies in the allocation of basic and applied science resources, both within given fields and programs and across time. Another major conclusion, arrived at in the context of examining policy measures and institutional reforms proposed to promote knowledge transfers between university-based open science and commercial R&D, is that there are no economic forces that operate automatically to maintain dynamic efficiency in the interactions of these two (organizational) spheres. Ill-considered institutional experiments, which destroy their distinctive features if undertaken on a sufficient scale, may turn out to be very costly in terms of long-term economic performance.
To say that economic growth in the modern era has been grounded on the exploitation of scientific knowledge is to express a truism. To say that what goes on within the sphere of human activities identified as ‘The Republic of Science’ has grown too important for the rest of society to leave alone is also something of a commonplace assertion.  Most of the industrial nations and many among the developing countries today acknowledge this, and virtually all societies in which modern science is practiced pay at least lip service to the belief that it is important to pursue some form of ‘science policy’.
Indeed, in the West national science policies have tended to become more strongly interventionist and more explicitly committed to ‘planning and management’. This is particularly so in the US and Great Britain, where the remarkable post-World War II autonomy of Western science communities in setting their own basic research agendas has undergone significant erosion since the end of the 1960s. To the fears of the public about scientific discoveries that may open the way to the creation of pernicious technologies, and the continuing calls for the exercise of greater ‘social responsibility’ on the part of concerned scientists, have been added two more recent arguments for closer governmental direction and control over publicly supported science. One is the generic demand for the application of fiscal restraint to a noticeably large category of expenditures, arising more from obsession with control-
1. For the origin of the term in quotes and its significance here, see Polanyi (1962) and note 5, below
ling government budget deficits than from any significant real growth in the volume of public funds being devoted to civilian R & D.  The other, however, reflects the more specific concerns of politicians and government administrators that the publicly supported parts of the scientific establishment were not addressing their efforts to the solution of national needs, and a growing conviction that the time had come to curtail the impulse of basic research scientists to ‘pursue knowledge for its own sake’ in order to redirect researchers to work on ‘applied’ projects that would bring more immediate discernable economic pay-offs. 
There is no little irony in the situation that consequently now presents itself. Supporters of political regimes that are ostensibly hostile to centralization and government planning in economic affairs, while jubilant over the comparative success of market capitalism and the manifest dysfunctions of socialist and communist systems, seem eager to subscribe to quite another position in so far as the management of science by the state is concerned; they view with deepening suspicion the largely autonomous, highly decentralized system for allocating resources that has been institutionalized and has flourished within ‘the Republic of Science’.  Perhaps because demands for closer management control over government-funded science and engineering research to improve its social payoffs do seem discordant when emanating from circles that, in other contexts, are instinctively doubtful of the public sector’s capability to allocate scarce resources efficiently, the idea of bringing the work of academic researchers into closer connection with market-oriented industrial R&D projects has lately been gaining a remarkable degree of support.
Thus, we in the West seem to have come full circle, and are approaching from a new direction the case for re-examining the institutions, organizational structures, and policies that constitute the mechanisms of resource allocation in that vital sphere of social activity we call Science. The popular press in the US has moved beyond its earlier mode of sympathetically reporting new conquests on Vannevar Bush’s ‘endless frontier’ of science (see Nelkin, 1987), along with speculations about the revolutionary consequences of such discoveries for our conception of the cosmos and the place of mankind within it, for the progress of industrial technology and health care, or for the sophistication and destructiveness of warfare. Instead, the newspapers increasingly divulge the more problematic aspects of the political economy of organized science, including some heated controversies over the criteria being used, to select among alternative science ‘projects’.
2. In the US, federal funding for civilian R&D in 1990 was only 3.7% larger, in constant (1982) dollar terms, than it had been in 1980, even though on a current dollar basis this expenditure item had swelled by 56.3% during the same period, to reach the $23.1 billion mark. However, total US expenditure for academic R&D, which in 1990 amounted to two-thirds of federally funded civilian R&D, had grown substantially more rapidly (in constant dollar terms) than the latter during the 1980s - having increased by 69.3% in real terms between 1980 and 1990 (see National Science Board (1991), Appendix Tables 4-17 and 5-1).
3. In Time Magazine for 23 November 1992, Walter Massey, then director of the National Science Foundation, is quoted as saying: “The public hears that we’re No. 1 in science, and they want to know why that fact isn’t making our lives better. The one thing that works in this country doesn’t seem to be paying off.” The story goes on to point out that the shift underway in the US government’s stance towards science was not the product of the new administration. “Long before the [presidential] election, policymakers were concluding that they should assert more control over research by telling many scientists precisely what to work on. ‘We’ve got to do some readjusting’, says Guyford Sever, co-chairman of a recent Carnegie Commission study on the future of American science.” see Thompson (1993) p. 34.
4. To appreciate fully the irony in this, one need only read the following extract from a well-known critique of the outcome of the enormous growth of a science establishment supported by the State and large business corporations:
… as these developments have proceeded in a uncoordinated and haphazard manner, the result at the present day is a structure of appalling inefficiency both as to its internal organization and as to the means of application to the problems of production or welfare. If science is to be of full use to society it must first put its own house in order.
The author was not a businessman taking his turn at bat as a member of the Reagan-Bush ‘teams’, nor a populist investigative congressman, out to expose the abuses of academic research institutions, nor a science advisor or minister in the Thatcher government. Rather, the passage is quoted from The Social Function of Science [10, p. xiii] by J.D. Bernal - a scientist and Marxist historian of science and technology, who took a leading role in the ‘social relations of science’ movement in Britain during the 1930s!
For example, has the US Congress been successfully lobbied to spend excessive sums on enterprises in ‘big science’, like the superconducting supercollider, at the sacrifice of a myriad undertakings in ‘little science’? Should we be trying to launch extensive, publicly funded programs like the effort to sequence the human genome, as soon as techniques and instrumentation that permits such research become available? Or, should some lines of investigation be deferred in favor of others? Should the availability of supplementary state and local government subsidies be made a deciding consideration when the National Science Board selects among groups of academic researchers (located at competing sites) those who are to receive the bulk of funds allocated for a particular scientific sub-field? Are public and private monies being channelled excessively into some fields of basic research, while other, potentially fruitful areas remain virtually drought-stricken? Is the amount and distribution of public funding for university laboratories insufficient to permit them to keep pace with rising costs of state-of-the-art equipment, such as is made available to researchers in industry?
Still other news reports reflect growing recognition of the subtle and complex issues facing those who must frame science policies affecting the conduct of basic research in an environment of global economic competition. Why should any nation continue to devote a significant portion of its public expenditure to advancing scientific knowledge if, through the global networks of the international science community, those new discoveries soon will be made available to allies and rivals alike? How can one tell whether or not it is on balance advantageous for a particular country to have its scientists participating in a unrestricted exchange of information with colleagues working abroad? Is the US failing to provide graduate (research) training for its nationals in the natural sciences, mathematics, and engineering, while at the same time subsidizing such training for too many foreign nationals? What is it about research universities that creates barriers impeding the easy and rapid ‘transfer’ of new scientific knowledge to the sphere of commercial applications - even in the US, where cultural traditions of academic separation from the world of business affairs are far less strong than remains the case in western Europe? To what extent is it desirable to modify modern university institutions and operating rules to permit and encourage closer integration of academic and corporate research activities? Is it a legitimate cause for national concern when foreign corporations actively recruit university scientists to staff company-run basic research institutes set up in their immediate vicinity (as was the case when a computer science research institute under Japanese corporate management was located on the doorstep of the University of California in Berkeley)? Would there be no equivalent cause for concern were the business corporations American in ownership? 
It is thoroughly understandable why economists have remained much preoccupied with theoretical and empirical studies of the sources of technological innovation and its connections to productivity growth and improvements in economic well-being and national economic power. At the same time, in the present setting, it is a surprising and rather regrettable fact that elaboration of the economic analysis of technology was for some time allowed to run far ahead of the economics of science. In the following pages we undertake to report on recent analytical developments that may enable the laggard member of the pair to catch up, and which hold out the prospect that the two naturally interrelated areas of study can begin moving forward more swiftly in tandem. Our economic analysis of the organization of research within the spheres of science and technology emphasizes the point that we are dealing with an interrelated system, comprised of distinct activities that may reinforce and greatly enrich one another, but, furthermore, that it is a system that remains an intricate and rather delicate piece of social and institutional machinery whose con-
5. For a sample of recent reporting of some of these issues, see: Japanese Labs in U.S. Luring America’s Computer Experts, New York Times, 11 November, 1990; Graduate Schools Fill With Foreigners, New York Times, 29 November, 1990; Foreign Graduate Students Know Hardship, New York Times, Letters, 13 December 1990; William J. Broad, Big Science - Is it Worth the Price?, New York Times, Feature Series, 27 May, 29 May (on superconducting supercollider), 5 June (human genome project), 10 June (NASA space station project), 19 June (NASA earth observing system project), 4 September (small-scale basic science funding impacts), 4 October (hot fusion project), 25 December 1990 (cutbacks funding for big science projects).
stituent elements also may become badly misaligned. Indeed, there does not seem to be an adequate appreciation of the vulnerability of the science-technology systems in the West today, for all the frequency with which their importance to the modern economy and polity is acknowledged; nor of the basic features that are common to these variegated institutional and cultural structures, and which render all of them susceptible to destabilizing and potentially damaging experiments which may soon be embarked upon in the earnest hope of more fully mobilizing the respective national scientific research communities in the service of national economic security - the successor goal to military security - that is now being promoted under the euphemism of ‘competitiveness’. Of course, the best preventative against blind and costly social experimentation that we can recommend is a prior investment in acquiring and disseminating deeper scientific understanding of the subject of concern. That, precisely, is our purpose in the present article.
We must begin by acknowledging that it would hardly be fair to charge economists with having simply ignored problems of resource allocation in relation to science. Before there could be a new economics of the subject there must have been an ‘old’ economics of science, and so there was. The emergence of ‘shortages’ in the market for scientific and engineering personnel attracted the attention of economists (e.g. Blank and Stigler, 1957; Arrow and Capron, 1959) early in the post-Sputnik era, and the literature that subsequently expanded on the economic determinants of the flow of newly trained scientists and engineers was greatly enriched by applications of concepts and empirical methods from the theory of human capital. Moreover, the seminal contributions to the modern analysis of the production and distribution of basic scientific knowledge date from the same period, notably the analytical work of Nelson (1959) and Arrow (1962), on the implications of the difficulties of privately appropriating the economic value of basic research findings. Closely related to those developments in the literature are the sequels to Griliches’ (1960) pioneering empirical effort in quantifying the economic ‘spillovers’, or ‘un-appropriated social benefits’ that might flow to society at large from science-based innovation.
The crop of economic writings that sprang from those few seeds constitutes an elaboration of three core propositions which are fundamental to most of what contemporary analysts have to say about the allocation of resources for basic scientific research.  First, the economic value of basic research is difficult to forecast, or even to gauge in retrospect. Economic payoffs entrained by scientific discoveries may come quickly, but more often are not realized for a long time. Since basic research occurs on the frontiers of knowledge, its outcomes are highly uncertain. Second, realization of economic rents (‘profits’) from a basic research advance also will be impeded to the degree to which property rights in such discoveries are intrinsically difficult to establish and defend, and because the organizational norms within which much of such research is conducted (by academic scientists) inhibits effective assertion of individual property rights that can readily be conveyed to other parties, such as business corporations. From this it follows that the private returns to investment in basic research are highly uncertain, especially in relation to the benefits that will accrue to society as a whole. The economic payoff to society can be quite large compared to the amount invested, because a fundamental advance in knowledge can serve as an input for applied research and commercialization efforts, leading to many new products and processes.
This divergence between the private and social returns to basic research outlays has led to the third and most influential proposition, namely, that there exists a systematic ‘market failure’ which, in the absence of remedial actions, would result in societal ‘underinvestment’ in science.  The modest share held by basic research within the R&D budgets of US corporations (about 5% in 1985, for example) lends a good bit of credence to this conclusion, even if it cannot indicate whether company financing of basic science is below the level that would be socially optimal.
6. The summary statement given here draws on the exposition in David et al. (1992)
Such qualifications have been left aside, and the foregoing ‘market failure’ argument has stood as the central economic rationale for the public funding of science for the past 30 years (see Mowery, 1983).
As impressive an achievement as that may be in its practical consequences, what we may respectfully refer to as ‘the old economics of science’ did not go much farther. Its approach to the production, dissemination and uses of scientific knowledge was partial in nature, and therefore largely incomplete in its ability to come to grips with most of the concrete issues in the formulation and implementation of public policies affecting basic science. The literature in question began by treating knowledge as a commodity, more specifically, as a durable public good. However, although the peculiarities of information as a commodity were recognized from the outset, for the purpose of analyzing the central issues of public finance posed by basic science, economists began by treating knowledge entirely on a par with other forms of durable public goods. Little effort was directed to exploring the resource allocation implications of the respects in which knowledge differs from other durable public goods, such as, for example, laws and constitutions, or lighthouses, for that matter.
One consequence was that the rules, regulations, and more generally ‘norms’ and ‘customs’ governing the production of knowledge and its uses by members of the various research communities, ranged under the headings of academic science, government science, and industrial science, were taken as given by the older tradition - if, indeed, they happened to be noticed explicitly. That these norms and characteristic organizational forms are in fact sharply differentiated across these communities of scientists, and quite unlike the background institutions relevant in the case of other durable public goods, was not seen as something of particular economic import, and therefore deserving of explanation. In the older tradition, neither the historical origins nor the possible present raison d’etre of these generalized ‘institutions’ are subjects of active inquiry among economists or economic historians.
In a pair of joint articles (Dasgupta and David, 1987, 1988) developed over the past several years we have tried, by synthesizing insights from economics and sociology, to carry the economics of science beyond the traditions of this older literature. A first goal was to provide an ‘explanation’ for the prevalence of distinctive norms, customs and institutions governing university science, on the one hand, and industrial R&D, on the other.  The phenomena that interested us naturally included those salient social arrangements that have occupied the attention of sociologists and philosophers of science: rules of priority and the role of validated priority claims in the reward structure of (academic) scientists, patenting and disclosure policies, and institutions associated with scientific communication and the functioning of a collegiate reputational reward system - ranging from ‘invisible colleges’ to organized, professional academies in various disciplines. This explanatory approach, however, departs from that of the sociologists and philosophers.
A considerable part of the characteristic style of analysis in the new economics of science derives from its recognition of three features of the processes for the production, dissemination, and use of knowledge. First is the fact that certain crucial inputs, such as research effort, care, innate scientific talent, and the realization of elements of chance in the process of discovery, are very costly for outsiders to monitor, and in most cases are not even observable jointly by the ‘principal’ who is sponsoring the inquiry (e.g. a private or public patron or employer) and the principal’s ‘agent’ (the researcher). Second is the observation that there often are significant aspects of indivisibility, and attendant fixed costs and ‘economies of scale’, inherent in the underlying processes of knowledge production. Third is the point that the knowledge generated in research activities, rather than being inherently available to others, can be kept from the public
7. The existence of a positive gap between the social rate of return on basic research outlays and the private rate of return on industrial capital expenditures has been repeatedly confirmed empirically by sample calculations that elaborate on the methodology of Griliches (1960) - see Mansfield (1991) for recent examples; but the problems of the representativeness of the sample cases studied remain formidable.
8. We associated university science with the world of ‘episteme’ and the republic of science, whereas the latter we associated with the realm of technology.
domain should the researcher so choose; the characteristics of the reward system, along with the costs entailed, determines what information gets disclosed fully, what is disclosed partially; and what is kept secret. Given these structural conditions, and the long-standing observation that the production of knowledge is shot through with uncertainty,  it is quite in order to re-examine features of organized research activities from the perspective offered by the recently developed analysis of resource allocation under conditions of asymmetric information.
As will be illustrated by this paper, the new economics of science has the two-fold ambition of (1) exposing the underlying logic of the salient institutions of science, and (2) examining implications of those differentiating institutional features for the efficiency of economic resource allocation within this particular sphere of human action. To carry out this program we will be building upon the foundations laid down by the classic contributions in the sociology of science,  adding to the insights provided by the ‘old’ economics of science some new ones that are drawn principally from the rapidly growing analytical literature that treats problems of behavior under incomplete and asymmetric information (including the economic theories of agency and optimal contract, or ‘mechanism design’ theory), as well as issues in the dynamics of racing and waiting games.
The remainder of this article has the following organization. Section 3 defines some terms relating to the concepts of knowledge and information, and introduces the generic problem of analyzing the relationship between market and non-market ‘systems’ for governing the production and distribution of knowledge. The logic of the reward system in academic science, its relationship to the Mertonian norm of ‘communality’ (complete free disclosure), and the central role played by priority of discovery, are briefly considered in Section 4. Section 4 then goes on to sketch an economic theory of departures from the norm of complete disclosure by treating as endogenous the location of the boundary-line between knowledge that is codified and that which is left tacit by scientists. The question of the efficiency of resource use under the institutional constraints of a collegiate reputation-based reward system funded by public patronage, such as exists in many Western nations, is considered next. Sections 5 and 6, respectively, focus on sources of misallocation within and between fields at a given point in time, and in the timing of support provided for research programs. It is shown that some of the sources of these allocative inefficiencies lie close to the institutional core of the system, so that the problems of correcting them are by no means straightforward.
Although most of the readily observed qualitative features familiar to sociologists of modern science, working at both the macro-institutional and micro-behavioral levels, can be accounted for by our analytical approach, it is symptomatic of the relatively underdeveloped status of the new economics of science that many of its implications remain unexplored and untested against systematic, quantitative data. Furthermore, as is generally the case when a new theoretical perspective is gained, new questions and puzzles arise even as old ones are settled. The agenda for future research in this field, therefore, remains both extensive and varied. Nevertheless, even in its present nascent state, the new economics of science can offer some measure of guidance for science policy, suggesting points at which public interventions in the resource allocation process are likely to be socially beneficial and others where they are probably best avoided - leaving discretion to individual scientists and scientific communities. Specific implications bearing on issues noticed in the introduction are remarked upon throughout the text and notes of Sections 3-6. Section 7 takes up a more extended analysis of several interrelated policy issues raised by current moves to: (1) readjust the balance of incentives influencing the production of Ph.D. scientists and engineers and to alter the way trained researchers
9. For early efforts to quantify the uncertainty surrounding the duration, costs, and success of industrial research projects, see Mocking (1962) and Norris (1971). We are unaware of comparable studies based on the experience of university scientific researchers, but there is a presumption that, because the latter are less constrained to select problems where the feasibility of a solution is less clearly established, the research duration and costs associated with a successful outcome typically are even more difficult to predict with accuracy.
10. For example, Merton (1973), Polanyi (1951, 1962, 1966), Hagstrom (1965), Ziman (1968), Gaston (1970), Zuckerman and Merton (1971, 1972), Ravetz (1971), Cole and Cole (1973), Salomon (1973), Blume (1974), Mulkay (1979). Luhmann (1979), Nelkin (1987), Whitley (1984).
distribute themselves among entities engaged in open science, mission-oriented government research, and proprietary R&D; (2) encourage ‘transfers’ of academic research results to industry, by modifying university regulations and contractual arrangements involving intellectual property rights. The brief concluding Section 8 therefore offers no more than a summary set of propositions to underscore the general cautionary message emerging from the analysis: short-run policies aiming to shift resources towards commercial applications of scientific knowledge (by dismantling bits of the distinctive institutional infrastructures of Science, or by weakening its ability to hold talented researchers in the face of competition from business) may seriously jeopardize a nation’s capacity to benefit, from a sustained flow of innovations based upon advances in scientific and technological knowledge.
As is the case in other specialized fields of inquiry, in this one some specific terminological conventions will be employed. A few initial definitions and distinctions will avoid later confusions, especially since many of the terms we employ are given somewhat different meanings in ordinary language usage and in the scholarly literature on the intellectual and social organization of the sciences.
By the term ‘information’ (following common usage in economics) we will mean knowledge reduced and converted into messages that can be easily communicated among decision agents; messages have ‘information content’ when receipt of them causes some change of state in the recipient, or action.  Transformation of knowledge into information is, therefore, a necessary condition for the exchange of knowledge as a commodity. ‘Codification’ of knowledge is a step in the process of reduction and conversion which renders the transmission, verification, storage and reproduction of information all the less costly.  Yet, somewhat paradoxically, this transformation makes knowledge at once more of what is described (in the public finance literature) as a ‘non-rival’ good, that is, a good which is infinitely expansible without loss of its intrinsic qualities, so that it can be possessed and used jointly by as many as care to do so.  Thus, codified scientific knowledge possesses the characteristics of a durable public good in that (i) it does not lose validity due to use or the passage of time per se, (ii) it can be enjoyed jointly, and (iii) costly measures must be taken to restrict access to those who do not have a ‘right’ to use it. 
In contrast, tacit knowledge, as conceptualized by Polanyi (1966), refers to a fact of common perception that we all are often generally aware of certain objects without being focused on them.  This does not make them less important: they form the context which makes focused perception possible, understandable, and productive. No less than other human pursuits, science draws crucially upon sets of skills and techniques - the
11. Shannon’s well-known measure of the quantity of information (see Shannon and Weaver, (1949)) was formulated to serve the specific needs of communications theory, and has been found unsuitable for application in economics. On this see, e.g., Arrow (1964) and Marschak (1971).
12. This usage is related to, but does not carry the broader implications of the concept of ‘codification’ in science as developed by Zuckerman and Merton (1972): “The consolidation of empirical knowledge into succinct and interdependent theoretical formulations.” According to the latter, scientific fields differ in this regard, with those disciplines that remain in what Kuhn (1962) would call a ‘pre-paradigm state’ being necessarily less ‘codified’.
13. Were person A to give person B a piece of information concerning Q, that would not reduce the amount of information concerning Q that was retained by A, (although, to be sure, the benefit to each would depend upon whether and in what manner B were to make use of the information). Romer (1990, 1993) has recently popularized this application of the term ‘non-rival’ good, but see David (1993b) for the proposal of ‘infinite expansibility’ as an alternative term that is less confusing, in that it acknowledges the possibility of rivalries for possession of new information.
14. Even in the case of codified knowledge the condition of strict non-excludability (namely, that once produced it is impossible to exclude anyone from benefitting from it), which is taken to be one of the hallmarks of a pure public good, does not hold; patents, copyrights, and trade secrecy are institutional devices for denying others beneficial access to information.
ingredients of ‘scientific expertise’ - that are acquired experientially, and transferred by demonstration, by personal instruction and by the provision of expert services (advice, consultations, and so forth), rather than being reduced to conscious and codified methods and procedures. The transfer process itself, as a rule, is a comparatively costly affair (in contrast to the case of codified knowledge) for both the provider and the recipient of tacit knowledge, but, like information, tacit knowledge can be swapped in transactions resembling ‘gift exchanges’, or sold for money, rather than being shared freely. 
Insofar as codified and tacit knowledge are substitutable inputs (at the margin) in the production of further knowledge, or in practical implementations, the relative proportions in which they are used is likely to reflect their relative access- and transmission-costs to the users. Similarly, differences in the extent to which knowledge generated by researchers in various fields gets codified for packaging as information, rather than retained in a tacit form, will reflect the reward structures within which researchers are working, as well as the costs of codification. Hence, variations in the relative importance of codified and tacit knowledge in the work of different research communities has no necessary connection with the ‘hardness’ or ‘softness’ of their respective disciplines. This perspective, afforded by the new economics of science, stands in contrast with the disposition of some philosophers and historians of science to associate a relatively high degree of codification with occupancy by the discipline in question of a superior position in some epistemological or methodological hierarchy.  Some recent discussions of the economics of R&D and technology transfers (e.g. Pavitt, 1987; Nelson, 1990; Rosenberg, 1990) continue to assign special significance to the tacit elements in technological knowledge, calling attention to the fact that the information contained in patents, blueprints and other codified forms of knowledge often are insufficient for successful implementation of the technical innovations they purport to describe; much complementary ‘know-how’ may be required, the acquisition of which, typically, is a costly business. While we regard the factual proposition to be beyond dispute, to our way of thinking its does not imply the existence of underlying, intrinsic differences in the nature of ‘technological’ as opposed to ‘scientific’ knowledge.  Certainly, we would dispute the assertion that technological knowledge, on that account, should be assigned a subordinate epistemological status. Furthermore, we find no compelling grounds for associating the tacit knowledge of either technologists or scientists necessarily with skills that are specific rather than ‘generic’ in their applicability.  Going back to Polanyi’s perceptual analogy, what gets brought into focus (and codified) and what remains in the back-
15. Whereas philosophers of science (e.g. Nagel, 1961; Hempel, 1966; Popper, 1968) emphasized the epistemological bases for distinguishing between the scientific and other modes of human understanding, stressing formal methodological rules of ‘scientific procedure’, Polanyi’s (1966) insights concerning the ubiquitous role of tacit knowledge has led some modern sociologists of science (e.g. Latour and Woolgar, 1979; Latour, 1987) to study the practice of laboratory science as a ‘craft’.
16. See Hagstrom (1965) on gift exchanges among scientists, and Arora (1991) on contractual arrangements for the sale of tacit information by business firms.
17. See, for example, Kuhn (1962), and also Nagel (1961), Hempel (1966), and Popper (1968), who suggest that it is the role played by formal methodological rules of ‘scientific procedure’ that provides epistemological grounds for distinguishing between (hard) science and other (softer) modes of human understanding.
18. This challenges Vincenti’s (1990a, p. 19) characterization of his writings as having shown that “engineering knowledge warrants recognition as an epistemological species in its own right”. In actuality, however, it is primarily on sociological and behavioral, rather than epistemological, grounds that Vincenti (1990a,b) shows engineering and science to be different pursuits. He says knowledge is an instrumentality in both, whereas they give priority to different ends: scientists to understanding of ‘how things are’, and engineers to ‘how things ought to be’, i.e. with solving practical problems. Even supposing that was so, it would not logically imply a difference in the nature of the means (knowledge) employed, or in their reliability. Furthermore, Vincenti’s differentiation on grounds of proximate goals rather closely resembles our view of Science as a sphere of activity whose organization is conducive to the rapid growth of the stock of knowledge, whereas Technology is concerned to achieve rapid growth of the material benefits from new knowledge.
19. Nelson (1990) makes the same point, but associates what he refers to as the ‘generic’ parts of technological information with codified knowledge that can be readily transferred, and which tends therefore to become public, whereas he identifies the tacit knowledge components with ‘specific’ technological information that is privately held.
ground (as tacit knowledge) will be, for us, something to be explained endogeneously by reference to the structure(s) of pecuniary and non-pecuniary rewards .and costs facing the agents involved. Although the position of the boundary between the codified information and tacit knowledge in a specific field of scientific research may be shifted endogenously by economic considerations, the complementarity between the two forms of knowledge has important implications for the way research findings can be disseminated. We comment on these below (in Section 7.3, especially) in connection with recent discussions of measures to promote the transfer of university research results to industry for further development and commercial exploitation.
Although the foregoing discussion has referred to scientists and technologists in ways that suggest they are different kinds of knowledge-workers, the crucial distinction we seek to draw is between the social organizations of Science and Technology. By capitalizing the terms we signify that these labels are being used here to refer to the two sets of socio-political arrangements and their respective reward mechanisms affecting the allocation of resources for scientific research. For our purposes, what fundamentally distinguishes the two communities of researchers is not their methods of inquiry, nor the nature of the knowledge obtained, nor the sources of their financial support. To be sure, differentiations between can be drawn along those lines, as already noted, but to our way of thinking much of the economics literature, like philosophy of science, has mistakenly focused upon science-technology distinctions that are epiphenomena of differences that lie at a deeper level. It is the nature of the goals accepted as legitimate within the two communities of researchers, the norms of behavior especially in regard to the disclosure of knowledge, and the features of the reward systems that constitute the fundamental structural differences between the pursuit of knowledge undertaken in the realm of Technology and the conduct of essentially the same inquiries under the auspices of the Republic of Science.
Loosely speaking, we associate the latter with the world of academic science, whereas Technology refers to the world of industrial and military research and development activities. What makes a knowledge-worker a ‘technologist’ rather than a ‘scientist’, in this usage, is not the particular cognitive skills or the content of his or her expertise. The same individual, we suppose, can be either, or both, within the course of a day. What matters is the socio-economic rule structures under which the research takes place, and, most importantly, what the researchers do with their findings: research undertaken with the intention of selling the fruits into secrecy belongs unambiguously to the realm of Technology. Secrecy, however, is more readily effected when knowledge is not codified in proprietary documents (e.g. blueprints, receipts for chemical syntheses) that can be purloined and published, but instead is retained in a tacit form. Training services that convey tacit information and .access to contracting with the trained are commodities that can be, and are, exchanged for value by business organizations operating within the realm of Technology, and also by academic research organizations, although the enforceability of contracts tends to differ between the two cases, and the terms on which tacit information can be sold are correspondingly different.  None of this implies that profit-seeking business firms would not find it to their advantage on occasion to invest some resources in ‘basic’ research, or organize research facilities in ways that emulated the open, cooperative environment characteristic of university campuses. Neither does it imply that academic scientists will never seek to benefit materially by patenting their inventions, or always refrain from rivalrously withholding research findings and
20. Of course, as we have acknowledged in the text above, epistemological and methodological differences exist between the two research communities. For example, it is widely believed that among applied industrial scientists, engineers, and like technologists the balance between codified and tacit knowledge (‘practical know how’) tilts more towards the latter than is the case among university researchers. Yet as we already have suggested, such a difference could well be a consequence of the different information disclosure rules and reward systems that obtain.
methods from university-based researchers in their field. 
It is equally possible to arrive at the foregoing, highly stylized view of the scientific research world via another route, namely, by starting from a consideration of the main alternative resource allocation mechanisms that can be used to produce and distribute scientific knowledge, and considering the efficiency with which they will perform those tasks. Of all possible resource allocation mechanisms, the one that has been most studied in economics is the ‘market mechanism’. As is now well known, if the market mechanism is not aided by further social contrivances, such as, for example, intellectual property rights, there is no basis for supposing it can sustain an efficient production of knowledge. The market mechanism has a tendency to discourage the production of public goods because of an inability on the part of producers to appropriate fully the value of the fruits of their efforts.  Three generic remedies have been devised to overcome the deficiency of the market in this regard: two of them do so by seeking to rectify the problem at its source, whereas the third solution applies correctives in the form of supplements to the market outcomes. Let us briefly look at each of these schemes in turn.
(1) The first scheme consists of the government engaging itself directly in the production of knowledge, allowing free use of it, and financing the production cost from general taxation. This was at the heart of Samuelson’s (1954) analysis of the efficient production of public goods. Government research and development (R&D) laboratories that publicly disclose their findings, such as agricultural research establishments, - are an ex ample of this. It is as well to note that under this scheme the volume of public expenditure in the production of knowledge, and the allocation of expenditure for the production of different kinds of knowledge, are both public decisions: they are decisions made by the government.
(2) A second possible solution is for society to grant intellectual property rights to private producers for their discoveries, and permit them to charge (possibly differential) fees for their use by others. This creates private markets for knowledge. Patents and copyrights are means of defining and protecting intellectual property rights, and as their strengths and weaknesses have been discussed extensively by economists over the years, we shall not embark upon that subject here.  It should be observed, nonetheless, that the producer (or owner) of a piece of information in this scheme, ideally, should set different prices for different buyers, because different buyers typically value the information differently.  One problem with such markets is that they are inevitably ‘thin’ (each market is essentially a bilateral monopoly, i.e. consisting of the seller and a single buyer), and therefore not a propitious environment for the emergence of prices that will sustain an efficient allocation of resources, as has been pointed out by Arrow (1971). Another problem with them arises from the fact that transactions in knowledge are shot through with leakage. The point is that for an exchange to be conducted efficiently, both parties need to know the characteristics of the commodity being transacted. In our example, the potential buyer needs to know what the piece of information is before the transaction is concluded, but once the potential buyer gets to know the information, it is difficult for the seller (e.g. the inventor-developer) to prevent the prospective buyer from benefitting from what she has learned, should the contemplated transaction not be concluded.  This feature of knowledge
21. It is the identification of Weberian ‘ideal types’ that endows the instances of deviance with interest and potential significance, as will be seen below in Section 4.2, for example.
22. This is the well-known ‘free rider’ problem in the supply of public goods. The earliest writers on this were Knut Wicksell and Eric Lindahi. See Musgrave and Peacock (1958) for abridged versions of the Wicksell-Lindahl analyses. Modern statements of the allocation problem associated with public goods are in Samuelson (1954) and Arrow (1971). The basic economics of public goods can be found in any textbook on public finance (e.g. Musgrave, 1968; Stiglitz, 1988.)
23. See for example, David (1993a,b) for citations to and discussion of recent contributions to the economics literature on intellectual property rights.
24. In economics, these variegated prices are called Lindahl prices, in honour of Eric Lindahl, who first articulated a scheme of this kind, (see Musgrave and Peacock, 1958).
25. Arrow (1962) argued that this is a particular difficulty whenever the transaction involves the fruits of fundamental research, for the findings of such research have possible applications in wide varieties of fields, some of which are not known to the would-be seller.
(that its value is often very difficult to quantify) will figure prominently in what follows. It means that the economic use-benefits of knowledge are often hard to appropriate privately, and therefore to market efficiently. This is so even when patent protection of knowledge gives the owner transferable legal rights to exclude others from using that knowledge in many specified contexts. 
The foregoing tells us that despite the limitations of the institutions of patents and copyrights (and, for that matter, the legal protections for secrecy among individuals and organizations), those ‘property-like’ contrivances provide means for privately appropriating profits from discoveries and inventions. In short, while information can in principle be used jointly, joint use can be prevented by legal prohibitions or through the practice of secrecy. This can be socially desirable, because even though monopoly in the use of knowledge is inefficient (it involves the under-utilization of knowledge), it can be offset by the act that the lure of monopoly profits makes researchers undertake R & D activity today. Therein lies the value of instituting patent laws, and of allowing secrecy to be practised by discoverers.
(3) A third possible scheme is for society to encourage private production of knowledge by offering public subsidies for its production, and by relying upon general taxation to finance these subsidies. A critical feature of this arrangement is that producers are denied exclusive rights to the output of their R&D activity: once it is produced, the knowledge is made freely available to all who care to use it.  In albeit imperfect forms, this scheme characterizes research activities carried on in public and private non-profit entities, such as universities, where much of the knowledge that is produced is prohibited from being patented by the private individuals involved in creating it, and where salaries, promotions and equipment are paid out of public funds (the subsidies!). We will study this resource allocation mechanism further in Section 4, where we will examine the role assumed -by a special form of intellectual ‘asset’, namely an accepted claim to priority of discovery or invention which, though not a legally recognized property right, nonetheless plays a crucial role in making the scheme more effective than it would otherwise be.
While the public finance literature provides some guidance to the problem of providing for the production of scientific knowledge, it does not take us very far. There are many, points where the standard prescription of subsidies remains unsatisfyingly vague. Should the subsidy be based on scientists’ research time? Or should it be adjusted according to the nature of their endeavors? Or should it instead be granted on a piece-rate basis for ‘output performance’ measured in some convenient way - such as publications in scientific journals, or the frequency with which their (codified) findings are cited in other scientific publications, and in patents? As it happens, this vagueness reflects a deeper weakness. The literature leaves unexplored the intricate institutional structure governing the practice of science. Nor does it build on the fact that research activity not only involves considerable uncertainties, but is shot through with choices the consequences of which are publicly unobservable. Without an understanding of these features of scientific activity, it is not possible to devise efficient subsidy schemes. Recent explorations in the economics of science focus their attention on these matters. We turn to them now.
It will be noticed from even the preceding sketch that the first and third resource allocation mechanisms in our list are similar in an important way: the output of research is publicly shared. In contrast, the second mechanism treats output as a private good. To be sure, the second scheme differs from the other two also as regards the source of funds: private versus public. But we maintain that to emphasize differences in the sources of funding is not a particularly illuminating way of distinguishing Science from Technology. Government laboratories engaged in military research are publicly funded, but the secrecy that
26. Matters are easier in the case of narrowly restricted knowledge about new technical processes and practical devices. This partly explains why it is a commonplace today to see A paying B a license fee for using B’s patent on the manufacture of a new product, or on a new process for manufacturing an old product.
27. See Pigou (1932), Baumol and Oates (1975) and Dasgupta and Heal (1979).
is maintained over their findings makes them part of Technology. 
So, for the purposes of the present discussion we can amalgamate the first and third schemes into one - on the grounds that their attitudes towards controlling the output of research are similar. Having thereby reduced the distinctive modes of organization from three or two, one may still ask: Is this still not one more than is really needed? The question can be answered in the negative, on the functionalist grounds that they are adapted to serving distinctive goals: the community of Science is concerned with additions to the stock of public knowledge, whereas the community of Technology is concerned with adding to the stream of rents, that may be derived from possession of (rights to use) private knowledge.  The arrangements within each can be shown to have an internal coherence and logic which, while not perfectly efficient, is quite well-suited to serve its (imputed) purpose.  Thus, because their functional attributes are not the same, there is no cause for thinking it is paradoxical for a given society simultaneously to organize scientific inquiries in two quite distinctive modes, maintaining the two sets of institutional mechanisms and behavioral norms characteristic of the research communities we have delineated. Although it is evident from the contradictory norms on which they are based that there is a tension between these two modes of economic and social organization, so that they do not ‘mix’ easily, the two are not mutually exclusive ways to successfully organize the pursuit of scientific knowledge within the same society. Indeed, we will argue that in order to ensure a reasonably efficient allocation of resources in the production of knowledge, modern societies need to have both communities firmly in place, and attend to maintaining a synergetic equilibrium between them.