Communication and Knowledge: How is the knowledge base of an economy constructed?

return Communication and Knowledge: How is the knowledge base of an economy constructed? 15 th Annual Meeting of the Society for the Advancement of Socio-Economics (SASE), Aix-en-Provence, 27 June 2003 Loet Leydesdorff University of Amsterdam, Sc ience & Technology Dynamics Amsterdam School of Comm unicat ions Research (ASCoR), Kloveniersburgwal 48, 1012 CX Amsterdam, The Netherlands http://www.leydesdorff.net ; loet@leydesdorff.net Abstract The competitive advan tages in a knowledge-based econo my can no longer be attributed to single nodes in the network. Political eco nomies are increasingly reshaped by knowledge-based develop ments that upset market equilibria and institution al arrangements. The network coordi nates the subdynamics of (i ) wealth production, (ii) organized novelty p roduction, and (iii) private approp riation versus public con trol. The interaction terms generate a com plex dynami cs which cannot be expected to cont ain central coo rdination. Howe ver, the knowledge infrastru cture of systems of innovations can be measur ed, for example, in terms of university- industry-government relations. The mutual information in th ese three dimensions indicates the globalization of the knowledge base. Patent sta tistics and dat a from the Internet are compared in terms of t his indicator. Introduction Whereas organizations and institutions can be id entified as observable units of analysis or nodes in a network, knowledge develops operationally in terms of reconstructions at the level of the links. Knowledge flows both within orga nizations and across in stitutional boundaries. In order to study organized know ledge production, therefore, one first has to distinguish analytically between the in tellectual and the institutio nal organization of knowledge production systems (Whitley, 1984 ). The intellect ual organization functions over tim e in terms of communications and their codificat ion s, whereas the institutional organization provides structural coordina tion at each moment in time (Luhmann, 1984; Cowan & Foray, 1997). The ‘knowledge base’ can thus be consider ed as an overlay of mutual expectations that feedback on the institutional arra ngements among the knowledge organizers (Leydesdorff & Etzkowitz, 2001). Network arrangements provide the background for knowledge flows (Castells, 1996; David & Foray, 1994). In a knowledge-based ec onomy the institutional arrangements among knowledge organizers (e.g., univers ities, industries, and governm ental agencies) can become a necessary condition for both producing and reta ining wealth from knowledge (e.g., Popper & Wagner, 2002; Steinmueller, 200 2). Because of the potential overlap in networks at different levels, one can no longer expect th e organization of knowledge to be contained within a single organization. In other words, the ‘knowledge base’ of an economy generates a dynam ics orthogonal to that of the knowledge infrastructure of a political economy. The latter provides arrangem ents that stabilize the market system at each moment , while knowledge flows through these networks in fluxes with different speeds. The inte raction among the flows puts pressure on the previously established boundaries. For exam ple, pharmaceutical corporations can nowadays no longer carry the costs of biotechnologica l innovations without relying on knowledge networks (Owen-Smith et al ., 2002). Corporate boundaries increasingly function as mechanisms for the appr opriation and shielding of com petitive advantages from the knowledge fluxes through the networks. Knowledge-based innovations change the interfaces between supply-side agencies producing novelty (e.g., R&D) and—market or non-market—selection environm ents. In this process the relevance of previously defined boundaries can be redefined. When the new boundaries become functional for the reproduction of the system s, new retention mechanisms can also become institutionalized. Knowledge-based i nnovations can thus be considered as the evolutionary operators that ch ange the network structures in which they are reflexively generated (Fujigaki, 1998; Leydesdorff, 2001a). The knowledge-based economy In the period before the oil crises of the 1970s , that is, in the decad es after W orld War II, social functions were largely organized into institutions on a one-to-one basis (Merton, 1942; Bush, 1945). The global effects of the oil crises made clear that advanced industrial nations could outcompete low-wage countri es only on the basis of the syst ematic exploitation of their respective knowledge bases (e.g., Nels on & Winter, 1977, 1982; Freeman, 1982). Collaboration across instituti onal boundaries, however, im plies transaction costs (Williamson, 1975). The new relatio ns may also generate longer-term revenues and sy nergies (e.g., Faulkner & Senker, 1995). The transaction costs can be considered as a m acro-investm ent in establishing new structures of collaboration and competition at the nationa l level. Thus, a dynamic view of a knowledge- based system could be generated in which instit utional agents have continuously to trade-off among optimizations using a variety of criteria (Galbraith, 1967). The trade-off between short and long-term costs and benefits brought govern ments into play in th e interaction between R&D and the economy (OECD, 1971). Technolog ical innovation policie s were increasingly formulated (OECD, 1980). During the 1980s, the techno-sciences like bi otechnology, information technologies, and new materials rapidly became the top priorities for stimulation policies in th e advanced industrial countries. Because these ‘platform sciences ’ (Langford & Langford, 2001) are based o n rearrangements across disciplinary lines—that is, recom bination at the intellectual level— competitive advantages through synergies at th is level are to be exploited for economic development at the institutional level (A ndersen, 1994; Leydesdorff & Gauthier, 1996). 1 The reorganization and stimulation of university -industry relations at th e institutional level thus became a second point of attention for S&T policy makers (Rothwell & Zegveld, 1981; 1 Previous attempts at a more direct mission-oriented steering of the sciences ha d at that time been e valuated as less successful (Van den Dael e, Krohn, & Weingart, 1977 ; Studer & Chubin, 1980). 2 OECD, 1988). Why had some countries been m ore successful than others in exploiting their knowledge-base (Hauff & Scharpf, 1975; Irvine & Martin, 1984)? Why ha d within countries certain sectors (e.g., chemistry, aircraft) been mo re successful than others in m aintaining knowledge-intensive relations (N elson, 1982)? Could lessons be le arned from best practices across sectors and might such practices be transf erable from one national context to another? In the U.S.A. the national system experimented with granting universitie s the right to patents on the basis of federal funding (the Bayh-Dole Act of 1980), and systematic efforts w ere made to raise the level of knowledge-intensity wi thin industry both at the level of the states and by stimulation programs at the level of the federal governm ent (Etzkowitz, 1994; Spencer, 1997). Universities became increasing ly players in the patent system of the U.S.A. (Henderson et al ., 1998) Thus, their role as systemic knowledge organizers in innovation networks became increasingly important (E tzkowitz & Leydesdorff, 2000; Owen-Smith et al ., 2002). Figure 1 exhibits the percentage of patents that can be retrieved using th e word ‘university’ as a search term in the database of the U.S. Patent and Trade Office (at http://www.uspto.gov/ ). The second curve—of the percenta ge of universities that can be retrieved using ‘university’ as a search term among the assignees of patents— shows even more clearly that the effect of the Bayh-Dole Act began to peak in 1997. 0 2 4 6 1980 1985 1990 1995 2000 2005 % of US Patents 1981-2002 Re fe r e nce to 'U n i v ersi ty ' 'U n i v ersi ty ' as Assi g n ee Figure 1 Percentage of U.S. Patents (i) w ith a reference to the word ‘university ’ and (ii) a ‘university’ among the assignees In the period between 1981 and 1997, universities have thus been enrolled as new players in the patenting domain (Henderson et al ., 1998; Sampat et al ., 2003). But what does this indicator teach us with respect to the role of academic research in innovation processes (Rosenberg & Nelson, 1994; Cohen et al. , 2002)? Whereas this role can be analyzed historically for innovations on a case-by-case basis, a specificati on of the relevant system (s) of innovations is needed to determine this role at the aggregate level. 3 The delineation of systems of innovation The definition of a system of innovations in terms of nations, sectors (Pavitt, 1984 ), technologies (Carlsson, 2002; Carlsson & Stankiewicz, 1991), regions (Braczyk et al ., 1998), etc., brings players other than the traditiona l ones into view. From the mid-1980s onwards, for example, the European Union has develope d a series of Framework Program s containing new policies for science, technology, and i nnovation. Both trans-national cooperation and cooperation across sectors have be en systematically stim ulate d. Within the emerging context of the European system, regions have tried deliberately to promote their positions as a relevant level for the systematic developmen t of the knowledge infrastructure (Leydesdorff et al ., 2002) Has a European system of innovations emer ged in relation to the underlying national systems? Have regions (e.g., Catalonia, Flanders , etc.) b een successful in establishing their own specific systems of innovation (Riba-V ilanova & Leydesdorff, 2001)? Have sectors (e.g., ICT) been developed using patterns of i nnovation which differ from those established in the previous cycles of industrial devel opment (Barras, 1990)? How can systems of knowledge-based innovation be delineated a nd assessed if they cr oss national boundaries? These questions became ever more pressing durin g the 1990s when the Internet emerged. South- and East-Asian countries seemed initia lly better equipped for m oving ahead in the new e-environment given their specific mixes of human resources, flexib ilities in industrial structures, and prevailing knowledge infrastruc tures. How should the previously advanced industrial countries react ? Is it sufficient to stimulate ongoing processes of global change locally or should policy frameworks be propos ed that enable new partnerships to be developed at the global level? Which criteria for the optimization should then be used (e.g., national, transnational, sectoral)? In other wo rds, the stage was set for a deep reformulation of the very problem of science and technology policy-making in the first half of the 1990s. Science and technology policies in the 1990s A redefinition of the problem of science and technology policies became urgent as the Internet signaled its future economic success in the first half of the 1990s. The additional dimension of global communication could be envisaged as changing the p hase space of possibilities for internat ional collaboration in science and technology, international trade, and international relations. Structur al adju stments of existing arrang ements were likely to gain further momentum (Freem an & Perez, 1988). Gibbons et al . (1994) suggested making a distinction between ‘Mode 2’ and ‘Mode 1’ types of the production of scientific knowledge. Whereas ‘Mode 1’ refers to the trad itional shape, largely confined within institutional setti ngs, ‘Mode 2’ would be communication-driven. Knowledge can then be considered as a codification of communi cation. A scientific communication can be contained within an institu tion or even within an individual agent as ‘tacit knowledge’ and/or it can be ‘published’ and then brought into circulation. These dimensions of public and private communication of knowledge resonate with and disturb the establis hed public/private arrangements between industries and governments in the political economy. The knowledge component a dds a reflexive dynamic to the so-called 4 ‘differential productivity growth puzzle’ between various sect ors in the economy (Nelson & Winter, 1975). Existing trade-offs between public control and the private appropriation of competitive advantages can be expected to be increas ingly upset when innovations are systematically organized and stimulated (Nelson & Winter, 1977). New regulations (and perhaps new regulatory regimes) are needed when knowledge-based technologies restructure the sectoral organization (C allon, 1998). During the 1990s, in creased knowledge-intensity became thus a driver to the reform of political econom ies. In a number of papers, Henry Etzkowitz and I have proposed a neo-evolutionary m odel of a ‘triple helix of university- industry-government relations’ for these knowledge-based transformations of political economies (e .g., Etzkowitz & Leydesdorff, 1997 and 2000; Leydesdorff & Etzkowitz, 1998). As noted, three functions have to be fulfilled within a system of innovations: wealth generation in the econom y, n ovelty and innovation production that upset the equilibrium seeking mechanism s in market systems, and public control versus private appropriation at the interfaces between economic system s of exchange and organized novelty production. The knowledge infrastruc ture of university-industry-government relations (e.g., at the level of a nation state) can be consid ered as a specific retention mechanism among these three subdynam i cs. Th ese institutional arra ngements and their trajectories are under pressure fr om global developments that f unction as a next-order regime (Dosi, 1982). Advanced industrial states have historically generated ‘national systems of innovation’ during the past century or so based on the geogra phical proximity of the various subdynamics (Freeman, 1988; Lundvall, 1988, 1992; Nelson, 1993). The innovative knowledge flows, however, span boundaries and thus generate ne w types of comp etition at the global level (Krugman, 1996). In the Triple Helix model, this selection pr essure is represented as an overlay of communications among the institutional agencies whic h have hitherto carried the knowledge infrastructure: industry, academia, and governm ent. Each of these institutions is organized along international dimensions as well. At the level of the overlay of expectations, one can entertain and recombine possibilities other than thos e that have been realized hitherto. Thus, the linkages provide the carryi ng agencies with access to the knowledge-based system of coordination. 5 evolving knowledge bases mark et perspectives codifications in science and technology managerial and policy considerations knowledge-based innovations historical knowledge infrastructures university-industry- government relations; networks actor attributes knowledge claims, profit expectations, preferences Figure 2: The evolving knowledge-base feeds b ack on the historical knowledge- infrastructure. As the relative weights in the networks ch ange by ongoing processes of collaboration, appropriation, and competition, new balances and unbalances can be expected to generate feedbacks in the knowledge infrastructure at ot her ends. For example, trajectories can be formed historically at interfaces when technol ogies are ‘lock ed-in’ wi thin industries (e.g., the QWERTY keyboard or VHS tapes; cf. David, 1985; Arthur, 1989) or—s imilarly but between different subdynamics—when specific scientific expertise and government policies begin to co-evolve as they sometimes do in the en ergy and the health s ectors (Elzinga, 1985; McKelvey, 1996). The state and indu stry can also become ‘locked- in’ as in the former Soviet Union. Co-evolutions between two subdynamics con tinuously generate stabilities between counteracting mechanisms in processes of m utual shaping, whereas a third subdynamic potentially dissolves previous arrangements at a global level. The interacting subdynamics thus shape trajectories and regimes endogenously (Nelson, 1994; Leydesdorff & Van den Besselaar, 1998). Policies then have to vary accord ing to which ‘lock-ins’ can be expected to prevail, and whether and how they can be dist urbed. For example, the market mechanism can be expected to reintroduce flexibilities in th e case of a bureaucratic lock-in, whereas in the case of a technological lock-in government interventions may be needed to break monopolistic tendencies. Thus, the policies become increasingly a variable dependent on the evolutionary assessment of the knowledge-based system. Localizable trajectories and globalized regimes While the systems under study operate dynamically, knowledge flows between system s can temporarily be stabilized and furt her developed along the historical trajectories of institutions that have served the developments hitherto. Fo r example, the well-organized niches of nation 6 states can be considered as providing the stab ility that is nece ssary for accessing globalized— i.e., meta-stabilized—regimes (Luhmann, 2000, at p. 396). A global regim e can be expected to emerge from closer interactions between hith erto relatively separate subsystem s (Leydesdorff & Scharnhorst, 2003). The regime , however, contains a codification of the interactions among differently coded communi cation systems (e.g., the economy, science, and policy-making). The emerging configurations of mutual expectatio ns can be expected to change the selection pressure on the institutional ar rangements that were shaped historically. The networks supporting the exchange have then to be re structured (Freeman & Perez, 1988). Older arrangements can be expected to surv ive, but th ey may change in terms of their functions as well as their forms (Frenken, 2000; Kauffman, 2000). However, one should not reify the ‘global level agency’ as a metabiology or a supe rsystem. The various system s of expectations interact and produce an overl ay of global expectations within the network. The overlay globalizes the system by making representati ons available beyond the ones which could already be envisaged from the vari ous (subsystemic) perspectives. The envisaged recombinations can be attribu ted to a next-order or ‘gl obal’ system, but this evolution remains an internal dyn amics that is ad ded to the syst em as its globalization. This globalization can be entertained reflexively and thus enrich th e system. It provides a future- oriented knowledge-base that i nnovates the historically shaped structures with hindsight. The innovativeness is based on inventing new c odifications reflexiv ely by recombining perspectives at interfaces (e.g., betw een R&D and market perspectiv es). Thus, the dynamics of knowledge systematical ly organized as scie nce and technology induce a reflexive turn in social systems and, therefore, in the study of social systems. The ‘reflexive turn’ in science & technology studies (Woolgar & Ashm ore, 1988) first implied that the idea of a single and universalistic yardstick—as sought in the philo sophy of science (e.g., P opper, [1935] 1959)—had to be given up in favour of codes that can continuously be recombined and reconstructed. Unlike universal standards, however, asymmetry can be expected to prevail in exchange relations (Gilbert & Mulkay, 1984). The sy stems are able to exchange because they have different substances in stock. For example, the political system is interested in results from the science s ystem that inform decision making and legitimate polic ies, but wi thout being burdened with the overwhelm ing uncertainties that are intrinsic to scientific inference (Beck, 1 986). W ithin the science systems these uncertainties potentially raise new research questions. H owever, the science system can also develop in relation to problems arising in industrial cont exts. New possibilities to patent arise unexpectedly as external ities within the research pr ocess, whereas in other (e.g., industrial) contexts scientific progress can sometimes be considered as an unintended by- product when the focus was initially on the solution of production problems (Rosenberg, 1990). The interactive and non-lin ear dynamics in the developm ent of science, technology, and innovation also change professional pr actices. The new constellations drive the knowledge-based reconstructions of the political economies. 2 2 Reflexive mechanisms have increasingly been institution alized in advanced industrial systems since the scientific-technical r evolution of the p eriod 1870-1910 (Brav erman, 1974; Noble, 1977) . In the first stage, the reconstructions remained confined to the physical and chem ical properties of materials in the environment. More recently, this development has been reflected in sciences that reconstruct b iological and institutional systems (Fukuyama, 2002). 7 The study of knowledge-based communications Knowledge-based systems do not exist in terms of stable elements, but they develop in term s of operations. Operations, however, can be comb ined and recombined in a variety of ways. As noted, several authors (e.g., Lundvall, 1992 ; Nelson, 1993) have proposed considering ‘national systems of innovation’ as the appropr iate units of analysis for innovation studies. The choice of this national perspective allows for a direct link to the possibilities and limitations of policy making by national governm e nts, and it enables the researcher to use national statistics (Lund vall, 1988). From a reflexive angl e, however, an assum ption about a (national) communality in the data remains a hypothesis (Sko lnikoff, 1993; Andersen, 1994). For example, the notion of a national identity is nowadays historically changing from a European perspective. The subnational construc tion of regions has resounded in som e regions because of linguistic identities (e.g., Flanders, Ca talonia), but in other places (as in France) regional authorities had to be shaped in order to accomm odate European policies and harmonization. In other words, the units of an alysis and the systems of reference can be considered as constructs that pr ovide the analysis with a heuris tics. What is relevant from one perspective can be considered as contextual from another. Innovati ons and knowledge-based reconstructions occur by defin ition at interfaces and therefore allow for more than a single angle of theoretical appreciati on. Consequently, the categories in this reflexive field of studies have to be entertained re flexively, that is, as hypotheses. A second argument against reifyi ng one’s categories na turalistically follows from a reflection on how to declare the time axis in the research design. In contrast to a historical build up, the evolutionary dynamics continuously operate in the present and with hindsight, that is, upon the instantiations of the systems under st udy (Giddens, 1984). The addition of a virtual dimension to the system at th e Internet highlight s these evolutionary dynam ics. The global dimension tends to invert th e time axis in th e analysis by reconstruc ting the past from the perspective of other possibilities perceived in the present or more recen tly. Note that this development is only a tendency, since the gl obal developm ents remain embedded in historical ones. However, the re trospective view provi des us with an analytical angle for the construction of alternatives th at is knowledge-based, since it is no longer limited by what was already construc ted previously. For example, an analysis of the stren gths a nd weaknesses of a research portfolio does not by itself suggest that one should ‘pick the winners ’ in order to strengthe n one’s case globally, that is, at the system’s le vel (Irvine & Marti n, 1984). The ‘winners’ may have been yesterday’s winners and one may have other r easons to strengthen the hitherto r elatively weak groupings or clusters (P orter, 1990). Empirical anal yses inform us about the contingencies that can be expected; but since the dynamics are complex, unintended consequences and unforeseeable externalities are also expected (Callon, 1998). The evaluation provides us with sign als that can som etimes be made the sub ject of systematic analysis by taking another angle. 8 Operationalization How can one move from the analysis of knowle dge-based system s to a determination of the relative importance of the theo retical concepts in explaining an observable reality? How can a reflexive analyst make a convincing argument wh en the notion of a system of reference can always be deconstructed, and the time line ma y also be inverted in terms of what the historical accounts mean for the present? Since systems that contain and communicate kn owledge cannot be considered as given or immediately available for observation, one has to specify them anal ytically—that is, on theoretical grounds—before they can be indi cated and/or measured. To this end the quantitative measurement remains thoroughly de pendent on qualitative understanding. For example, one can raise the question of whether ‘Mode 2’ currently prevails in the production of scientific knowledge. What would count as a de monstration of this prevalence, and what as a counterargument? Can, for example, instances be specified in which one would also be able to observe processes of transition between the two modes? What should one m easure in such instances, and why? While qualitative analysis reduces the comple xity by taking a perspective, quantitative analysis allows us to raise questions about the extent to which a theoretical perspective highlights a relevant dimension. Can the curren t development of ‘biotechnology’ in Germ any be characterized as ‘Mode 2’? How can it be compared with ‘biot echnology’ in the United States? (e.g., Giesecke, 2000) A policy analyst may always be ab le to point to contingency, sameness and differences, continuities and change, but the quantitative analysis requires that these categories be specified as ex ante hypotheses so that the exp ectations can be updated by the research efforts. Empirical research enables us to specify the per centage of the variation that can be explained using one model or another. Whether ‘Mode 2 ’ is ‘old wine in new bottles’ (Weingart, 1997) or new wine in old bottles depends on the definitions of the bottles and the wines, and the processes of change that are analytically exp licated. In other words, the definitions of a knowledge-intensive system are them selves knowledge-intensive (Nowotny et a l ., 2001; Leydesdorff, 2001b). The observations and indica tors are also knowledge-intensive, since one can no longer assume that the overwhelmi ngly available information would answer the research questions precisely (Hicks & Ka tz, 1996). A crucial question becomes the theoretically informed specification of a se lection from the data. Which are the proper systems of reference? For example, what one understands nowadays under the name of ‘biotechnology’ is very different from what governments wanted to stimulate in the 1980s (Nederhof, 1988). Analogously, what industries subsume under the category of ‘biotechnology’ is different from what research cou ncils indicate with this sam e term. A modern society is pluriform and therefore differentiated in terms of its coordi nation mechanism s, codifications, and media of communication. The evolutionary perspective then demands an ex post delineation of the domains under study, but in the form of proposals and hypotheses. Thus, the indication of knowledge -based systems is based both on theoretical reflection and on methodological considerations about how one may be able to proceed from the specific choices to operationalizati on, and vice versa. For example, one can operationalize 9 ‘biotechnology’ in terms of a set of biotechnology journals in the Science Citation Index . If one fixes this journal set ex ante in order to make com parisons along the time line possible (Narin, 1976; Irvine et al. , 1985), one observes the devel opment of ‘biotechnology’ as conceptualized at the beginning of the data collection. If one defines the journal set dynamically, one studies the changing meaning of ‘biotechnology’ in rela tion to other journal clusters. If one determines the journal set ex post one refers to the most recently available understanding. The latter definition can be ma de relevant for policy, while the former definitions inform historical studies (Leydesdorff, 2002). In other words, the operationalization rema ins thoroughly dependent on the theoretical perspective. One looses the notion of indepe ndence of the external referent when studying knowledge-based systems from an evolutiona ry perspective (Luhmann, 2002). For example, using a journal set provides us with a focus on the scientific publication system . The use of patent data provides us with a focus on t echnological inventions. These two system s are differently codified and therefore can be expected to exhibit different dynamics. The methodological problems reflect decisions that have to be taken on analytical grounds. The theoretical consideration s, howev er, can only be made relevant for the measurement if they can be formulated as hypotheses th at are to be operationalized. Patent indicators In order to demonstrate my poi nt, let me provide data based on the U.S. national patent database, on the one hand, and based on the Inte rnet as a globally developing system, on the other. ‘University’, ‘industry’ , and ‘government’, and the vari ous combinations with Boolean ‘AND’ operators can be used as keywords in these databases. As above (see Figure 1), I searched the patent database for the num ber of occurrences of the terms in the file on a year- to-year basis. For reasons of comparison with the Internet searches (b elow), the time series is limited to the period 1993-2002. Table 1 first pr ovides the results of these searches. year University Industry Government UI UG IG UIG Total number of patents 1993 3063 9716 2619 401 588 334 63 110540 1994 3359 10568 2855 479 684 390 89 114564 1995 3710 10800 2828 529 771 410 93 114864 1996 4552 12147 3149 703 963 488 114 122953 1997 5406 12699 3604 814 1199 583 168 125884 1998 7623 17068 4708 1254 1658 807 266 166801 1999 8326 18553 4856 1352 1735 844 235 170265 2000 8488 19368 4831 1399 1776 865 267 176350 2001 9190 20812 5136 1591 1868 996 296 184172 2002 9228 21089 5242 1619 1928 1047 352 184531 Table 1 The number of hits for the sear ch term s ‘university,’ ‘industry,’ and ‘government’ and their combinations in the database of the U.S. Patent and Trade Office Note that these results do not indicate inte llect ual property or institutional relationship. The values for ‘university AND industry’ (UI), ‘university AND government ’ (UG), and ‘industry 10 AND government’ (IG) can be considered as indicator s of the bilateral links in the discursive domain of the database, wher eas the value of UIG represen ts the trilateral communality between these three concepts. In general, this type of data enables us to span a three-dimensional model as exhibited in Figure 3. IG U I UI Univ ersity Government Industry UG UI G G Figure 3 University-Industry-Government re lations in three dimensions As different from co-variation between two dim ensions or co-occurrence measurement, mutual information or transmission can be defined analytically in three dim ensions (Abramson, 1963). 3 Two states of a triple helix configuration can then be distinguished: in Figure 4 the three sets exhibit an overlap, whereas in Figure 5 this overlap has vanished. The mutual information in three dimensions (T UIG ) may become negative in the la tter case, while this indicator has a positive value in the f ormer. 3 The transm ission in three di mensions (x, y, z) can be defined as foll ows (Abramson, 1963, at p. 12 9): T(xyz) = Σ xyz P(xyz) log {[P (xy).P(xz).P(y z)] / [P(x).P(y).P(z).P (xyz)]} Or in anot her notation: T(xyz) = H(x) + H(y) + H(z) – H(xy ) – H(yz) – Hxz) + H(xyz) In the first formulation, P(x) stands for th e probability of an ev ent x and P(xy) for the probability that x and y occur together, etc. These probabilities can be measured by counting freq uencies of (c o-occurrences) of even ts, as will be shown in the empirical examples below. 11 I G UIG U UG IG UI I G U UG IG UI Figure 4 Three subsystems with a center of coordination Figure 5 Three subsystems without a center of integration When the three sets of documents containing the search term s ‘unive rsity’, ‘industry’, and ‘government’ are closely coupled by sharing a communality in the variation (e.g., in the case of neo-corporatist arrangements), the value of th is transmission is positive. W hen the three subsets are completely uncoupled, the mutual inform ation vanishes (T UIG = 0). However, when the three domains are operationally coup led through uncoordinated bi-lateral relations, the indicator can also become negative. Thus, this indicator provides us with a m easure for the state of a Triple Helix system whenever the dim ensions can be specified so that the relevant relations can be counted. Conceptually, the generation of a negative entropy such as mutual information corresponds with the idea of complexity th at is contained or ‘sel f-organized’ in a netw ork of relations that lacks central coordination. The system then pr opels itself in an evolutionary mode. The ‘global’ reduction of the uncertainty by the ne gative transmission is a result of the network structure of bi-lateral relations (Figure 6). 12 U UI I time G Figure 6 Three subsystems with hypercyclic in tegration in a globalized dimension The next-order structure operates globally by constraining and enabling local substructures. However, the overall structure cannot be perceive d completely from any of the positions in this networked system since there is no cen ter of coordination. However, an evolving structure in a virtual dimension can be hypothesized and then also be attributed a value using the algorithm as an indicator. The globalizable expectations remain embedded in the local situations, albeit in a distribut ed and therefore uncertain m ode. The em bedded uncertainties cannot be observed, but by using an algorithmic indicator one can appr eciate this latent structures of coordination. Figure 7 provides the value of T UIG for the time-series of pa tent data during 1976-2002. The figures show that co-occurrences be tween two of the three terms prevail to the extent that the value of T UIG is negative, but that the discourse in the U.S. national patent system becam e further integrated in terms of making references to univers ity-industry-governm ent relations during the 1990s. -0. 25 -0. 20 -0. 15 -0. 10 -0. 05 0. 00 1975 1980 1985 1990 1995 2000 2005 T(uig) Figure 7 The mutual information among ‘university,’ ‘i ndustry,’ and ‘governm ent’ relations in the database of the U.S. Patent and Trade Office. (The curve added depicts the two-year moving averages.) 13 As noted, the Bayh-Dole Act can with hindsight be considered as havi ng provided the patent system with one more degree of freedom , that is, by allowing universit ies increasingly to become players in this institu tional field (Sampat et al ., 2003). The patent system , however, is a system of legal control by a national gove rnment and therefore under the pressure of integration. New players can be e xpected to be enrolled within this discourse, but it takes time to reshape the m utual perspectives. 4 Webometric data Despite the poor operationaliza tion of the industrial dimension when using the word ‘industry’ as a search term, 5 the increasing integration in the patent datab ase is not a trivial result. This is demonstrated by the next tes t: Figure 8 is based on performing precisely the same exercise at the Internet using th e AltaVista Advanced Search Engine . 6 In this case, the mutual information in three dimensions d ecreases during the sec ond half of the 1990s. -0 .4 0 -0 .3 0 -0 .2 0 -0 .1 0 0.00 1992 1994 1996 1998 2000 2002 T( uig) title : fr eet ext Figure 8 The mutual information among ‘university,’ ‘industry,’ and ‘governm e nt’ relations as retrieved at the Internet using the Altavista Advanced Search Engine. (The curves are based on two-year moving averages.) In order to control for the eff ect of using these search terms without control, the searches were additionally conducted with the title-words . Title words are deliberately provided by the web-authors as meta-tags. Although the numbers of the retrievals are orders of magnitude smaller, the results are similar in exhibiting the trend. After a pe riod of initial construction of the system (1992-1995), the value of the indicator decreases steadily. This ‘self-organization’ of the Triple Helix relations in the gl obal dimens ion at the Internet seems to have flattened in the most recent years. Pe rhaps the flattening of the curve illu strates that the process of endogenous expansion of the Internet has been interrupted temporarily as the e-business has gone into a recession during these last years. 4 During the period 1976- 1992, T UIG had remained equal to – 0 .190 ± 0.008. 5 More than 50 % of the patent s contains an i ndustrial address (Ja ffe & Trai tenberg, 2002), w hereas only 10 -20% are indicated under ‘industry’ in Ta ble 1. 6 I used th e AltaVista Advance d Search Engine because this engine is unique in allowing searches with bot h Boolean ope rators and ti me delineations. For the met hodological proble ms involved i n using this t ool, see (Leydesdorff, 2001a). 14 year University Industry Government UI UG IG UIG “url:*” (total) 1993 2205 441 1041 49 49 46 25 18437 1994 12722 2178 3579 1007 1174 719 391 135265 1995 66719 13190 21187 5140 6861 4541 2036 640967 1996 216548 45938 66839 16257 21729 15894 6945 2308162 1997 478164 110434 166550 37122 51259 35230 16224 5740624 1998 842665 243611 343066 71306 95478 78922 32318 14379504 1999 1415659 471387 669844 131979 178892 157446 61899 33053057 2000 3005285 975976 1385296 245470 342218 298731 117318 86537251 2001 5381142 2419632 3014141 523922 724722 679407 247734 186175482 2002 10408179 7779754 7301276 1216090 1646210 1567669 550263 492815972 Table 2 The number of hits for the free-text search term s ‘university,’ ‘industry,’ and/or ‘government’ and their combinations using the AltaVista Advanced Search Engine (May 15, 2003) 7 Table 2 provides the data underlying this represen tation in a format sim ilar to that of Table 1. As in the case of the patent data, the changes are not apparent by visual inspection of the data. Unlike variables, the study of fluxes requires an algorithmic approach and the results can therefore be counter-intuitiv e. Note that the In ternet data are time-stamped in the p resent (in this case at May 15, 2003). As the Internet evol ves, previous representations are continuously overwritten. The search engines also change, using additionally their own reflexive dynamics (Leydesdorff, 2001a). While the words and title-words can be cons idered as variation, following the hyperlinks enables us to map the selecti ons that the authors of the webpages m ake from the materials previously made available. In this dimensi on the authors can be expected to integrate references into their text, whereas they are ex pected to reach out using w ords and title words (Leydesdorff, 1989). The AltaVista search engine enables us to map these hyperlinks to the relevant dom ains in terms of their institution al affiliation using the ex tensions ‘.edu’, ‘.com’, and ‘.gov’ as proxies. Note that these proxies ar e limited to the U.S.A. in the case of the .edu and .gov-domains, whereas the .com-domain is used worldwide. The resulting figure (Figure 9) shows a mirror im age to the cu rves exhibited in Figure 8. The selecting documents differentiate using their ow n codes in the present by using the selected documents as their knowledge-base. The knowledge ba se is thus integrated into the relatively stabilized instantiations taki ng part in the observable knowledge infrastructure (Giddens, 1984). 7 The se arch “ url:* ” provided a t otal of 1,50 4,185,772 hi ts using t he Altavista Advanced Search Engine on May 15, 2003. 15 -0 .3 0 -0 .2 0 -0 .1 0 0.00 1992 1994 1996 1998 2000 2002 T( uig) link: Figure 9 The mutual information among ‘link:.edu,’ ‘link: .com ,’ and ‘link:gov’ re lations as retrieved at the Internet using the Altavista Advanced Search Engine (15 May 2003). year .edu .com .gov edu AND com edu AND gov com AND gov edu AND com AND gov Link:* (total) 1993 721 753 26 32 16 21 13 140631 1994 10653 5969 5070 1281 454 1657 264 155429 1995 58559 85344 63208 16060 4168 30666 2707 971806 1996 185571 213755 40505 52853 13816 15191 9713 4215445 1997 383999 586804 76767 118249 25447 29842 18723 8410235 1998 714592 1512795 206683 177352 49238 59734 33695 21190676 1999 1410789 3372441 341635 346610 92354 126961 63192 42521722 2000 2212642 10057844 577433 622780 194573 244278 151641 92177426 2001 3722856 30497559 1328142 1344270 373437 599161 305180 196204140 2002 8564790 81698935 4035084 3058198 1159347 1758589 757120 501734312 Table 3 The number of hits for the search terms ‘link:university,’ ‘lin k:industry,’ and/or ‘link:government’ and their combinations using the AltaVista Advanced Search Engine (May 15, 2003) Table 3 shows the number of links involved. The nu mber of links with .com is an order of magnitude larger than the other ones and among the bilateral links the co-‘sitation’ of .edu and .com by the referring documents prevails (R ousseau, 1997). 8 Although the more skewed shape of these distributions considerably redu ces the values of the entropies involved, the interaction effect among the entropies is not visi ble on the basis of the values of the different fluxes without further computation. 8 The num ber of links to .c om pages is als o an order of m agnitude larger than t he number of hits using ‘industry’ as a free text term (Table 2). 16 The measurement of complex and codified communications What can the above pictures teach us? As note d, the word ‘university ’ can be expected to mean something different in a patent appli cation than on the Intern et. Furthermore, the meaning of a word may change over tim e. For example, it m ay have become more important for an applicant to make his or her collabo ration with a university visible in a patent application without necessarily implying that these collaborati ons did not exist previously. A pervasive problem with measurem ent in the case of complex dynami cs is that both the values of the variables and also the meanings of th e variables may change with the choice of the system of reference and over time. If one tries to m easure change in both the meanings of the variables and the values of the variables using a single design, the unders tanding tends to become confused because one loses a clear definition of a baseline (Studer & Chubin, 1980, at pp. 269 ff.). Knowledge-based developments cannot be equa ted with the developm ent of institutional units (Collins, 1985) or with fixed journal sets (Narin, 1976) . The evolutionary focus on flows of communications makes it necessary first to hypothesize what each system of communications is communicating when it operate s. For example, a system of references (citations, outlinks) can be expected to co mmunicate differently from a system of co- occurrences of words or a (re-)dis tribution of institution al addre sses. Citations relate papers along trajectories over time, wher eas institutional addresses of coauthorships, for example, can also be used for the mapping at specific m oments in time. The specification of a system of reference in terms of an operation —as different from a unit of analysis—extends the anal ysis with a reflection on the time horizon. In the historical dimension, I have elaborated above o n the issue of inverting the tim e axis because statisticians have been inclin ed to build on databases using a historical perspective. Historically interested sociolog ists and socio-con structivists share th is interest in temporal order in the materials: the quant itative data can then be used mainly as illustrations f or the narratives when ‘following the actors’ (Latour, 1987). The study of knowledge-intensive developments, however, requires us to take a refl exive turn towards the data gathering process, both in the quantitative and in the qualitative domains. The focus is no longer on the actors, but on the emerging order in their communications (U rry, 2000; Leydesdorff, 2002). Conclusion I have argued that a fundamental reformulation of the problems of sc ience, technology, and innovation policies became urgent during the 19 90s. Three models have been proposed for the study of innovation systems: (i) the distinction of a ‘Mode 2’ type of knowledge production, (ii) the model of ‘national systems of innovation,’ and (iii) th e Triple Helix model of university-industry-government relations. The authors of the ‘Mode 2’ thesis (Gibbons et al., 1994) have argued that the new configuration has led to a de-d ifferentiation of the relations between science, technology, and society. Internal codification mechanisms (lik e ‘truth-finding’) were discarded by these authors as an ‘objectivity trap ’ (Nowotny et al., 2001, at pp. 115 ff.). From this perspective, all scientific and techn ical communication can b e translated and compared with other 17 communication from the perspective of scien ce, technology, and innovation policies (Callon et al., 1986; Latour, 1987). In my opinion, the ‘Mode 2’ model focused on th e perform ative integration of representations of systems that are otherwise different and continuously also differentiating. The systems under study are asymmetrically integrated at the hi storical interfaces, for example, in the case of successful innovations. However, they can be expected otherwise to restore their own orders by differentiating again in terms of the specificity of their respective communication codes. This asymmetry of the differentiation is needed in order to perform a next cycle of integration. Differentiation and integration do not exclude one another, but rather depend on one another as different dimensions of the communicati on over time. A specific integration can be expected to mean something di fferent in the various dimensi ons that were integrated. The communication enables us to construct and so metim es stabilize an integration, but the underlying systems compete both in terms of their definitions of social realities and in term s of the representations that they construct at the localizable interfaces . Systems of innovations solve the puzzles of how to interface different functions in the communication. The solutions and failures are manifest at the level of historical organization. The latter can then also be reshaped. Evolutionary economists have argued in favor of studying ‘national systems of innovation’ as hitherto the most relevant le vel of integration. In deed, they have provi ded strong arguments for this choice (Lundvall, 1992; Nelson, 1993; Skolnikoff, 1993). However, these systems are continuously being restructured under the driv e of a global differentia tion of expectations. Economies are interwoven both at th e level of markets and in terms of m u ltinational corporations, sciences are organized internationally, a nd governance is no longer limited within national boundaries. The mo st interesting innovations ca n be expected to involve boundary-spanning mechanisms. In sum, I concur with the ‘Mode 2’-model in assuming a focus on communication as the driver of systems of knowledge production and control. However, the problem of structural differences among the communications and the or ganization of interfaces rem ains crucial to the understanding of innovation in a global and knowledge-based economy. The wealth of knowledge and options for further developmen ts have to be retained by reorganizing institutional arrangements with reference to global horizons. The Triple Helix model tries to capture both dynamics by introducing the notion of an overlay that feeds back on the institutional arrangem ents. Each of the helices develops internally, but they also inte ract in terms of exchanges of both goods and services and in terms of knowledge-based expectations. The vari ous dynam ics have first to be distinguished and operationalized, and then some tim es they can also be measured. I have argued that the dynamics among the dimensions can then be measured using algorithm ic indicators. The strength of this research program is that it is no l onger assumed to be possible to generalize on the basis of intuitions and naturalistic assumptions about the data. Em pirical results are expected to inform us, but the resu lts can also be counterintuitive. One may be able to appreciate unexpected results by innovating one’s theore tical assumptions about the 18 relevant systems of reference. If the various subdynamics can be better understood, one m ay also be able to develop simulation mode ls on the basis of the reconstructions. There is an intimate connection between th e algorithm ic evaluation of indicators and simulation studies. When analyzing knowledge-b ased system s, (scientometric) indicators enable us to study knowledge production and co mmunication in terms of the traces that communications leave behind, while simulations try to capture the operations and their possible interactions. The common a ssumption is that knowledge production, communication, and control are cons idered as operations that ch ange the materials on which they operate. The historically ob servable units of analysis ar e reflexively supplemented with units of operation that can only be specified on the basis of theoretical ass umptions. Because of this dependency on the theoretical specifi cation for its measurem ent, a knowledge-based economy can be expected to reinforce its development. References Abramson N. 1 963. Informat ion Theory an d Coding . New York, etc.: McGraw-Hill. Andersen ES. 1994. Evolutionar y Economics: Post -Schumpeteria n Contributio ns. London: Pint er. Arthur WB. 1989. Com peting Technologies, Increasing Ret urns, and Lock -In by Histori cal Events. Economic Journal 99 : 116-131. Barras R. 1990. Interactive Inn ovation in financial and business serv ices: The vanguard of the service revolution. Research Policy 19 : 215-237. Beck U. 1986 . Die Risiko-Gesellschaft . Fran kfurt a.M.: S uhrkamp. Braczyk H-J, Cooke P, Heid enreich M (eds.). 1998. Regi onal Innovation Systems. London/ Bristol PA: Univ er sity C olle g e London Pr ess. Braverm an H. 1974. Lab or and Monopol y Capital. T he Degradati on of Work i n the Twentiet h Century. Ne w York/London: Mon thly Review Press. Bush V. 1945. The Endless Frontier: A Report to the Presiden t . Reprinted, New York: A rno Press, 1980. Callon M. 1998. An essay on framing and overflowing: economic externalities revisited by sociology. In The Laws of th e Market . Mald en, MA, and Oxford: Blackwell; 244-269. Callon M, Law J, Rip A. (eds.). 1986. Mappin g the Dynamics of Science an d Technology. London: Mac millan. Carlsson B. (ed.). 2002. New Technol ogical Systems in t he Bio Industries -- an Inter national Study . Boston/Dordrech t/London: Kluwer Academic Publishers. Carlsson B, Stankiewicz R. 1991. On the Nature, F unction, and Compos ition of Technological Systems. Journal of Evolutio nary Economics, 1 (2): 93-118. Castells M. 19 96. The Rise of the Network Society . Malden, MA, and Oxford: Blackwell. Cohen WM, Nelson RR, Walsh JP. 2002. Links and Im pacts: The Influence of Pu blic Research on Industrial R&D. Management Scie nce 48 (1): 1-23. Collins HM. 1985. The Possibilities of Scien ce Policy. Social Studie s of Science 15 : 554-558. Cowan R, F oray D. (19 97). The Econom ics of Codification and t he Diffus ion of Kno wledge. Ind ustrial and Corporate Ch ange 6 : 595-622. David PA. 1985. Clio and the Economics of QWERTY. American Economic Review , 75 : 332-337 . Dosi G. 1982. Technol ogical Paradigm s and Technological Trajectories: A Suggest ed Interpretati on of the Determinants and Directions of Technical Change. Research Policy , 11 : 147-162. Elzinga A. 1985. Research, bur eaucracy and the drift of episte mic criteria. In The University Research System . Wittrock B, Elzinga A. (eds.). Sto ckholm: Almqvist & Wiksell; 191-217. Etzkowitz H. 1994. Techno logy Centers and Indu strial Po licy: the Emergence of the Intervention ist State in the USA. Science and Public Policy 21 (2): 79-87. Etzkowitz H, Leydesdorff L (eds.) .1997. Universities in the Global Knowledg e Economy: A Triple Helix of University-Industry-Government Relations . London: Pinter. Etzkowitz H, L eydesdorff L. 200 0. The Dynam ics of Innovat ion: From Nati onal Systems and ‘M ode 2’ to a Triple Helix of Univ ersity-I ndustry-Government Relations, Research Policy , 29 (22): 10 9- 123. Faulkner W, Se nker J. 1995. Know ledge frontie rs, public sector rese arch and i ndustrial innovati on in biotechnology, engineering c eramics, and p arallel comp uting. Oxford: Oxford University Press. Freeman C. 1982. The Economics of Industrial Innovation . Harmondsworth: Pengui n. 19 Freeman C, Perez C. 1988. St ructural crises of adjustment , business cycles and in vestment behavi our. In Technical Change and Econom ic Theory , Dosi G., Freem an C., Nelson R., Silve rberg G., Soete L. (e ds.). London: Pinter ; 38-66. Frenken K. 2000. A Complexity Ap proach to Innovation Networks. The Case of the Aircraft Indu stry (1909- 1997), Research Policy 29 (2): 257-272 . Fukuyama F. 2002. Our Posthuman Future: Political Consequences of the Biotechnology Revolutio n. New York: Farrar Straus & Giroux. Galbraith JK. 1967. The New Industrial State. Penguin: Harmondsw orth. Gibbons M, Li moges C, Now otny H, Schwa rtzman S, Scot t P, Trow M. 19 94. The new prod uction of knowledge: t he dynamics of science and rese arch in contem porary societie s. London: Sage. Giddens A. (1984). The Constitution of Socie ty . Cambridge: Polity Press. Giesecke S. (2 000). The Cont rasting Roles of G overnment in t he Developm ent of Biotechn ology Industry in the US and Germ any. Research Policy , 29 (2): 205-223. Gilbert GN, Mulkay MJ. 1984. Opening Pand ora’s Box. A Sociologic al Analysis of Scientist s’ Discourse. Cambridge: Cambridge Uni versity Press . Hauff V., Scharpf FW. 1975. Mod ernisierung der Volkswirtsch aft. Technologiepolitik als S trukturpolitik . Frankfurt a.M: EVA. Henderson R., Jaffe A., Trajtenberg M. 1998. Univers ities as a Source of Commercial Technolog y: A Detailed Analysis of University Patenting, 1965-1988. Review of Economics and S tatistics , 80 (1): 1 19-127. Hicks D, Katz JS. 1996. Science policy for a highly collaborativ e science system. Science and Pu blic Policy , 23 : 39-44. Irvine J, Ma rtin B, Peacock T , Turner R. 1985. Charting the decline of British science. Natur e , 316 : 587-590 . Irvine J, Ma rtin BR. 198 4. Foresight in Science: Picking the Winner s . London: Frances Pinter. Jaffe AB, Trajtenberg M. 2002 . Patents, C itations, a nd Innovations : A Window on the Knowle dge Economy . Cambridge, MA/London: MIT Press. Kauffman SA. 2000. Investigations. Oxford , etc.: Oxford Univ ersity Press. Krugman P . 1996. The Self- Organizing Economy . Malden, MA, and Oxford: Blackwell. Langford C, L angford M. 20 01. Funding St rategic Univers ity Research in Ca nada. Paper p resented at the Annual Meeting of the Society fo r the Social Studies of Science (4S), Boston, MA, Novem b er 2001. Latour B. 1987. Science in Actio n . Milton Keynes: Open University Press. Lewison G., Cunningham P. 1991. Bibliometric Stud ies for the Eval uation of Trans -National Resea rch. Scientometrics 21 : 223-244. Leydesdorff L. 1989 . Words and Co-Word s as Indicators of Intellectual Org anization. Research Po licy , 18 ; 209- 223. Leydesdorff L. 1995. The Challenge of Scientometrics: the developm ent, measure ment, and self-organization of scientific communications . Leiden : DSWO Press, Leiden University. http://www.upub lish.com/books/leydes dorff-sci.htm [1 June 2003] Leydesdorff L. 200 1a. Indicators of Innovation i n a Knowledge -based Econom y. Cybermetrics , 5 (1), Paper 2, at http://www.cindoc.csic.es/cybermetrics/articles/v5i1p2.html [1 June 2003] Leydesdorff L. 200 1b. Re-Thinki ng the Relati ons Between Text s and Context s in Science. Science & Publ ic Policy 26 (6): 484-486 . Leydesdorff L. 2001c. A Sociological The ory of Com munication : The Self- O rganizatio n of the Kn owledge- Based Society . Parkland, FL: Universal Publishers. http://www.upub lish.com/books/leydesdorff.htm [1 June 2003] Leydesdorff L. 2002. Dynamic and E volutionary U pdates of Classificato ry Schemes in Scie ntific Journal Structures. Journal of the Ameri can Society for Inf ormation Science and Technology , 53 (12): 987-994. Leydesdorff L, Van den Besselaar P. 1998. Technological Development and Factor Substitution in a Non-lin ear Model. Jour nal of Social and Evolutio nary Systems 21 : 173-192. Leydesdorff L, Cooke P, Olazaran M. 2002 . Region al Innovation Systems in Europe (Special Issue). Journal of Technology Tr ansfer 27 (1): 5-145. Leydesdorff L, Etzkowitz H. 1998. The Triple Heli x as a model for i nnovation st udies. Science and Public Policy 25 (3): 195-203 . Leydesdorff L, Et zkowitz H. 200 1. The Transform ation of U niversity-Ind ustry- Govern ment Relati ons. Electronic Jou rnal of Sociolo gy 5 (4), at http://www.sociology.org/content/vol 005.004/th.html [1 June 2003]. Leydesdorff L, Ga uthier É. 1996 . The Evaluat ion of Na tional Perform ance in Selected Priority Areas using Scientometri c Methods. Researc h Policy 25 : 431-450. 20 Leydesdorff L, Heimeriks G. 2001. The Self-Organizatio n of the European Informa tion Society: The c ase of ‘biotechnol ogy’. Journ al of the Americ an Society f or Information Sci ence & Technology 52 (1 4): 1262- 1274. Leydesdorff L., Scharnh orst A. (2003). Measuring the Know ledge Base: A Pr ogram of Inn ovation Studi es . Report to the German Bundesminister ium für Bildung und Forsch ung. Berlin: Berlin-Bran denburgische Akademie der Wissenscha ften; at http://sciencepolicystudies.de/ Leydesdorff&Sc harn horst.pdf [1 June 2003] Luhmann N. 1984. Sozial e Systeme. Grundriß eine r allgemeinen Theorie . Frankfurt a. M: Suh rkamp [ Social Systems (Stanf ord Universit y Press, Stanford ) 1995]. Luhmann N. 2000. Organi zation und Entscheidung. Opla den: Westde utscher Verlag. Luhmann N. 2002. The Mode rnity of Scie nce. In Theories of Distinction: Redescribing the Description of Modernity . Rasch W (ed.) Stan ford: Stan ford Univ ersity Press, pp. 61-75. Lundvall B-Å. 1988. Innovation as an interactive process: from user-producer interaction to th e national system of innovat ion. In Techni cal Change and Economic The ory. Dosi G, F reeman C, Nelson R , Silverberg G, Soete L. London: Pinter; 349-369 . Lundvall B-Å (ed.). 1992. National Systems of Innovation . London: Pinter. McKelvey MD. 1996. Evolutionary Innovations: The Business of Biotechnology . Oxford : Oxford U niversity Press. Merton RK . 1942. Science a nd Technology in a Dem ocratic Order. Journal of Legal and Political Sociology 1 : 115-126. Narin F. 19 76. Evaluative Bibliometrics: The Use of Publication and Cita tion Analysis in the Evaluation of Scientific Activity . Washington, DC: Nation al Science Foundation. Nederhof AJ. 1988. Changes i n publication p atterns of bi otechnologists: A n evaluation of the impact of governm ent stim ulation progr ams in six i ndustrial nations,. Scientometrics 14 : 475-485. Nelson RR (ed.). 1982. Government and Techni cal Progress: a cross -industry analysis . New York: Perg amon. Nelson RR (ed.). 1993. Nationa l Innovation Systems: A comparative ana lysis . New York : Oxford Univ ersity Press. Nelson RR. 1994. Economic Growth via the Co ev olution of Technolog y and Institutio ns. In Evolution ary Economic and Chaos Theory: New d irections in technology studies , Leydesdorf f L, Van den Besselaar P (eds.), Lon don/ New Yo rk: Pinter; 21-32. Nelson RR, Winter SG. 1975. Growth Theor y from an Evolutionary Perspective: The Differential Productivity Growth Puzzle. American Economic Review 65 : 338-344. Nelson RR, Winter SG . 1977. In Search of Useful The ory of Innovatio n. Resear ch Policy 6 : 35-76. Nelson RR, Winter SG. 1982. An Evolutionary Th eory of Economic Change Cam bridge, MA: Belknap Press of Harvard University Press. Noble D. 1977. America by Design . New Yor k: Knopf. Nowotny H., Scott P, Gibbo ns M. 2001. Re-Thinking Scie nce: Knowledge and the Public i n an Age of Uncertainty . Cam bridge, etc: Polity. OECD. 1971. Science, Growt h and Society: A New Perspective . Paris: OECD. OECD. 1980. Technical Change and Economic Po licy. Paris: OECD. OECD. 1988. Biotechnology and the Changing Role of Governm ent. Paris: OECD. Owen-Smith J., Riccaboni M, Pamm olli F, Powell WW. 2002. A Comparison of U.S. and European University- Industry Relations in the Life Sciences. Management Science 48 (1): 24-43. Pavitt K. 1984. Sectoral patterns of technical change: towards a theory and a taxono my. Research Policy 13 : 343-373. Popper KR. [1 935] 1959. The Logic of Scientific Discovery . London: Hutchinson. Popper SW, Wagner CS . 2002. New Fo undation for Growt h: The U.S. Innovatio n System Today and Tomor row (No. MR-1338.0-O STP). Washington, DC: RAND. Porter ME. 1990. Th e Competitive Advantage of Nations . London: Macm illan. Riba-Vilano va M., Leydesdorff L. 2001. Why Catalonia cannot be considered as a Regional Inno vation System . Scientometrics 50 (2): 215- 240. Rosenberg N. 1 982. How exogenou s is science? In Inside the Bla ck Box: Technology and Economics . Cambridge: Cambridge Uni versity Press; 141-159. Rosenberg N. 1 990. Why do fir ms do basic resear ch (with their o wn money)? Research Policy 19 : 165-174. Rosenberg N., Nelson RR. 1994. American universities and technical advance in ind ustry. Research Policy 23 : 323-348. Rothwell R, Zegveld W. 1981. Industrial In novation and Public Policy . London: Pinter. Rousseau R. 1997. Sitati ons: An Expl oratory Study. Cybermetrics , 1 (1), http://www.cindoc.csic.es/cybermetrics/articles/v1i1p1.html [1 June 2003]. 21 22 Sampat BN, Mowery DC, Sidonis AA. 2003. Changes in University Patent Quality after the Bayh-Do le Act: A Re-Examinat ion. International Journal of Indu strial Organization, forthco ming. Skolnikoff EB. 1993. Th e Elusive Transformation. Princeton, NJ: Princeton University Press. Spencer J-C. 1997 . Publicly supported non-defence R&D: the USA's Advan ced Technology Program. Science and Public Po licy , 24 (1): 42 -52. Steinmuel ler WE. 2002. Kn owledge-Base d Economi es and Inform ation and Comm unication Technologies. Internatio nal Social Sci ence Journ al , 54 (171): 141-153. Studer KE, Ch ubin DE. 1980. The Cancer Mission. Social C ontexts of Biomedical Research . Beverly Hills, etc.: Sage. Urry J. 2000. Sociolog y Beyond Societies: Mobilities for the Twenty-First Centu ry . London and Ne w York: Routledge. Van den Daele W ., Krohn W, W eingart P (eds.) . 1977. Geplante Forsch ung: Vergleichen de Studien übe r den Einfluss politischer Programme auf die Wissenschaftsentwicklu ng. Frankfu rt a.M.: Suhrkamp. Weingart P. 1997. From “Finalization” to “Mode 2”: old wine in new bottles? Social Scie nce Information 36 : 591-613. Whitley RD. 1984. The Intellectual and Soc ial Organizat ion of the Scienc es . Oxford: Ox ford University Press. Williamson OE. 1975. Markets and Hierarchies: Analysis an d Antitr ust Implica tions. A study in the Economics of Internal Organiz ation . New York: Free Press. Woolgar S, As hmore M (eds.). 1988. Knowledge and Reflexivity: New Fro ntiers in the Sociology of Knowledge . London: Sage. Wouters P, A nnerstedt J, Ley desdorff L. 1999. The European Guide to Science, Technolog y, and Innovation Studies . B russels: Eur opean Commi ssion, DG XII. http://www.chem.uva.nl/sts/guide/ [1 June 2003]. return