The Knowledge Base of an Economy - Semantic Scholar

The Knowledge Base of an Economy: What is it? Can it be measured? Can it be modeled? Loet Leydesdorff Science & Technology Dynamics, University of Amsterdam Amsterdam School of Communications Research (ASCoR) Kloveniersburgwal 48, 1012 CX Amsterdam, The Netherlands [email protected] ; http://www.leydesdorff.net

1. Introduction Hardly ever has a concept introduced by evolutionary economists been more successful in the political discourse than the one of a ‘knowledge-based economy’ (Foray & Lundvall, 1996; Abramowitz & David, 1996). This assumption of a qualitative transition in the economic conditions has become commonplace among policy-makers and mainstream economists (OECD, 1996). The abbreviation ‘KBE’ has even become a standard to the extent that it is often used without further explanation. For example, the European Summit on 23-24 March 2000 in Lisbon was specifically held “to agree a new strategic goal for the Union in order to strengthen employment, economic reform and social cohesion as part of a knowledge-based economy.” The conclusion of the meeting was, among other things, that “the shift to a digital, knowledge-based economy, prompted by new goods and services, will be a powerful engine for growth, competitiveness and jobs. In addition, it will be capable of improving citizens’ quality of life and the environment.”1 Can such a large impact on the real economy be expected from something so elusive and poorly defined as a knowledge-based economy? Or should one with hindsight consider this concept as a rhetorical reflection of the optimism during the second half of the 1990s about the potential impacts of ICT and the Internet (Godin, 2003)? In which respects can a knowledge-based economy be expected to differ from a market economy or a political 1

at http://www.europarl.eu.int/summits/lis1_en.htm#b .

economy? In this contribution, I shall argue that one can expect a knowledge-based economy to exhibit a dynamics different from a market-based or political economy because organized knowledge production and control adds a coordination mechanism to the social systems of economic exchange and political control. The expected interactions among these coordination mechanisms can be specified from an evolutionary perspective. 1.1

What is the knowledge base of an economy?

In their introduction to a special issue on the topic David & Foray (2002) more recently voiced a caveat against using the metaphor of a knowledge-based economy. They signal that the terminology was coined only recently and note that “as such it marks a break in the continuity with earlier periods, more a ‘sea-change’ than a sharp discontinuity. This transformation can be analyzed at a number of different levels” (ibid., p. 9). These authors argue that ‘knowledge’ and ‘information’ should be more carefully distinguished and they propose to analyze the development of a knowledge-based economy in terms of codification processes (Cowan & Foray, 1997). The main focus of most economic contributions, however, has remained on the consequences of the knowledge-based developments, including the effects of globalization on the relations among competitors and labour markets. The knowledge-based economy is thus mainly invoked as a factor explaining the historical developments and changes, but the evolutionary dynamics itself often remains unexplained. I do not wish to question the social relevance of the historical transitions and their impacts on the economy. On the contrary, my argument implies that a knowledge-based dynamics can be expected to provide mechanisms qualitatively different from the system out of which it emerged on the basis of interactions. The dynamic of knowledge production and control adds a degree of freedom to the complex system of social relations and coordination. In other words, I consider the knowledge-based economy as an explanandum that has to be explained rather than as an explanans for its possible implications. I will focus on the evolutionary mechanisms and the specific conditions under which a knowledge-based economy can be expected. In order to operationalize, model, and perhaps even measure the knowledge base of an economy we first have to flesh out the meaning of the concept. On the basis of a better 2

understanding, one is able to discuss among other things the question of whether and how the emergence of a knowledge-based economy is related to ‘globalization.’ Why and how can a knowledge-based economy be considered a driving force for social transformation? What may count as indicators of the knowledge base operating within an economy? After this first theoretical part with a focus on the specification, I will turn to the question of how the knowledge base of an economy can be operationalized and to whether a knowledge base can also be measured and simulated. In other words, it will be argued here that the concept of the knowledge base of an economy can be unpacked, and that this analysis results in an apparatus which provides us with a heuristics for empirical research and simulation studies. Organized knowledge production and control can be specified as a subdynamic of the socio-economic system (Noble, 1977; Whitley, 1984). The knowledge base operates in terms of informed expectations in the present. The expectations and anticipations drive a knowledge-based economy increasingly more than its historical conditions. In other words, representations of possible futures increasingly feed back on the historical process by inverting the time axis locally. However, in order to arrive at a model of anticipation as a mechanism of social coordination, we first have to define a knowledge-based economy more precisely as an evolutionary achievement. Let us follow David & Foray (2002) when distinguishing between information and knowledge. Knowledge operates by codifying the information or more precisely by codifying the meaning of the information within a system. Providing meaning to an uncertainty can be considered as a first-order codification. Knowledge enables us to discard some meanings and retain others in a second layer of codifications. Knowledge itself can also be codified. Codified knowledge can, for example, be commercialized. Thus, a knowledge-based system operates in recursive loops that are increasingly selective. A knowledge base of a social system can be shaped over time. The trading of knowledge emerged historically as different from the trading of commodities within the context of a market economy, that is, before the emergence of a knowledge-based economy. For example, the patent system can be considered as a typical product of industrial competition in the 19th century, but it was fundamental for regulating intellectual property thereafter. A knowledge market could increasingly be created in chemistry and then also in 3

electrical engineering (Noble, 1977; Van den Belt & Rip, 1987). Patents provide a format for codifying the knowledge contents for purposes other than the internal one of quality control within the scientific communication system. Patents package scientific knowledge so that new knowledge can function at the interfaces of science with the economy and, for example, be used in science-based innovations (Jaffe & Traitenberg, 2002). Schumpeter (1939) argued that the dynamics of innovation tend to upset the market mechanism (Nelson & Winter, 1982). While the market forces seek equilibrium at each moment of time, novelty production generates an orthogonal subdynamic along the time axis. Knowledge production and economic exchange relations can thus be considered as two analytically different dimensions that potentially interact in the case of innovations (Sahal, 1981). A third dimension relevant to the subject is the geographical—and potentially national—distribution of whatever is invented, produced, traded, and retained. For example, nation states can be expected to differ in terms of the mixes between the economy and their respective knowledge bases (Lundvall, 1992; Nelson, 1993). Given these specifications one can draw a model of the three relevant dimensions and their interaction terms as follows:

Knowledge Knowledge Infrastructure

Innovation

Economy

Political Economy

Geography

Figure 1. Three dimensions and their three first-order interaction terms 4

Figure 1 extends the conceptualization with the interaction terms between each two of the three dimensions. These interaction terms, however, are not synchronized ex ante. Therefore, they can be expected to generate an evolutionary dynamics in the historical system. For example, under the condition of knowledge production first being considered as a given exogenously to the political economy, nation states could be formed increasingly during the 19th century. Thereafter, under the condition of constitutional stability in the various nation states, national systems of innovation could be developed among the axes of economic exchange and institutionally organized knowledge infrastructures (Rosenberg, 1976 and 1982). In general, dynamics based on two subdynamics can be expected to co-evolve along trajectories when the third dynamic can be considered relatively constant. Two dynamics can over time lock-in into each other in a process of mutual shaping. However, the erosion of the stability in the third axis can be expected to change the systemic conditions in the longer run (Callon et al., 2002). When three dynamics interact, a more complex behaviour of the resulting systems can be expected. For example, a coevolution can then be broken by a bifurcation in a production process and a diffusion process.2 Although the historical progression varies among countries, the post-war stability in the national systems has changed more recently. When Lundvall (1988) proposed to consider the nation as ‘a first candidate’ for the integration of national innovation systems, he formulated carefully as follows: The interdependency between production and innovation goes both ways. […] This interdependency between production and innovation makes it legitimate to take the national system of production as a starting point when defining a system of innovation. (Lundvall, 1988: 362)

This assumption of a national level integrating innovation into production has an analytical advantage: the nation state provides us with a system of reference. This specification enables

2

The formal condition is that the diffusion parameter D is larger than the rate constant a divided by two. The one eigenvalue of the matrix representing the two coupled subdynamics becomes then positive, while the other is negative. A saddle point is thus generated and a bifurcation in the system is the consequence. (Rashevsky, 1940; Rosen, 1985, at pp. 183f.; Turing, 1952).

5

the analyst to delineate an empirical domain, for example, when studying the so-called ‘differential productivity growth puzzle’ (Nelson & Winter, 1975). This puzzle is generated by the different speeds of development among the industrial sectors because of innovations, but it cannot be studied without the specification of a system of reference (Nelson, 1982, 1993, 1994). Although the national level of integration undoubtedly still plays a major role in systems of innovation (Skolnikoff, 1993), the emergence of transnational levels of government like the European Union and the increased awareness of regional differences within and across nations have changed the functions of government (Storper, 1997; Braczyk et al., 1998). Governments have evolved from single (fixed) points of reference into variables with potentially a variety of levels. Larédo (2003), for example, argued recently that this polycentric environment of stimulation has become a condition for innovation policies in the Europe Union. The new configuration of three possible degrees of freedom—markets, governance, and knowledge production—can be captured by the model of a Triple Helix of universityindustry-government relations (Etzkowitz & Leydesdorff, 1997; Leydesdorff & Etzkowitz, 1998). In this model, the complex dynamics is indicated in terms of three relations as different from the dialectics prevailing in a political economy. Government can be considered as the variable that instantiates and organizes in the geographical dimension of the model, while industry is considered as the main carrier of economic production and exchange, and the university takes a major part in organizing the knowledge production function (Godin & Gingras, 2000). In an evolutionary model, however, these institutional dimensions can no longer be expected to correspond one-to-one to the functions carried by and among these agencies. Each university and industry, for example, has also a geographical location. The functions no longer develop only locally, that is, contained within the institutions. The interaction terms generate the evolutionary dynamics of change at the network level. University-industry-government relations, in other words, develop in terms of arrangements that recombine three functions of the socio-economic system: (1) wealth generation and retention, (2) novelty production, and (3) normative control at the interfaces between these two subdynamics. While the first two functions (economy and science) can be considered as 6

open and ‘universal’—and therefore continuously differentiating (Parsons, 1951; Luhmann, 1984, 1990)—normative control bends the space of possibilities reflexively back to the position of the operating units (e.g., the firms and the nations) in the market place and at the research front, respectively. In this dimension the question of what can be retained locally during the reproduction of the complex innovation processes becomes crucial. The knowledge becomes incorporated. The subdynamics develop functionally along the dimensions (eigenvectors) of the network, but the system needs also mechanisms for the integration and the retention. The knowledgebase of an economy can be considered as a second-order interaction effect in the historical trade-offs between these functions. This second-order interaction can also be considered as a resonance among the first-order interaction effects. However, the resonances can be expected to remain incomplete because they are disturbed by the interactions among the network carriers as the first-order retention mechanism. The knowledge base should therefore not be reified. However, overtones can sometimes be stabilized as innovations. As the innovations are further innovated, a knowledge base is increasingly available for the restructuring of the historically developed configurations. One can visualize the knowledge base as an overlay system that is added to the previous representation as follows:

7

Knowledge Knowledge Infrastructure

Innovation

Knowledge-based Economy

Economy

Political Economy

Geography

Figure 2. The first-order interactions generate a knowledge-based economy as a next-order system (from: Leydesdorff & Meyer, 2003). In other words, the carriers entertain a dually layered network: one layer of institutional relations in which they constrain each other’s behaviour and one layer of functional relations in which they shape each other’s expectations. From a systems-theoretical perspective the second-order interaction term can be considered a historical result of the first-order interactions depicted in Figure 1 under the condition that the various dimensions can be rearranged and reconstructed by the interactions. The dynamics of the system thus tend to shift from agent-based to network-based and then also knowledge-based innovations (Leydesdorff, 2003a). The additional degree of freedom between functional and institutional differentiation is needed for finding possible interactions among the interactions. The availability and strengthening of a knowledge base reinforces the capacity to develop solutions that improve on the combinations found hitherto among the three functions of wealth retention, novelty production, and control. The knowledge base is not a given, but a construct that emerges endogenously, locally, and in a distributed mode, and that is expected to remain under reconstruction. For example, it can be expected to change when the first-order interaction terms change importantly during historical transitions (like in Eastern Europe).

8

The interacting expectations, however, provide a basis for changes in the behaviour of the carrying agencies different from institutional imperatives and market incentives that have driven the system historically. While the institutions and markets develop along the time axis, the knowledge-based structure of expectations drives the system into an anticipatory mode. Planning cycles can become more important than current market trends. The informed anticipations increasingly change the dynamics of the system from a (locally oriented) agentbased perspective towards a more abstract knowledge-based one. One can model this as a change in the relative weights of the different subdynamics. The knowledge-based subdynamic has been reinforced during the last century by the social organization of knowledge production and control (Whitley, 1984). In other words, the institutional dynamics develop along historical trajectories, but the knowledge base can be expected to function evolutionarily as the technological regime of the same system (Dosi, 1982). The emerging regime is not phenotypically present as an observable variation, but it remains pending as selection pressure generated and reproduced by the lower-level interactions. The three subdynamics—which continue to develop recursively along their respective axes— are expected to interact in the complex dynamics of a knowledge-based economy as a historical phenomenon. The political (national and industry-based) economies have mainly stabilized an institutional infrastructure for trade, while the knowledge-based economy can be expected to expand as a cycle of communications among differently codified expectations at the network level. It operates by reconstructing the past in the present on the basis of representations (e.g., curves and functions).3 The emerging order of expectations can be accessed by reflexive agents. On the basis of the hypothesis of the operation of a knowledge base one is sometimes able to gain an understanding of the restructuring of the expectations at interfaces within the systems under study. For example, in a knowledge-based economy the price-mechanism of a market-based economy can increasingly be refined in terms of price/performance ratios and expectations about lifecycles of technologies (Galbraith, 1967; Heertje 1973).

3

While the first-order interactions led to functions that remained latent for the interacting agents (Lazarsfeld & Henry, 1968), these structural dimensions additionally operate over time by using a virtual dimension (Giddens, 1979).

9

As noted, trajectories can be stabilized whenever two of the three subdynamics co-evolve in a process of mutual shaping. The third subdynamic, however, potentially meta-stabilizes a knowledge-based innovation system into its global regime. For example, when a sector is innovated technologically, a ‘lock-in’ into a market segment may first shape a specific trajectory of innovations (Arthur, 1994). Learning curves can then be steep (Arrow, 1962). In other words, the trajectory follows an ‘up-hill’ search in a (hypothesized!) phase space of possible solutions (Allen, 1994; Ebeling & Scharnhorst, 2000; Kauffman, 1993). At the regime level of the phase space one can also compare different trajectories, but only by using a model (Scharnhorst, 1998). Analogously, when a science-based technology locks into a national state (e.g., in the energy or health sector), a monopoly can be immunized against market forces for considerable periods of time. Over longer periods of time, however, these ‘lock-ins’ can be expected to erode because of the ongoing processes of ‘creative destruction’ (Schumpeter, 1943). The creative destruction is based on recombinations of market forces with new insights and intuitions. These may lead to new bifurcations (destabilization; segmentation) and potential lock-ins (stabilization). The interaction effects, however, may also lead to global crises that require investments in restructuring the institutions (Freeman & Perez, 1988). 1.2

The operation of the knowledge base

The dynamics of a complex system of innovations based on second-order interaction effects are by definition non-linear. The non-linearity is a consequence of the interaction terms among the subsystems and the continuous recursivity in each of them operating at the same time. The non-linear terms, however, can be expected to outweigh the linear (action) terms in the longer run because of the higher exponents in the equations. For example, the interaction effects between ‘demand pull’ and ‘technology push’ become over time more important for the systemic development of innovations than the sum of the linear action terms (Kline & Rosenberg, 1986; Mowery & Rosenberg, 1979, 1989). Historically, interactions among the subdynamics were first enhanced by geographical proximity (for example, within a national context), but as the systems globalize, dynamic scale effects can become more important than the static ones for the retention of wealth. Such 10

dynamic scale effects through innovation were first realized by multinational corporations (Galbraith, 1967; Granstrand et al., 1997; Brusoni et al., 2000). However, they became a concern of governments in advanced industrialized countries after the oil crises of the 1970s (OECD, 1980). Improving the knowledge base of the economies of these nations became a priority during the 1980s because knowledge-based innovations could increasingly be considered as providing the competitive advantages of these economies (e.g., Rothwell & Zegveld, 1981; Freeman, 1982). In other words, the relatively stabilized system of a political economy (e.g., within a nation state) endogenously generates the meta-stability of a knowledge-based system when the nation states begin to interact more intensively. Under the condition of the need to restructure the institutional make-up of the national systems, recombinations at the functional level can induce ‘an oscillation’ of the system into its globalization.4 A system globalizes with reference to its next-order or regime level as an order of expectations. The knowledge base emerges by recursively codifying the expected information content of the underlying arrangements like in a bootstrapping operation (Berger & Luckman, 1966; Leydesdorff, 2001; Luhmann, 1984). Knowledge-based innovations can be considered as the historical carriers of this emerging system because the innovations instantiate interfaces (Fujigaki, 1998). The innovations change the systems in the present and potentially restructure existing interfaces from a hindsight (e.g., evaluative) perspective. For example, if one introduces high-speed trains, the standards for constructing the railways and the rails have also to be reconsidered. These materials may have to be innovated. Once in place, a knowledge-based system thus feeds back on the terms that went into its construction by offering improvements and advantages in comparison to the solutions hitherto found locally (e.g., on the basis of crafts and skills). Knowledge-intensity drives differentiation at the global level by providing us with a phase space of alternative possibilities, while the emerging system continues to operate locally in terms of institutions and solutions that organize and produce observable integration across interfaces.

4

The European Union and its Commission can perhaps be considered as catalysts by stimulating the interactions among universities, industries, and governance levels on non-national grounds.

11

The expectations are heavily structured and invested with interests because they are based on differentiations. Some authors (e.g., Gibbons et al., 1994; Nowotny et al., 2001) have claimed that the emergent system’s level (‘Mode 2’) exhibits a de-differentiation process between policy-making, economic transactions, and scientific insights because of their mutual ‘contextualization.’ The ongoing restructuring guided by the knowledge base, indeed, can be expected to induce new institutional arrangements in which sometimes industry can take the role of university, and vice versa (Etzkowitz et al., 2000). The exchanges among codified expectations, however, are expected to remain highly structured and reproducing the differentiation for evolutionary reasons. Knowledge-based systems would loose degrees of freedom and therefore competitive competencies by de-differentiation (Shinn, 2002). In other words, the complex system needs both integration in organizational formats (stabilization) and differentiation (globalization) in order to enhance further growth. The functions develop at the same time interactively and along their own axis (and on top of the exchanges among the institutions). At the interfaces among the codes (e.g., the price mechanism and the heuristics in R&D processes) translation mechanisms can be further developed that structure and codify the coevolutions. Since the systems of social coordination, communication, and control in a knowledge-based economy no longer provide a single frame of reference, both integration (for example, in action) and differentiation can be expected at the various interfaces. In dynamic terms differentiation and integration can be considered as two sides of the same coin: integration may take different forms and differentiations can be relatively integrated (as subsystems). The evolutionary question is where the relevant puzzles can be solved and where competitive edges can be maintained? The horizontal and vertical crisscrossing of systems and subsystems can be considered a hallmark of a knowledge-based economy. The definition of a system of reference itself becomes increasingly knowledge-based when the subsystems are differently codified, but yet interacting in the reproduction of the system. For example, one expects systemic integration at levels other than individual agency (e.g., in transnational corporations or institutional arrangements) and relative autonomy of functionally differentiated subsystem based on integration by codification (Luhmann, 1997). Governance of a knowledge-based economy can only be based on a set of assumptions that

12

are predictably in need of more informed revisions because one expects new formats to be invented at hitherto stabilized interfaces. 1.3

A micro-foundation of the Triple Helix model

Before I elaborate on the Triple Helix model in empirical terms, let me provide a microfoundation of the above conceptualization of the knowledge-based economy. Note that this micro-foundation cannot imply that the resulting phenomena can be disaggregated because the evolutionary model is based on the interaction terms. Instead of a static aggregation, however, a micro-foundation can in this case be considered as a nesting of operations that are expected to interact as different subdynamics. The subdynamics stabilize and globalize the systems selectively and in recursive loops along potentially different axes. As the system develops historically, the axes can be expected to rotate because of interactions and then also to become more complex. While the assumption of the homo economicus exclusively positioned the actor in the single dimension of the rational exchange, the systems view is grounded in actors that have both (geographic) positions and entertain exchange relations. Furthermore, the actors are reflexive. However, the reflection is historically later than the primary system’s operation in terms of positions and relations (Burt, 1982). The reflection can be considered as a second-order operation. The expected information content of the network arrangements remains uncertain until it is provided with meaning by reflexive agency or by a network of relations among such agents. In the latter case, the various meanings can also be recognized and selected among the agents.

13

Expected information value of the network arrangements

situational meaning; discursive knowledge

private meaning; tacit knowledge

Exchange relations (media of communication)

networks of nodes and links

Positions (persons; firms; countries)

Figure 3. Micro-foundation of the Triple Helix Model of Innovation In this configuration the rationality of the exchange is bounded because of the positional constraints, and the information is therefore incomplete (Alchian, 1950; Simon, 1955). The complex arrangement of positions and economic relations provides the historical network with an expected information value in a third dimension. However, this analytical dimension has still to be recognized (Maturana, 1978). Knowledge can be generated with reference to this axis because of the reflexive capacities organized along the other two axes. Recognition in either of these two dimensions—that is, incorporation of knowledge in agency or puzzle-solving capacity building in networks—provides the information with potentially different meanings. The two meanings can be differently codified because the respective axes are orthogonal. I indicated this in Figure 3 with ‘tacit meaning’ when the meaning is incorporated in the position of the reflecting system and I used the concept ‘situational meaning’ (from symbolic interactionism; Blumer, 1969) if the meaning is generated in the network of relations. Meaning provided at the network level can be further codified into discursive knowledge (Leydesdorff, 1998). Meaning provides the information with a first coding; knowledge can be considered as a meaning that makes a difference for the receiving system. Keep in mind that knowledge can

14

thus be generated and meaning be provided with reference to two different dimensions: the social exchange relations in the network among actors and their positions. While ‘structurally coupled’ (Maturana & Varela, 1980) these two systems of reference (networks and agents) are analytically different (Andersen, 1994; Leydesdorff, 2001; Luhmann, 1984). Knowledge production and control can thus become a social coordination mechanism in addition to being an embodied asset of agency (e.g., entrepreneurs). The organization of knowledge production in scientific discourses enables us to specify meaning with increasing precision at the supra-individual level and using a variety of disciplines. When this secondorder coordination mechanism is increasingly interfaced with the economy, a second-order dynamics can be expected to emerge. 2.

How can the knowledge base of an economy be measured?

When a system is no longer historically contained and integrated ex ante like in terms of, for example, a nation state, integration has to be considered a local effect of the system’s operation. Integration then emerges ex post and requires explanation. Lundvall’s (1988) assumption of an a priori integrating system is in this case no longer legitimate. Figures 4 and 5 illustrate the difference between these two models of integration:

ij

ij

ijk

i ik

Figure 4

il ik

Figure 5

15

il

Three subsystems emanating from a

Three subsystems can use a common network of mutual

common origin (i) and therefore with ex

relations; integration emerges ex post

ante integration in the origin While the configuration of Figure 4 contains a common zone (i) from which the subsystems emanate, Figure 5 exhibits an interaction among the subsystems at the level of the network of relations. In formal terms, the two configurations differ in the sign of the intersection: when the information content of the intersection in Figure 4 is considered as having a positive value, it would correspondingly have a negative value in Figure 5. In Figure 5 the network of relations reduces the uncertainty for the subsystems over time in a mode very different from the integration at each moment in time in Figure 4. 2.1

Next-order systems formation as a result of innovations at interfaces

The information transfer at the links can be formalized in terms of the mutual information or transmission T between the systems. It can be shown that the mutual information in a relation between two systems is always positive or zero (in the case of no relation), but never negative (Theil, 1972). The mutual information in three dimensions, however, can become negative if the bilateral relations form a network system (Abramson, 1963: 129 ff.). When the communality among the systems is no longer focused in a center, the dynamics of the system can increasingly be determined by the communication links among the subsystems. Note that such a system is no longer agent-based at the node, but communication-based in terms of the network of relations. The network among the nodes can subsequently develop an interaction term adding to the aggregate of the mutual relations. This overlay system may function increasingly as another subdynamic of the complex system. Let me visualize the working of this latent system of relations by penciling an overlay system into Figure 5. This leads to the following configuration:

16

ijk

ij

ik

time

il

Figure 6. Hyper-cyclic integration generates a negative entropy If all the relevant relations function at one time or another, the network can resonate into a hyper-cyclic integration (ex post). Note that this integration through the network relations requires time as a degree of freedom within the system. Adjustments among the three systems over time can lead to a surplus value over time—in terms of reduction of the uncertainty— when a next-order cycle can be induced. In economic theorizing, this degree of freedom added by the time dimensions has been expressed, for example, by Galbraith (1967: 178n) when he distinguished between growth maximization over time and profit maximization at each moment in time. How can a negative entropy that is produced at the network level, reduce the uncertainty within a complex system? For example, in a family a child can entertain relations with both its parents. But in addition to these two components of the expected information content of this family system and the common component in the aggregate (when the three are together, for example, at dinner time), the relation between the two parents can reduce the uncertainty for the child beyond its control. Thus, the latent structure of relations can sometimes reduce the uncertainty locally as a feedback mechanism on the historically unfolding events. In other words, the time dimension provides us with another degree of freedom for making an assessment about the integration of the system. This degree of freedom is declared in the Triple Helix framework as an overlay of expectations that continuously restructures the

17

underlying arrangements, but to a variable extent. The mutual information in three dimensions can be used as an indicator of this operation because it is based on entropy statistics. Unlike most other statistics, information theory provides us with a calculus which is able to indicate not only the state, but also the dynamics of the system(s) under study (Theil, 1972; Leydesdorff, 1995). 2.2

Operationalization in the terms of the Triple Helix model

In the case of university-industry-government relations, the mutual information T between two dimensions of the probability distribution (for example, in university-industry (UI) relations) is defined as follows: TUI = HU + HI – HUI

(1)

In this formula HU stands for the uncertainty in the distribution of the variable denoting the university, and HI for the uncertainty in the other dimension. The relationship reduces the total uncertainty in the relating systems (HUI) with the transmission (–TUI).5 It can be shown that the mutual information in three dimensions is equal to: TUIG = HU + HI + HG – HUI – HIG – HUG + HUIG

(2)

While each of the interacting systems (HU, HI, and HG) adds to the uncertainty, their interactions reduce the uncertainty at the system’s level by the mutual relations at the interfaces between them. As in the two-dimensional case, the systems condition and determine each other mutually in their relations. The three-dimensional uncertainty in the overlap (HUIG), however, adds again positively to the uncertainty that prevails at the network level. Because of the alteration in the signs the three-dimensional transmission can become negative when the reduction of the uncertainty by the bi-lateral relations prevails over the central coordination. Note that the value of TUIG can be calculated directly from the relative frequency distributions of the variables and their co-variation. 5

In other words, the two systems condition each other asymmetrically, while determining each other in the overlap symmetrically. This configuration can also be formulated as follows: HUI = HU|I + TUI + HU|I

18

How can one operationalize the indicator in relation to measurable variables? The evaluation can always be made if the three dimensions can be operationalized independently and if the covariations are also measurable. In other words, one needs not only information about how many universities collaborate with industry, but also more detailed information about which universities collaborate with which industries, and an independent operationalization of the role of the state. The distributions in two dimensions (e.g., university-industry) can then be tabled in matrix format. The margin totals indicate the information content in each dimension, while the cell values contribute to the mutual information. In all the covariations in the three dimensions of university-industry-government relations can be measured, one obtains a cube of information. If one extends this with the time dimensions (e.g., for different years) one obtains an hypercube of information which is not easily conceptualized, but which can be organized elegantly in terms of relational database management. The algorithms operate on the relations among the databases. In another context I elaborated several examples of the measurement of this mutual information in three dimensions using scientometric, webometric, and U.S. patent data (Leydesdorff, 2003b). Figure 7 provides one example, using the co-occurrences of the words ‘university,’ ‘industry,’ and ‘government’ retrieved using the AltaVista Advanced Search Engine for different years. The development of this data over time shows the increasingly negative value of the mutual information in these three dimensions over time. The pattern follows the globalization of the Internet after its initial phase of 1993-1995.

19

T(uig)

0.00 1992 -0.10

1994

1996

1998

2000

2002 title:

-0.20

freetext

-0.30 -0.40 Figure 7. The mutual information among ‘university,’ ‘industry,’ and ‘government’ relations as retrieved at the Internet using the Altavista Advanced Search Engine. (The curves are based on twoyear moving averages. The measurements were performed on 15 May 2003).

While the freetext words and title-words can be considered as variation generated by the authors to reach out in different dimensions, the hyperlinks enable us to follow the selections which the authors of webpages make from the materials already available to them. By acting selectively the authors can be expected to integrate references into their text (Leydesdorff, 1989). The AltaVista Advanced Search Engine enables us additionally to count these outgoing hyperlinks from the webpages to the relevant domains. Using the extensions ‘.edu’, ‘.com’, and ‘.gov’ as proxies, Figure 8 shows the mutual information among the cooccurrences. (Note that these proxies are limited to the U.S.A. in the case of the .edu and .gov-domains, whereas the .com-domain is used worldwide.)

0.00 1992

1994

1996

1998

2000

2002

T(uig)

-0.10 link: -0.20

-0.30

Figure 8. The mutual information among ‘link:.edu,’ ‘link:.com,’ and ‘link:gov’ relations as retrieved at the Internet using the Altavista Advanced Search Engine (15 May 2003). 20

Figure 8 shows a mirror image to the curves exhibited in Figure 7. The selecting documents use their specific content in the present for selecting the hyperlinked documents as their knowledge-base. The knowledge base is thus integrated into the instantiations taking part in the knowledge infrastructure observable as networks at the Internet. In summary, the observable stabilization can be provided with different meanings in terms of the dimensions that were considered relevant on theoretical grounds, i.e., as expectations. The instantiations organize the relevant dimensions as observable variation, but the dimensions select upon this variation in the different directions. These selection environments are not given—like in ‘natural selection’—but remain theoretically constructed. A knowledge-based innovation system can thus be hypothesized and tested in terms of the data (variation) deemed relevant to the selecting units of operation. 3.

Can the knowledge base (of an economy) also be simulated?

While the institutions develop historically and in relation to one another, they can under specific conditions develop configurations among them which recombine the expectations in a decentralized but networked mode. This overlay of expectations can operate as a feedback mechanism from a hindsight perspective, that is, against the time axis. In this latter case, it modifies the historical behaviour of the carrying agents by adding a negative entropy to the network among them. In other words, social systems can be expected to process expectations in addition to and on top of the historical exchange relations. The systematic exploitation of this additional dynamic increasingly generates a knowledge base within the socio-economic system. Note that the crucial point in my argument has been the generation and circulation of knowledge as a social coordination mechanism in addition to the generation of knowledge within each of the agents. Thus, the knowledge is not only positioned by a system of reference, but it has to be exchanged in relations among systems before social (i.e., relational) systems can become knowledge-based. The differentiation of knowledge production and control in two dimensions (see Section 1.3. above) provides the social system with one more degree of freedom for developing knowledge than systems which are only positioned. 21

The historically stabilized systems can be considered as niches that can be meta-stabilized when this degree of freedom is added. The meta-stabilization may lead to globalization on the side of the social system.6 Globalization, in other words, should not be considered as an indicator of the independent existence of a global ‘world system’ (Wallerstein, 1974), but as an indicator of the stabilization and potential globalization of systems of expectations. Thus, the globalizing dynamics are part of the differentiation in knowledge-based system, that is, the knowledge base provides the system with an internal representation of its next-order system. The representations, however, contain only a reference to a next-order or future system. Gobalization as another subdynamic within knowledge-based systems—other than, for example, stabilization—should not be reified by assuming the independent existence of a global system as an inference. Since the next-order system is expected to emerge at a next moment in time, the representation represents a state later in time than the system itself. Thus, the representations operate as an anticipatory feedback in the represented system of which they are at the same time a reflexive subroutine. This potentially inverts locally the time dimension. In the previous section, I discussed how a negative entropy can be produced in the relations among more than two subdynamics of the system. The measurement, however, remained at the system’s level. In this section, I shall specify the anticipatory mechanisms that stabilize and potentially globalize a system of expectations as the knowledge base of an economy. 3.1

Anticipation at the level of the social system

Knowledge can be considered as a further codified expectation. Scientific (or more generally, discursive) knowledge operates by reflexively entertaining a model of the system under study. However, the systems under study can only be accessed in terms of the understanding hitherto. This understanding is reflected, for example, in the scientific journal literature. The communicative operation in discursive knowledge generation and validation is thus recursive on previous communications. Scientific communication proceeds by improving the quality of

6

Whereas the reaction-diffusion dynamics can be considered as de-stabilizing the historical configurations in the longer run (with the time axis), the same mechanism operates as meta-stabilizing when considered evolutionarily (against the arrow of time).

22

the communication in recursive loops, but with reference to an external reality. However, this external reality is only represented in the communications. A communication system which operates in terms of improving an internal model, can be considered as an “anticipatory system” (Rosen, 1985). The reflexive angles enable us to explore alternative options in the system before realizing them. However, a system which contains a model of itself generates a paradox because the model also belongs to the system to be modeled, etc. This paradox has generated a lot of confusion in second-order systems theory (Kauffman, 2001). In my opinion, the paradox is only generated when one uses a timeindependent logic and/or a geometrical metaphor. When time is declared as a degree of freedom, the paradox can sometimes be solved by using an algorithmic approach. For example, ‘computing anticipatory systems’ has emerged recently as another specialty in computation and applied mathematics (Dubois, 2000). An anticipatory system can contain a prediction of the state of the system at t+1 while the system on which the model reflects still dwells in the present time t. Thus, the two subdynamics of the system (anticipation and reproduction) differ in terms of their respective clock speeds. If the reflexive model operating at t+1 is available as a subsystem within the system in the present, a feedback arrow can operate from t+1 on the state of the system at t. Dubois (2000) considered this as ‘weak anticipation.’ Mental models, for example, provide us with means to anticipate on changes in the environment. Social systems, however, can be considered as ‘strongly anticipatory systems’ because they contain one more subdynamic of anticipation than weakly anticipatory systems. In addition to containing weakly anticipatory subsystems that can make informed predictions (i.e., agents and/or models), the interaction among the various subdynamics of a social system can be expected to generate a feedback on the time axis that can be considered as an anticipation. In the previous section, this feedback was indicated as a negative entropy, but let us now try to specify the mechanism. The crucial point is that these subsystems are not ex ante synchronized and therefore time provides us with a degree of freedom. When the subsystems are no longer ex ante synchronized, each subsystem windows upon the others using a representation of the other subsystem in its own code. For example, when a 23

new technology is introduced in the market, the technology can be considered as new with reference to its previous version at t-1, that is, along the historical axis, but it experiences a feedback from the market in the present (at t). One can use this difference of a time-step ∆t for modeling the social system as an anticipatory system (Leydesdorff & Dubois, 2003). The subsystems update in relation to one another by interacting on the basis of representations of each other. Thus, the complex social system contains subsystems which model one another in the present, while the subsystems continue to develop recursively with reference to their respective pasts. A strongly anticipatory system not only anticipates, but it is able to construct its own future like in a techno-economic evolution. The two anticipatory mechanisms can operate upon each other in a coevolution. The social system is strongly anticipatory because it contains two anticipatory mechanisms that may coevolve: (a) the weak anticipation contained at the level of the carrying systems who provide meaning to the observable phenomena, and (b) the structural anticipation at the interfaces among subsystems of the social system. As noted, the second type of anticipation emerges only under the condition of functional differentiation among subsystems within the social system (Parsons, 1951; Luhmann, 1984). It is communication-based unlike the agency-based reflections in the first form of anticipation. 3.2

Anticipatory modeling

The growth of a technology under selection pressure, for example, can be modeled using the logistic equation. The discrete formulation of this equation reads: x(t) = a * x(t-1) * [1 – x(t-1)]

(3)

As a system becomes larger the growth experiences a feedback term [1-x(t)] increasingly. (In the logistic map, the parameter a is well-known to generate progressive bifurcations at higher values leading eventually to chaos at a = 4.) An anticipatory formulation of this same equation can be formulated as follows (Dubois, 1998): x(t) = a * x(t-1) * [1 – x(t)]

(4)

24

The system x(t) still refers to its previous state x(t-1) historically, but it experiences a feedback from the environment [1 – x(t)] in the present. As noted, such a model is more realistic in the case of the market introduction of a new technology than the alternative of assuming feedback from a previous configuration in the market. The anticipatory character of Equation 4 can be made explicit by moving one time-step ahead, i.e., by writing the same equation as follows: x(t+1) = a * x(t) * [1 – x(t+1)]

(5)

In general, if the two subdynamics—in this case, technology development and market selection—are differently coded one expects their relation to function as a feedback in the present and not in the past. Each of the subsystems is updated in terms of its own historical development and using its own (symbolically mediated) code. The updates in the two systems are not synchronous ex ante (for example, within a single organization), but synchronized by the interaction ex post. Each subsystem contains in its historical coding an expectation of the other based on previous interactions which is updated by the next interaction. The asynchronicity among the subsystems provides the emerging system with time as a degree of freedom for the integration. This degree of freedom can be exploited by reflexive agencies at interfaces using both their own axis for the weak anticipation and additionally the time-differences among subsystems for the generation of strong anticipation, that is, the knowledge-based reconstruction of the social system. How can anticipations be simulated? The anticipatory formulation of Equation 5 can be written as an endless recursion by replacing x(t+1) in the right-hand term of the equation with the entire formula: x(t+1) = a * x(t) * [1 – x(t+1)]

(5)

x(t+1) = a * x(t) * {1 – a * x(t) * [1 – x(t+1)] }

(6)

etc.

25

This generates the above noted paradox in the description at each moment in time. However, the series has also an analytical solution as can be seen from the following rewrite: x(t+1) = a * x(t) * [1 – x(t+1)]

(5)

x(t+1) = a * x(t) – a * x(t) * x(t+1) x(t+1) + a * x(t) * x(t+1) = a * x(t) x(t+1) * [1 + a * x(t)] = a * x(t) x(t+1) = a * x(t) / [1 + a * x(t)]

(7)

This model can be simulated and the results show properties very different from the logistic map. For example, this map does not exhibit chaotic behaviour at a = 4. In other words, the introduction of an anticipatory selection mechanism dampens the entropy production. In general, a selection mechanism operates ex post on the variation. While one is able to observe the variation phenotypically, one can expect a selection mechanism to have operated when one observes a non-random variation. As noted, when the selections are no longer provided ‘naturally’ like in biology, these mechanisms are hypothesized. The anticipatory model, however, provides us with an opportunity to study the interaction between variation and selection with different parameters by using the difference ∆t between the expectation and the observation along the time axis, but in a backward mode.7 The backward mode enables us to analyze the evolutionary mechanisms of selection and globalization, while models that follow the time axis in a forward mode exhibit the historical variations and stabilizations (e.g., Malerba et al., 1999). Dubois (2001, 2002) used the anticipatory formulation of the logistic equation to show that in the case of synchronization an anticipatory system is able to predict the behaviour of a complex and even chaotic dynamic when the latter contains a systematic delay τ. The anticipation can then be synchronized with a state of the system which is not yet manifest. The author used the spread of an epidemic as the example. However, the development of a technology before entering the market can analogously be predicted when systemic delays—

7

The anticipatory model is possible because the discretization of the differential equation can be based on the forward or the backward evaluation along the time axis (Leydesdorff, 2003c).

26

for example, in terms of its development phases—can be expected. Figure 9 provides a result of this simulation.

Figure 9. Anticipatory synchronization of I*(t) and I(t) with delay τ = 50 (Dubois, 2001: 15) Unlike time lines, cellular automata enable us to visualize the interactions among subsystem spatially. The lower half of Figure 10, for example, shows a recursively evolving system that develops over time. The development in this lower half is represented in the upper half of the screen using the anticipatory formulation of the logistic equation (Equation 7 above). As expected, the representing system contains a model of the represented system. In the language of second-order systems theory one might formulate that an observer is generated within the system under study (Brown, 1969; Maturana, 1978; Luhmann, 1995).

27

Figure 10. The top-level screen produces a representation of the bottom-level one by using an anticipatory algorithm In other words, the top-level screen selects in the present on the variation previously produced in the system represented at the bottom-level screen. Thus, the selection mechanism can be made visible by the simulation. What seemed to have become volatile as a mechanism that could only be hypothesized on theoretical grounds, can be studied systematically in terms of its operation over time by using the algorithmic approach of simulations. The visualization, for example, illustrates that not all properties of the represented system (e.g., the technology) are relevant for its reflection in the representing system (e.g., the market). However, the ‘technical characteristics’ remain relevant for the further development of the technology along the time axis (Frenken, 2001). 3.3

Strong anticipation and technological determination

When two structurally different anticipations can operate upon one each other, the subdynamics of ‘representing’ and ‘represented’ can be considered at each moment as orthogonal axes informing each other through the covariation in terms of their mutual 28

information (Figure 11). When this co-variation is repeated over time, a coevolution between the two subdynamics may lead to a ‘lock-in.’ The representing/represented dimensions can then begin to alternate like in a coevolution. But when the emerging coevolution develops further and stabilizes into a third dynamic, one expects non-linearities and therefore the emergence of the species of chaotic behaviour (e.g., crises). Thus, the coevolving system construct the next-order (three-dimensional) stage as a bifurcation. As noted, a negative entropy can be a consequence of the mutual information in three dimensions.

representing ↑

representation

represented → stabilization of systemness with coevolution over time

Figure 11. Evolving systemness in a representation using time as another dimension The production of a new system’s dimension can be considered as a so-called diffusionreaction mechanism.2 The crux of this rather technical argument is that the two (sub)systems can first ‘lock-in’ into each other (like in a coevolution) when developing increasingly over time. However, the co-evolution of two subsystems becomes unstable as the diffusion parameter becomes larger. Thus, while the two systems couple first in a coevolution, the expectation is that a bifurcation will occur and a third axis will eventually become independent.8 In other words, the coevolution is unstable in the longer run, but it can be 8

In formula format: f(xy) = a xα + byβ + c(xy)γ

29

considered as a stabilization during considerable stretches of time. Thus, the organizational layer provides the next-order system with mounting blocks that will be left behind when the system further develops. These expectations have still to be further elaborated in a program of simulations (Leydesdorff, 2001, 2002, 2003c; Leydesdorff & Dubois, 2003). In general, anticipation has a function in suppressing chaotic behaviour in interactions by synchronizing subdynamics (Kampmann et al., 1994). Strong anticipation reinforces the subdynamics of anticipation to the extent that one is able to reconstruct the systems under study. Thus, the techno-economic evolution can increasingly become knowledge-based. From this perspective, the technological artifacts can be considered as provisional stabilizations of an evolving system of expectations. 4.

The complex dynamics of a knowledge-based economy

Three functionally different subdynamics were specified as spanning a knowledge-based innovation system by interacting: (1) economic exchanges on the market, (2) geographical variation, and (3) the systemic production of novelty based on the organization of knowledge. Along these axes differentiations continuously expand the system because of ‘lock-ins’ and consequently emerging dimensions. Various forms of integration have historically been organized at the interfaces. For example, a political economy interfaces the economy within the geographical domain of a nation state (Nelson, 1993). Organized knowledge production, however, continuously upsets the historical arrangements using the mechanism of codified (and potentially counter-intuitive) expectations. In principle, dissemination of organized knowledge in (semi-)markets can generate wealth, but this global process has to be retained locally (Krugman, 1996). During the 20th century the knowledge production system became increasingly organized and controlled. If the interaction term (xy) can recursively be stabilized, for example, because of coevolution and/or lock-in, a trajectory can be shaped along a third axis z. While initially developed as a function of the interaction (xy), this axis can increasingly become orthogonal to the interacting systems x and y. Thus, the system can endogenously gain a degree of freedom. When this happens, all mutual informations among x, y, and z can be declared in the next stage as dimensions of a new (and more complex) system. The emergence of a new dimension in the system dimension can be assessed using the Markov property, while as argued above the self-organization of the newly emerging system can be measured in terms of the mutual information among the three axes (Leydesdorff, 2003b; Leydesdorff & Van den Besselaar, 1998).

30

Furthermore, the social subsystem of knowledge production and control was interfaced with the economy and with government, to the extent that the upsetting forces of innovation could no longer be contained within the institutional arrangements of a political economy constructed in a previous period. Structural adjustments are driving the system into a knowledge-based economy as a next-order configuration. A knowledge-based economy is different from a political or a market economy in entertaining a horizon of expectations with reference to relevant (e.g., global) environments. The expectations are endogenous to the networked systems. They can be expected to develop along the eigenvectors of the network among the carriers and sometimes to resonate into new solutions. The next-order level of development provides the carrying systems with their respective knowledge bases. A knowledge base cannot be observed, but it can only be expected. Entertaining this hypothesis, however, may enable us to improve on the expectations. The dynamics in the resulting knowledge base were first shown in terms of the production of a negative entropy as a result of the interaction among subdynamics. A negative entropy can be measured when the selecting systems of reference are theoretically specified. The mechanism of this feedback on the time axis can be studied using the model of anticipatory systems. Anticipations operate as a selection mechanism at the next-order level by enabling the system(s) to filter the variations that are generated historically. From the empirical perspective, this selection mechanism could only be hypothesized. The simulations, however, enabled us to specify how the selection mechanisms can operate as a feedback. When socially organized, the feedback can grow so strong that it provides sufficient stability for inducing historical change. A knowledge-based system gains this degree of freedom along the time axis that can be exploited as a competitive advantage. Since the algorithms are nested—and not aggregated—the micro-foundation can no longer be formulated in terms of observable units of analysis. One has to account for the interactions at the network level, for example, among the links in niches. The niches are communicationbased—as different from agent-based—and these communications can endogenously become more knowledge-based under specifiable conditions of cultural evolution (e.g., functional differentiation). The functional subdynamics at the level of the eigenstructure of the network 31

can gain importance over time when compared with the institutional subdynamics at the level of the observable relations because of the potential to further develop endogenously. The mechanism of further development is the recursivity in the operation of codification. Codifications can thus stabilize and then—because of the expectation of bifurcation among layers—globalize bodies of knowledge as a basis for economic development. References Abramowitz, M., & P. A. David. (1996). Measuring Performance of Knowledge-Based Economy. In Employment and Growth in the Knowledge-Based Economy (pp. 35-60). Paris: OECD. Abramson, N. (1963). Information Theory and Coding. New York, etc.: McGraw-Hill. Alchian, A. A. (1950). Uncertainty, Evolution, and Economic Theory. Journal of Political Economy, 58, 211-221. Allen, P. M. (1994). Evolutionary Complex Systems: Models of Technology Change. In L. Leydesdorff & P. v. d. Besselaar (Eds.), Evolutionary Economic and Chaos Theory: New Directions in Technology Studies (pp. 1-18). London/ New York: Pinter. Andersen, E. S. (1994). Evolutionary Economics: Post-Schumpeterian Contributions. London: Pinter. Arrow, K. J. (1962). The Economic Implications of Learning by Doing. Review of Economic Studies, 29, 155-173. Berger, P. L., & T. Luckmann. (1966). The Social Construction of Reality: A Treatise in the Sociology of Knowledge. Garden City: Doubleday. Blumer, H. (1969). Symbolic Interactionism. Perspective and Method. Englewood Cliffs: Prentice Hall. Braczyk, H.-J., P. Cooke, & M. Heidenreich (Eds.). (1998). Regional Innovation Systems. London/ Bristol PA: University College London Press. Brown, G. S. (1969). Laws of Form. London: George Allen and Unwin. Brusoni, S., A. Prencipe, & K. Pavitt. (2000). Knowledge Specialization and the Boundaries of the Firm: Why Do Firms Know More Than They Make? Administrative Science Quarterly, 46, 597-621. Burt, R. S. (1982). Toward a Structural Theory of Action. New York, etc.: Academic Press. Callon, M., C. Méadel, & V. Rabeharisoa. (2002). The Economy of Qualities. Economy and Society, 31 (2), 194-217. Cowan, R., & D. Foray. (1997). The Economics of Codification and the Diffusion of Knowledge,. Industrial and Corporate Change, 6, 595-622. David, P. A., & D. Foray. (2002). An Introduction to the Economy of the Knowledge Society. International Social Science Journal, 54 (171), 9-23. Dosi, G. (1982). Technological Paradigms and Technological Trajectories: A Suggested Interpretation of the Determinants and Directions of Technical Change. Research Policy, 11, 147-162. Dubois, D. M. (1998). Computing Anticipatory Systems with Incursion and Hyperincursion. In D. M. Dubois (Ed.), Computing Anticipatory Systems, Casys-First International Conference (Vol. 437, pp. 3-29). Woodbury, NY: American Institute of Physics. Dubois, D. M. (2000). Review of Incursive, Hyperincursive and Anticipatory Systems -Foundation of Anticipation in Electromagnetism. In D. M. Dubois (Ed.), Computing 32

Anticipatory Systems Casys'99 (Vol. 517, pp. 3-30). Liege: Amercian Institute of Physics. Dubois, D. M. (2001). Theory of Incursive Synchronization and Application of a Chaotic Epidemic. International Journal of Computing Anticipatory Ssystems, 10, 3-18. Dubois, D. M. (2002). Theory of Incursive Synchronization of Delayed Systems and Anticipatory Computing of Chaos. In R. Trappl (Ed.), Cybernetics and Systems (Vol. 1, pp. 17-22). Vienna: Austrian Society for Cybernetic Studies. Ebeling, W. & A. Scharnhorst (2000). Evolutionary Models of Innovation Dynamics. In D. Helbing, H. J. Herrmann, M. Schreckenberg, D. E. Wolf (Eds.), Traffic and Granular Flow ’99 – Social, Traffic, and Granular Dynamics (pp. 43–56). Berlin: Springer. Etzkowitz, H., & L. Leydesdorff (Eds.). (1997). Universities in the Global Knowledge Economy: A Triple Helix of University-Industry-Government Relations. London: Pinter. Etzkowitz, H., A. Webster, C. Gebhardt, & B. R. C. Terra. (2000). The Future of the University and the University of the Future: Evolution of Ivory Tower to Entrepreneurial Paradigm. Research Policy, 29 (2), 313-330. Foray, D., & B.-A. Lundvall. (1996). The Knowledge-Based Economy: From the Economics of Knowledge to the Learning Economy. In OECD Documents: Employment and Growth in the Knowledge-Based Economy (pp. 11-32). Paris: OECD. Freeman, C. (1982). The Economics of Industrial Innovation. Harmondsworth: Penguin. Freeman, C., & C. Perez. (1988). Structural Crises of Adjustment, Business Cycles and Investment Behaviour. In C. F. Giovanni Dosi, Richard Nelson, Gerald Silverberg, and Luc Soete (Ed.), Technical Change and Economic Theory (pp. 38-66). London: Pinter. Frenken, K. (2001). Understanding Product Innovation Using Complex Systems Theory. University of Amsterdam, Amsterdam. Fujigaki, Y. (1998). Filling the Gap between Discussions on Science and Scientists' Everyday Activities: Applying the Autopoiesis System Theory to Scientific Knowledge. Social Science Information, 37 (1), 5-22. Galbraith, J. K. (1967). The New Industrial State. Penguin: Harmondsworth. Gibbons, M., C. Limoges, H. Nowotny, S. Schwartzman, P. Scott, & M. Trow. (1994). The New Production of Knowledge: The Dynamics of Science and Research in Contemporary Societies. London: Sage. Giddens, A. (1979). Central Problems in Social Theory. London, etc.: Macmillan. Godin, B. (2003). The Knowledge-Based Economy: Conceptual Framework or Buzzword Godin, B., & Y. Gingras. (2000). The Place of Universities in the System of Knowledge Production. Research Policy, 29 (2), 273-278. Granstrand, O., P. Pattel, & K. Pavitt. (1997). Multitechnology Corporations: Why They Have 'Distributed' Rather Than 'Distinctive' Core Capabilities. California Management Review, 39, 8-25. Heertje, A. (1973). Economie en Technische Ontwikkeling. Leiden: Stenfert Kroese. Kampmann, C., C. Haxholdt, E. Mosekilde, & J. D. Sterman. (1994). Entrainment in a Disaggregated Long-Wave Model. In L. Leydesdorff & P. v. d. Besselaar (Eds.), Evolutionary Economics and Chaos Theory: New Directions in Technology Studies (pp. 109-124). London/New York: Pinter. Kauffman, L. H. (2001). The Mathematics of Charles Sanders Pierce. Cybernetics & Human Knowing, 8 (1-2), 79-110. Kauffman, S. A. (1993). The Origins of Order: Self-Organization and Selection in Evolution. New York: Oxford University Press. 33

Kline, S., & N. Rosenberg. (1986). An Overview of Innovation. In R. Landau & N. Rosenberg (Eds.), The Positive Sum Strategy: Harnessing Technology for Economic Growth (pp. 275-306). Washington, DC: National Academy Press. Krugman, P. (1996). The Self-Organizing Economy. Malden, MA, and Oxford: Blackwell. Larédo, P. (2003). Six Major Challenges Facing Public Intervention in Higher Education, Science, Technology and Innovation. Science and Public Policy, 30 (1), 4-12. Lazarsfeld, P. F., & N. W. Henry. (1968). Latent Structure Analysis. New York: Houghton Mifflin. Leydesdorff, L. (1989). Words and Co-Words as Indicators of Intellectual Organization. Research Policy, 18, 209-223. Leydesdorff, L. (1995). The Challenge of Scientometrics: The Development, Measurement, and Self-Organization of Scientific Communications. Leiden: DSWO Press, Leiden University; at http://www.upublish.com/books/leydesdorff-sci.htm. Leydesdorff, L. (1998). Theories of Citation? Scientometrics, 43 (1), 5-25. Leydesdorff, L. (2001a). A Sociological Theory of Communication: The Self-Organization of the Knowledge-Based Society. Parkland, FL: Universal Publishers; at http://www.upublish.com/books/leydesdorff.htm . Leydesdorff, L. (2001b). Technology and Culture: The Dissemination and the Potential 'Lock-in' of New Technologies. Journal of Artificial Societies and Social Simulation, 4 (3), Paper 5, at http://jasss.soc.surrey.ac.uk/4/3/5.html. Leydesdorff, L. (2002). The Complex Dynamics of Technological Innovation: A Comparison of Models Using Cellular Automata. Systems Research and Behavioural Science, 19 (6), 563-575. Leydesdorff, L. (2003a). The Construction and Globalization of the Knowledge Base in InterHuman Communication Systems. Canadian Journal of Communication, 28 (3), 267289. Leydesdorff, L. (2003b). The Mutual Information of University-Industry-Government Relations: An Indicator of the Triple Helix Dynamics. Scientometrics, 58 (2), 445467. Leydesdorff, L. (2003c). Anticipatory Systems and the Processing of Meaning: A Simulation Using Luhmann's Theory of Social Systems. Paper presented at the European Social Simulation Association (SimSoc VI workshop), Groningen. Leydesdorff, L., & P. v. d. Besselaar. (1998). Technological Development and Factor Substitution in a Non-Linear Model. Journal of Social and Evolutionary Systems, 21, 173-192. Leydesdorff, L., & D. Dubois. (2003). Anticipation in Social Systems. Paper presented at the Paper presented at the International Conference on Computing Anticipatory Systems CASYS'03, 11-15 August 2003, Liège. Leydesdorff, L., & H. Etzkowitz. (1998). The Triple Helix as a Model for Innovation Studies. Science and Public Policy, 25 (3), 195-203. Leydesdorff, L., & M. Meyer. (2003). The Triple Helix of University-Industry-Government Relations: Introduction to the Topical Issue. Scientometrics, 58 (2), 191-203. Luhmann, N. (1984). Soziale Systeme. Grundriß Einer Allgemeinen Theorie. Frankfurt a. M.: Suhrkamp. Luhmann, N. (1990). Die Wissenschaft Der Gesellschaft. Frankfurt a.M.: Suhrkamp. Luhmann, N. (1997). Die Gesellschaft Der Gesellschaft. Frankfurt a.M.: Surhkamp. Lundvall, B.-Å. (1988). Innovation as an Interactive Process: From User-Producer Interaction to the National System of Innovation. In G. Dosi, C. Freeman, R. Nelson, G.

34

Silverberg & L. Soete (Eds.), Technical Change and Economic Theory (pp. 349-369). London: Pinter. Malerba, F., R. Nelson, L. Orsenigo, & S. Winter. (1999). ‘History-Firendly’ Models of Industry Evolution: The Computer Industry. Industrial and Corporate Change, 8 (1), 3-35. Maturana, H. R. (1978). Biology of Language: The Epistemology of Reality. In G. A. Miller & E. Lenneberg (Eds.), , Psychology and Biology of Language and Thought. Essays in Honor of Eric Lenneberg (pp. 27-63.). New York: Academic Press. Maturana, H. R., & F. J. Varela. (1984). The Tree of Knowledge. Boston: New Science Library. Mowery, D. C., & N. Rosenberg. (1979). The Influence of Market Demand Upon Innovation: A Critical Reveiw of Some Empirical Studies. Research Policy, 8 (102-153. Mowery, D. C., & N. Rosenberg. (1989). Technology and the Pursuit of Economic Growth. Cambridge: Cambridge University Press. Nelson, R. R. (1994). Economic Growth Via the Coevolution of Technology and Institutions. In L. Leydesdorff & P. v. d. Besselaar (Eds.), Evolutionary Economic and Chaos Theory: New Directions in Technology Studies (pp. 21-32). London/ New York: Pinter. Nelson, R. R. (Ed.). (1982). Government and Technical Progress: A Cross-Industry Analysis. New York: Pergamon. Nelson, R. R., & S. G. Winter. (1975). Growth Theory from an Evolutionary Perspective: The Differential Productivity Growth Puzzle. American Economic Review, 65, 338344. Nowotny, H., P. Scott, & M. Gibbons. (2001). Re-Thinking Science: Knowledge and the Public in an Age of Uncertainty. Cambridge, etc: Polity. OECD. (1980). Technical Change and Economic Policy. Paris: OECD. OECD. (1996). OECD Economic Outlook, No. 60. Paris: OECD. Parsons, T. S. (1951). The Social System. New York: The Free Press. Rashevsky, N. (1940). Bull. Math. Biophys., 1, 15-20. Rothwell, R., & W. Zegveld. (1981). Industrial Innovation and Public Policy. London: Pinter. Scharnhorst, A. (1998). Citation-Networks, Science Landscapes and Evolutionary Strategies. Scientometrics, 43 (1), 95-106. Schumpeter, J. (1943). Socialism, Capitalism and Democracy. London: Allen & Unwin. Shinn, T. (2002). The Triple Helix and New Production of Knowledge : Prepackaged Thinking on Science and Technology. Social Studies of Science, 32 (4), 599-614. Simon, H. A. (1955). A Behavioral Model of Rational Choice. Quarterly Journal of Economics, 69, 99-118. Skolnikoff, E. B. (1993). The Elusive Transformation: science, technology and the evolution of international politics. Princeton, NJ: Princeton University Press. Storper, M. (1997). The Regional World - Territorial Development in a Global Economy. New York: Guilford Press. Theil, H. (1972). Statistical Decomposition Analysis. Amsterdam/ London: North-Holland. Turing, A. M. (1952). Philos. Trans. R. Soc. B., 237, 5-72. Van den Belt, H., & A. Rip. (1987). The Nelson-Winter-Dosi Model and Synthetic Dye Chemistry. In W. E. Bijker, T. P. Hughes & T. J. Pinch (Eds.), The Social Construction of Technological Systems. New Directions in the Sociology and History of Technology (pp. 135-158.). Cambridge MA: MIT Press.

35

Wallerstein, I. (1974). The Modern World System: Capitalist Agriculture and the Origins of the European World Economy in the Sixteenth Century. New York: Academic Press. Whitley, R. D. (1984). The Intellectual and Social Organization of the Sciences. Oxford: Oxford University Press.

36