return to homepage

Are EU Networks Anticipatory Systems?
An empirical and analytical approach

Loet Leydesdorff

Department of Science & Technology Dynamics
University of Amsterdam
Nieuwe Achtergracht 166, 1081 WV Amsterdam, The Netherlands
<loet@leydesdorff.net>; http://www.leydesdorff.net/

forthcoming in: Daniel M. Dubois (Editor),
Computing Anticipatory Systems -- CASYS'99
Woodbury, NY: American Physics Institute, 2000 

Abstract

A social system can be considered as distributed by its very nature. Social communication among humans can be expected to be reflexive. Thus, this system contains uncertainty and the uncertainty is provided with meaning. This dual-layeredness enables the network to organize itself ("autopoietically") into an anticipatory mode. The extent to which anticipatory functions have been developed can be observed, notably in the case of intentional constructions of reflexive layers of organization. In this study, the "Self-Organization of the European Information Society" is analyzed from this angle. Using empirical data, I argue that the increasing unification in representations at the European level allows for another differentiation in terms of the substantive communications that are represented. Insofar as the reflexive layers are differently codified, the anticipatory functions of the system can be strengthened.

INTRODUCTION

Anticipatory systems are able to operate upon a representation of themselves.1,2 The networks currently under construction by the European Union can be considered as systems of this kind: EU member states select upon the European organizations that they have produced relationally (e.g., the European Monetary Union), and the European organizations (e.g., the European Central Bank, the European Commission, or the Parliament) are able to act upon the networks among the member states. Mutual expectations are among the factors shaping future behaviour at both levels. Over time, what is selected can alternate with what is selecting, thus driving the system of networks potentially into a dynamics of its own ("lock-in").3

The European networks operate in an open environment with the rest of the world. These relations constitute yet another network system, for example, in the case of global developments like the techno-sciences. A system can be considered as self-organizing to the extent that it is able to reproduce a representation of its own boundaries in relation to such a changing environment.4

In this study, I summarize results from three studies of such dynamics among representations. [Note 1: The studies were part of an EU-project entitled "The Self-Organization of the European Information Society," nr. SOE1-DT97-1060 of the program for Targeted Social-Economic Research of the European Commission. The author is grateful to Nienke Oomes and Gaston Heimeriks for co-authoring two of the case studies (Leydesdorff & Oomes 1999; Leydesdorff & Heimeriks, forthcoming), and to other colleagues in this project for further collaborations and stimulating discussions.] First, I raise the question of whether the European Monetary System (EMS) can be considered with hindsight as a system anticipating the introduction of the Euro. In addition to analyzing the development of the nominal exchange rates between the European currencies over time, I explore the interaction of the monetary system with the economic system in terms of real economic exchange rates or purchasing power parities.

Second, the statistical instruments for answering this question will be applied to the question of whether the EU can also be considered as a single publication system. Thirdly, I analyze in greater detail the interactions between the global network system and the European publication system by focusing on "biotechnology" as a priority area in science and technology policies.
 
 

SOCIAL SYSTEMS

Social Systems Self-organize the Communication

Autopoietic or self-organizing networks have been defined in biology as networks which are able to maintain an organization in relation to a structure. The system is operationally closed, but this organization remains structurally coupled to environmental dimensions. For example, the social system operates in terms of communications among human carriers, but it is not able to function without the latter.

The specific organization operating on structural dimensions is expected to guide the reproduction as in a coevolution. Thus, a (dynamic) retention mechanism can be specified in the interaction between the two layers of the system. For example, as Maturana (1978, at p. 43)5 put it:

"Since it is a constitutive feature of an autopoietic system to maintain homeostatically invariant its organization under conditions of structural change, the realization of the autopoiesis of a plastic living system under conditions of perturbation generated by a changing medium must result in the selection of structure in the living system that incorporates, in its autopoietic network, specific processes (changes of state) that can be triggered by specific changes of state of the medium; otherwise the system disintegrates."

In contrast to biological systems, social systems remain distributed, that is, they can be expected to fail to integrate. In other words, they do not perform a life cycle in the biological sense, but they exhibit complex (that is, recursive and interactive) dynamics. Luhmann6 has analyzed the network of communications among human beings in terms of the self-organization of the social system. The human carriers ("consciousness systems") are structurally coupled to the network, analogously to the coupling of brain cells to the neural net.

Following Bateson,7 Luhmann6 focused on the communication of meaningful information. However, he did not specify the selective operation by which Shannon-type information or uncertainty is used to generate meaningful information. In my opinion, a sociology has to specify this operation reflexively: sociological hypotheses can meaningfully be specified with reference to uncertainties. The distributions under study remain uncertain, and therefore one is able to specify an expectation. The uncertainty in the observable distribution is then expected to contain (Shannon-type) information. These observations enable us to update the expectations. Society is organized by recursively selecting this information as meaningful information in increasingly higher order of codification of the communication.8

The recursivity is epistemologically implied in the notion of probabilistic entropy: a probabality distribution can have a probability and a selection may be selected (e.g., for stabilization over time). In other words, the self-organization of the information society can be specified in terms of the operation of the communication networks upon themselves (as in scientific reflections) or among each other as in translations among codes of subsystems.9 As the selections are stabilized over time, social institutions are shaped that can be considered as the retention mechanism of the social system.
 
 

From Social Systems to Reflexive Anticipations

Thus defined, social systems are extremely flexible since they contain (uncertain) information expected to be communicated, while these communication processes are only conditioned ("enabled and constrained")10 by the reflexive reconstruction of their respective trajectories in terms of institutions. Thus flexible, yet reflexive systems are under evolutionary pressure to develop the code of the communication into possibly anticipatory configurations. National states, for example, can be considered as competing in terms of the institutional arrangements to solve the problem of differentiation.

The adjustment of the institutional retention becomes particularly urgent when the codes are further refined by scientific reflections, that is, in a knowledge-based economy. The functional coding of languages generates a global overlay which, as a hypercycle, drives the underlying systems into exploring fits more optimal than the "naturally" (or "nationally") given ones. While a specific match between functional and institutional differentiation is expected to lock-in, the complex social system may drift into new options for recombination by using its reflexive mechanisms of variation (e.g., translation).11

The European network systems can be considered as such a reflexive construction: the overlay has emerged in the relations among nation states, and since World War II it has been increasingly codified into European organizations. The construction of European technological artifacts (like CERN and Airbus) and monetary unification (the Euro) can be expected to feed back onto the underlying national systems levels, thus, in principle, freeing up existing lock-ins with alternative options.

For example, the development of the Airbus allowed European national aircraft industries to recombine strengths and weaknesses at the network level without sacrificing national advantages (Frenken, forthcoming). Thus, the European system can be expected to delineate itself in terms of the communicative competencies of the contributing members (e.g., states, industries, academics). Such systems are expected to organize themselves in anticipation of their own potentials because they have access to uncertain and subsymbolic representations of themselves.
 
 

METHODS AND DATA

Can systemness be indicated using the distribution of the European networks, and what might be the implications of the respective operations of these networks for the future of Europe? Can transnational feedback be organized programatically in a feedforward mode? How does the European network system operate internally, that is, in terms of variation among nation states, and how does it relate to its global competitors (e.g., the American and Japanese systems)?

The first two studies share a methodological apparatus in using entropy statistics for the operationalization of processes like systemness, path-dependency, co-variation, and co-evolution.12 Parametric statistics are used in the third study because of the much larger data sets, but the reasoning is similar. While the first two studies focus on the time dimension, the third one analyzes the complexity as a finger-print at a certain period of time, namely the year 1996.

Data

Monetary and real economic exchange data are based on statistics provided by the OECD (1996) and the IMF (1997). Time series data were available for the periods 1980-1996 and 1984-1996, respectively.13 Publication data for the various EU countries, the U.S.A., and Japan were collected from the Science Citation Index in February 1999. The on-line version of this database at DIALOG can be searched conveniently by using institutional addresses, delineations between calendar years, and Boolean OR-statements for combining sets such as the EU. We use only articles, reviews, and letters.14 [Note 2: "Notes" have been dropped from the on-line version of the database since 1996.] The data cover the period 1974-1998, but for various reasons the data for 1974-1979 have been discarded.15

Publication data for five core journals in "biotechnology" were downloaded from the CD-ROM editions of the Science Citation Index for 1996.16 In this case, we used only research articles.
 
 

Methods

Longitudinal studies

A self-organizing system is expected to maintain its structure despite variations at other levels. Thus, it is buffered against variations at the level of subsystems and it operates vis-à-vis competing systems by reorganizing its resilience. From the European perspective, the national systems are expected to follow their own trajectories because they are relatively firmly integrated. The potential differentiation among these relatively indepent trajectories is expected to disturb the European integration. As integration succeeds, predictions on the basis of systemness at the European level will gradually improve in comparison to the normalized sum of the predictions based on the individual trajectories of the member states.

The two predictions can be tested against each other. As a diachronical indicator of systemness, I shall use the Markov property which states that the best prediction of a next state of the system under study is based on the maintenance of its present state. In other words, the current distribution is not expected to change in the case of a system. Any unexpected change can then be expressed in terms of bits of information using Shannon's17 formula for probabilistic entropy.18
 

Country

1980

....

1995

1996

Country A

 

 

 

 

Country B

 

 

 

 

...

 

 

 

 

Country N

 

 

 

 

FIGURE 1: Time series of rows versus systemness over the columns



Whereas the current distribution can be considered as a column in a series of time series organized in rows, each row contains a time-series expectation of the next state for a (national) sub-system. The two representations are analytically independent (Fig. 1). They are expected to disturb each other in the event (that is, the cell values). However, one expects to find both European systemness and national (sub-)systemness. The question is whether systemness is a better predictor than subsystemness, or vice versa.

For the subsystemic prediction based on each time series from year m to year n, it can be shown19 that the best prediction for the year (n + 1) is expected to be:

(Smn Fi) - Fm

Fn+1 = { ________________ } * Fn

(Smn Fi) - Fn

This prediction is non-parametric, and the failure of the prediction can be expressed in terms of bits of information.20 After proper normalization (and given the best fit for using all possible values for m), the longitudinal expectation can be compared with the multivariate one (based on the Markov property) in terms of the respective qualities of the predictions. Furthermore, the relative improvement of prediction can be decomposed in terms of the contributing subsystems. A negative sign indicates a worsening of the prediction.

In summary, the test for systemness is based on a comparison of a table of indicators which contains both multi-variate and longitudinal data. Such tables are produced regularly by various bureaus of statistics. One can also test for whether, for example, monetary and economic integration are co-evolving by combining more than a single table in a multi-dimensional array. In the study of the scientific publication system, we also assess the extent to which the Markov prediction itself is improving over time as a predictor.
 
 

Functional versus national differentiation

In the case of the biotechnology data we will use another type of table, with scientometric data. Each document is organized as a row in a matrix with its title words as variables along the column dimension. The documents, however, also contain institutional addresses. Thus, the cases can be grouped in terms of (one or more) national (sub-)systems of scientific publications. Since the set of documents was originally selected on the basis of the criterion of core journals of biotechnology, the ensemble of title words contains a representation of the global organization at the level of the intellectual field. Each document can be considered as a vector in the eigenstructure of this global network. Note that the two operationalizations are again analytically independent since they are based on the orthogonal axes of rows and columns.

We used factor analysis to explore the structural dimensions of the co-occurrences of title words at the field level. The quality of the national distinctions can be assessed using discriminant analysis over the groupings both within Europe and by comparing the EU with the U.S.A. and Japan. How specific are the word distributions in these subsets? The correctly classified document sets provide us with word frequency lists. These can be compared with factor loadings for different factors in the global analysis I will use Pearson's r for the evaluation of the comparison (since these analyses were based on parametric statistics).

If the three subsets of the EU, the U.S.A. and Japan were completely uncoupled (N= 3 and K= 0),21 a three-factor solution would provide a best fit between the results of the two types of analyses. A single global dimension would add a fourth dimension on top of these three systems. If, however, the coupling is more complex (as expected), a further increase in the number of factors can be expected to improve the fit. Note that in the abstract, a system with three nodes (N = 3) and two links (K = 2) is expected to contain two suboptima. With increasing numbers of N (as in the case of the decomposition of the EU), the number of possible suboptima increases with N in the exponent.22
 
 

RESULTS

The European Monetary System

The analysis was performed in 1997, that is, before a decision was made about the precise inclusion of currencies in the Euro-zone. We distinguished between the twelve countries of the European Community (EC) that signed the Maastricht Treaty in 1991 and the fifteen member states of the European Union (EU). Belgium and Luxembourg, however, use a single currency, so that N = 11 in the case of monetary integration. Greece was not included in the IMF-data for the period under study.

Table One exhibits the improvement of the prediction for the Markov assumption. The European system can be shown to develop increasingly as both a monetary and an economic system, but the two developments do not reinforce each other. Tables Two and Three show that the decomposition is sometimes different in sign for the various countries involved, particularly for the 1990s.

Decomposition of the data for EC member states shows that those nations that have firmly coupled their currency rates to the German Mark have been less successful in the European economic integration than some of the other countries (U.K., Italy, and Spain). Perhaps it is difficult to keep pace with the German economy without a degree of freedom in monetary policies.

In short, the results of our analysis suggest that the European system contains an internal dynamics between a DM-zone and a U.K. Sterling zone. At the global level, this internal dynamics operates as a differentiation in opposition to the U.S. Dollar zone which includes many other currencies.

TABLE 1. Improvement (or worsening) of the prediction on the basis of the assumption of systemness in the nominal exchange data in comparison to univariate timelines, for the case of 15 EU member states and 11 EC member states (from: Leydesdorff & Oomes 1999, p. 74).
 

year

 

15 member
states (EU)

 

11 member
states (EC)

 

REER
(EC)

1987

 

-0.82607438

 

-0.86647414

 

0.02511461

1988

 

1.68470963

 

1.86778215

 

1.24406229

1989

 

-0.01974935

 

0.29255241

 

-0.11642926

1990

 

-0.08202988

 

0.31599596

 

0.29265731

1991

 

-0.24174252

 

0.48141053

 

-0.27362359

1992

 

-0.70750947

 

-0.05812540

 

-0.09436024

1993

 

-0.73487024

 

-0.54354992

 

0.76952849

1994

 

2.40052927

 

0.18931945

 

0.52320784

1995

 

-0.09895706

 

-0.05664408

 

0.41080807

1996

 

2.38339315

 

2.09119657

 

1.15037528

TABLE 2. Improvement of the prediction on the basis of the Markov assumption for 10 European currencies in relation to the German Mark, based on nominal exchange rates (from: Leydesdorff & Oomes 1999, p. 76).
 

COUNTRY

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

 

BELGIUM/

LUX

6.21

8.19

3.25

-0.74

1.85

2.15

1.55

5.73

5.77

6.19

 

DENMARK

5.38

5.86

2.35

-1.78

0.52

1.51

0.52

5.86

4.75

5.39

 

FRANCE

4.17

3.03

-1.04

-1.24

1.40

0.64

0.79

6.15

4.97

3.88

 

IRELAND

2.46

-1.91

-3.42

-0.38

0.79

0.89

0.89

4.81

4.03

1.54

 

ITALY

-1.25

1.77

-1.56

-0.94

0.88

-1.26

-3.37

-14.60

-2.07

-10.72

 

NETHERL.

7.26

10.47

4.78

-0.08

0.87

0.82

1.24

7.04

5.69

5.32

 

PORTUGAL

-14.09

-13.14

-9.35

-5.19

-5.35

-3.25

2.95

1.85

-7.07

0.01

 

SPAIN

-3.91

-3.63

2.34

6.90

5.47

1.25

-0.94

-10.86

-13.72

-4.67

 

UK

-7.09

-8.77

2.94

3.76

-5.95

-2.79

-4.18

-5.78

-2.28

-4.85

 

EC

-0.87

1.87

0.29

0.32

0.48

-0.06

-0.54

0.19

-0.06

2.09

 

TABLE 3. Improvement of the prediction on the basis of the Markov assumption for 10 European currencies in relation to the German Mark, based on real economic exchange rates (from: Leydesdorff & Oomes 1999, p. 77).
 

COUNTRY

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

BELGIUM

-1.81

-1.75

0.72

3.01

1.06

0.94

0.27

-0.71

-3.71

-3.29

DENMARK

-2.25

-4.28

-1.90

1.81

1.19

1.42

3.26

-1.02

-1.28

-2.15

FRANCE

-1.57

0.02

1.76

3.22

1.96

2.50

2.88

-1.86

-1.86

-0.82

IRELAND

-2.70

-0.64

4.48

4.16

1.36

2.03

1.68

-1.98

-1.48

-0.14

ITALY

-1.30

-3.50

-0.70

-0.96

-1.96

-0.44

-1.05

-3.79

-2.64

6.49

LUXEMB.

0.54

3.20

1.72

0.74

1.28

0.70

1.44

-0.48

-2.55

-0.16

NETHERL.

1.07

-1.99

1.14

5.00

4.40

3.15

1.72

-2.57

-3.11

-3.08

PORTUGAL

2.91

4.85

1.50

-3.64

-3.81

-7.50

-10.64

-4.12

1.86

-0.76

SPAIN

-2.11

-3.67

-4.16

-7.45

-6.11

-1.86

0.96

7.45

9.42

2.50

UK

7.24

9.02

-4.68

-5.61

0.36

-1.04

0.25

9.60

5.76

2.58

EC

0.03

1.24

-0.12

0.29

-0.27

-0.09

0.77

0.52

0.41

1.15


 

If one combines the Markov forecasts in the two dimensions of both nominal and economic integration, the prediction for 1995 is considerably worsened (Fig. Two). Thus, the interaction between economic and monetary integration can be negative, showing the open yet differentiated character of the emerging European system.

FIGURE 2: The improvement of the prediction on the basis of the Markov assumption for nominal, real exchange rates, and their combination
 
 

The European Publication System

Figure 3 provides a graphic representation of the statistics of the test for systemness in both the EU data (15 member states) and the data for comparison at the global level between the EU, the U.S.A., and Japan. Since the average value of this measure is very close to zero both for the EU and for the global comparison, this data shows mainly randomness, in my opinion. In other words, the prediction on the basis of the Markov assumption is neiter improved nor worsened when compared with the assumption of independent trajectories for the nation states. Moreover, this picture is not significantly affected by focusing on "articles" only.

FIGURE 3: Improvement of the prediction on the basis of the Markov assumption versus the assumption of historical trend lines
 
 

The extreme values for the EU data in 1992 and 1993 are partly due to the German unification process (1991). If this event is taken into account (that is, by including the data for the German Democratic Republic in the time series for Germany), the extreme effects in 1993 are mitigated, but they do not disappear in 1992. In other words, the EU publication system was not severely upset by German unification because the EU set was never a system in the first place.

FIGURE 4: Decrease of the expected information content of the update in the case of the EU compared with the global system.
 
 

Multivariate prediction (based on the assumption of the Markov property) can itself also be considered as a measure of systemness. Figure 4 shows the trendlines for a comparison of the EU-countries, a global comparison, and thirdly the limitation to research articles only. The graphs for the EU and the global comparison cannot be compared directly in terms of absolute values (since the number of categories and therefore the maximum entropy is different for the two systems of reference). However, the slope in the EU case is negative, while it is slightly positive in the case of the global comparison. This can be considered as a weak indicator of increasing systemness in the EU data set (since the Markov property would mean that the occurrence of the event does not provide us with new information, and thus the value of I tends to vanish). The correlation between the fluctuations in the curves suggests a coupling between the global and the EU "system."

In summary, these results indicate that in terms of publications, the EU does not yet exhibit systemness. However, the alternative hypothesis of trajectories for individual member states is also not corroborated by these results. The situation seems rather indecisive. Furthermore, there is no sign of an increase in systemness over time. The systemness hypothesis was rejected in most years both in the case of the distribution of EU member states and in the comparison of the EU data with those of the U.S.A. and Japan.[Note 3: The precise values of the mean in the data (of Figure Three) are -0.0115 mbit for the EU (articles, reviews, and letters), -0.0094 for the EU in the case of only articles, and -0.089 for the global comparison between the EU, the U.S.A., and Japan. Thus, the assumption of systemness in the data is always rejected, but the hypothesis remains somewhat stronger in the European case than in the global comparison.]

The European RTD-programmes may have been helpful as a mechanism of resource allocation for national systems, but these results provide insufficient evidence that a supra-national system is a relevant level for understanding the system's integration. Given this conclusion, a best forecast for 1999 can be provided (in Table Four) on the basis of the trend lines for individual EU states.

TABLE 4. Prediction of national percentage shares of publications for 1999
 

 

1998

1999

(prediction)

first year included in the prediction

Belgium

Denmark

France 

Germany

Greece

Ireland

Italy

Luxembourg

Netherlands

Portugal

Spain

U.K. *)

(EC total

Austria

Finland

Sweden

(EU total

U.S.A.

Japan

1.3165

1.0198

6.5159

8.8291

0.6026

0.3300

4.1350

0.0110

2.4491

0.3270

2.8465

8.9366

33.8616

0.9026

0.8947

1.9542

35.7794

31.6623

9.4394

1.3597

1.0589

6.6476

9.2176

0.6480

0.3475

4.2133

0.0122

2.4540

0.3621

3.0380

8.9468

34.3386

0.9441

0.9142

1.9673

37.3091

31.0931

9.7012

1991

1996

1996

1996

1995

1996

1996

(1988)

1996

1996

1995

1980

1995)

1994

1996

1995

1996)

1996

1994

*) 1980 was the first year included in the longitudinal data.

As noted, our time series are based on data since 1980. The rightmost column of the table provides the beginning date on which each best estimate is based. With the exception of the U.K. and Belgium(/Luxembourg), only the most recent years are needed for the prediction in the case of the EU countries, the U.S.A., and Japan. These results suggest that the network of lateral relations among the countries works to mitigate the influence of the historical trajectories.
 
 

The Case of "Biotechnology"

The delineated set of biotechnology papers in 1996 provided us both with an operationalization of the European dimension in terms of institutional addresses and with an independent operationalization of the intellectual space in terms of words and co-occurrences of words in the titles of the document sets. We limited the analysis to a comparison among the documents with addresses in the EU, the U.S.A., and Japan (n= 778). Additionally, the EU set (n= 394) was decomposed in terms of a comparison among its member states. [Note 4: Since the set did not contain data with an address in Luxembourg, N = 14.]

The discriminant analysis using institutional addresses results in 77.6% correctly classified documents in the case of a global comparison between the U.S.A., the EU, and Japan. For Japan, this correct match is as high as 86.7% showing its relative specialization. The decomposition for the E.U. similarly shows a correct classification in 86.5% of the cases. (The overall picture from the analysis under 3.2 confirms that the Japanese set document behaves like those for the major European countries.)

FIGURE 5: Pearson correlations between Word Frequency Lists and Factor Loadings for seven European nations given different numbers of factors.
 

The main finding is that the papers from different European states couple to the international vocabulary in varying degrees (Figure 5 shows the results for seven European countries). Strong coupling can be observed for the U.K., the Netherlands, Sweden, and particularly Denmark. The French document set exhibits the weakest coupling, possibly because of the implied use of English title words for the operationalization of the intellectual organization.

FIGURE 6: Pearson correlations between Word Frequency Lists and Factor Loadings for a variable number of factors.

The variation between the European countries is larger than that between the EU, the U.S.A., and Japan (Fig. 6). At the world system level, competition and collaboration between the EU and the U.S.A. is predominant. The Japanese document set is related mainly to the American set, but at a lower level of integration.

FIGURE 7: Effect of wordfrequency threshold on the coupling of the American document set, but not on the European one

Although the European dimension displays a significant coupling on the global dimension, Fig. 7 shows that this coupling is not sensitive to selections in terms of substantive words, while the American one couples better under this "higher selection pressure." If one uses, for example, a word frequency cutoff level at nine instead of six words, this has a significant influence in the case of the U.S.A. coupling, but not in the case of the EU. In my opinion, these results suggest that the European coupling is not cognitive but institutional, while the national systems are also coupled substantively.
 
 

CONCLUSIONS

The data which we used for this research is itself a manifestation of the increase in the use of representations to map ongoing developments in the information society. Public policy makers and the strategic staff of corporations are continuously supplied with these types of indicators. Is this information recombined at the European level to develop intelligent policies anticipating future positions? Is a European system emerging?

In the case of the publication system, the answer has been negative. The metaphor of world trade, which is more dense among European member states than across industrial blocks, seems to describe the situation more accurately. While firmly integrated at national levels, the R&D systems seem to function globally in relating to the international environment, for example, by surveying and disseminating new opportunities. European programs mainly stimulate the further participation of countries which have hitherto remained behind at the international level. Whether they also reach their stated objectives of stimulating the functional integration of the European science system into regional innovation networks and localizable technology developments cannot be evaluated in terms of this data.

The study of monetary and economic integration also shows differentiation rather than harmonization between the dimensions of the coordination problem. The case of "biotechnology" indicates that systemic stimulation at the European level has remained external to substantive developments. In other words, the self-organization of the European information society provides an additional (and potentially orthogonal) dimension to the self-organization of the cultures of the member states. European action has hitherto reinforced the distributedness of the development. It can therefore be considered as an enabling (as opposed to a constraining) condition.

The networks develop variety and differentiation rather than the intended unification. European policies stimulate unification in the representation, but not necessarily in terms of what is represented by the representations. Paradoxically, the unification may further the anticipatory strength of the underlying systems by allowing for a more complex pattern of differentiation. An overlay of communication emerges that is no longer based on national differences or institutionalized divisions of labour, but on a next-order functional differentiation. In this case, the European level may have been able to remain subsymbolic because it has not (yet) been needed for the symbolic reproduction of the national systems on which the EU networks build.

ACKNOWLEDGEMENT

The author acknowledges partial funding by the program for Targeted Social-Economic Research of the European Commission, project nr. SOE1-DT97-1060 ("The Self-Organization of the European Information Society").
 
 

REFERENCES

1. Rosen, Robert, Anticipatory Systems, Pergamon Press, 1985.

2. Dubois, Daniel, "Introduction to Computing Anticipatory Systems," in: Daniel Dubois (ed.), Abstract Book of the First International Conference on Computing Anticipatory Systems, Liège: Chaos, 1997, pp. 1-5.

3. Arthur, W. Brian, Increasing Returns and Path Dependence in the Economy. Ann Arbor: University of Michigan Press, 1994.

4. Leydesdorff, Loet, Int. J. Systems Research and Information Science 6, 219-30, 1994.

5. Maturana, Humberto R., "Biology of Language: The Epistemology of Reality," in: G. A. Miller and E. Lenneberg (Eds.), Psychology and Biology of Language and Thought. Essays in Honor of Eric Lenneberg. New York: Academic Press, 1978, pp. 27-63

6. Luhmann, Niklas [1984], Social Systems. Stanford: Stanford University Press, 1995.

7. Bateson, Gregory, Steps to an Ecology of Mind, New York: Ballantine, 1972, at p. 489.

8. Leydesdorff, Loet, Systems Research and Behavioral Science (forthcoming).

9. Leydesdorff, Loet, International Sociology 12, 25-45, 1997.

10. Giddens, Anthony, The Constitution of Society, Cambridge: Polity Press, 1984.

11. Leydesdorff, Loet, "The New Communication Regime of University-Industry-Government Relations," in: Henry Etzkowitz and Loet Leydesdorff (Eds.), Universities and the Global Knowledge Economy: A Triple Helix of University-Industry-Government Relations. London: Cassell, 1997, pp. 106-117.

12. Leydesdorff, Loet, The Challenge of Scientometrics: The development, measurement, and self-organization of scientific communications, Leiden: DSWO/Leiden University Press, 1995.

13. Leydesdorff, Loet, and Nienke Oomes, Social Science Information38, 57-85, 1999.

14. Braun, Tibor, Glänzel, Wolfgan, and Schubert, Andras, Scientometrics15, 165-70 (1989).

15. Leydesdorff, Loet, Science and Public Policy15, 149-152, 1988.

16. Leydesdorff, Loet, and Heimeriks, Gaston, "The Self-Organization of the European Information Society: The case of `biotechnology'," presented at the Annual Meeting of the Society for the Social Studies of Science (4S), Halifax, October 28- November 1, 1998.

17. Shannon, Claude E., Bell System Technical Journal, 27, 379-423 and 623-56, 1948.

18. Theil, Henry, Statistical Decomposition Analysis. Amsterdam: North-Holland, 1972.

19. Leydesdorff, Loet, Social Networks 13, 301-45, 1991.

20. Theil, Henry, Applied Economic Forecasting. Amsterdam: North-Holland, 1966.

21. Kauffman, Stuart A., The Origins of Order: Self-Organization and Selection in Evolution, Oxford: Oxford University Press, 1993.

22. Kauffman, Stuart A., At Home in the Universe, New York/Oxford: Oxford University Press, 1995, at pp. 166 ff.

return to homepage