Mark W. Johnson:

Leydesdorff’s Compass: A Review of “The Evolutionary Dynamics of Discursive Knowledge,”

Cybernetics and Human Knowing. 29(3/4) (2022); [preprint-version]


Over the last 40 years, Loet Leydesdorff’s project has been to develop practical and empirical applications of second-order cybernetic thinking with the aim of providing better metrics of socio-economic effectiveness. This is a project which has necessitated a deep reading not just of key cybernetic thinkers, but the philosophical and sociological provenance of cybernetic thinking, as well as technical development and application of models and algorithms. In recent history, there are few comparable examples of the depth and scope of this kind of intellectual enterprise – perhaps only those of some of the scholars to whom Leydesdorff’s thinking is closest - most notably, Niklas Luhmann.

His new open-access book, “The Evolutionary Dynamics of Discursive Knowledge” is an important summation of the synthesis of Leydesdorff’s ideas and practical investigative techniques. Reflecting the richness of Leydesdorff’s intellectual concerns, there is a lot of territory to cover. To navigate this requires the use of a good compass, and here Leydesdorff has provided considerable clarity in the presentation of his ideas. Close reading rewards with profound insights into intellectual history and provides illumination on current trends such as “big data”. The book’s significance lies not just in the unique way Leydesdorff has reassembled and operationalised cybernetic sociology with Shannon’s information theory, or how he has reunited cybernetics with its philosophical predecessors (particularly Husserl), but also the fact that Leydesdorff himself is a witness to the principal intellectual currents of late third of the 20th century and the opening decades of the 21st. Luhmann was a personal friend, while Leydesdorff’s broader intellectual circle includes Latour, Habermas, Giddens and Simon as well other luminaries from cybernetics and evolutionary economics. Key intellectual moments are recounted – not least, the Habermas-Luhmann debate -  which has been so crucial to the future directions of sociology, against the backdrop of the social unrest of 1968, the Prague Spring (both of which he narrowly avoided getting mixed-up in!) through to the development of Science and Technology policies in the OECD and European Union.

The Evolutionary Dynamics of Discursive Knowledge begins with an overview of themes which recur throughout the book. It sheds light on the social and philosophical context from which Leydesdorff’s ideas come. For future generations of readers, this telling of the story of “how ideas come to be thought” is particularly important, and it makes for a compelling introduction to what follows.  We are gently introduced to the “back-story” of Leydesdorff’s thinking: from the challenges to Marxist dialectic presented by the information-driven economy, to 20th century phenomenology, and the fall-out from the Husserlian intersubjective view of consciousness, through to Parsons, Schutz, Luhmann and Simon. Behind all of it is the concept of “meaning” – and here we find one of the key themes in Leydesdorff: Husserl’s “horizon of meaning” becomes a “horizon of expectations”, and with this, Leydesdorff is able to invoke the biology of “anticipatory systems” from Robert Rosen, and a more mathematical representation produced by one of Rosen’s students, Daniel Dubois. These fundamental elements are then gradually unpacked through the three parts of the book.

Part 1 deals with the concept of scientific knowledge and the way in which an empirics of knowledge can be operationalised. Sitting on Luhmann’s shoulders allows Leydesdorff to present knowledge as a mechanistic communicative process which is grounded in selections of utterances steered by “codes of communication” which in turn produce the functionally differentiated society in which we all live. Scientific knowledge is one result of this functional differentiation, where what matters is what is selected as “meaningful” by groups of scientists. Leydesdorff convincingly argues that this selection of meaning leaves an “imprint” in the citation practices of scholars in academic journals. He argues that “the modern sciences are discursive and mediated” producing “variation” in texts and “selection” in citation practices. Citation for Leydesdorff provides a clear example of a process of codification - so citation practices are used as a vehicle for understanding the underlying dynamics of codes of communication in scholarly discourse.

This idea forms the basis of a large amount of empirical work which Leydesdorff has conducted using bibliometric data and other forms of text analysis. Following Luhmann, he argues that the sciences “self-organize into disciplines and specialties using specific codes”, and the selection of meaning is determined by the operation of these codes. Part 1 explains some of the analytical techniques which can be used to expose a dynamics of “meaning selection” in science. Using Shannon’s information theory (which ultimately is all about “selection”) Leydesdorff shows how Shannon’s measure of “transmission” (otherwise called “mutual information”) in citation practices can act as an index of extent to which options are realised and organised within a system, while the unrealised options are represented by the generation of multiple versions of the same thing, which in Shannon is called “redundancy”. He argues that the complex dynamics of economic systems arise with the interaction between transmission and organisation on the one hand, and the generation of redundancy and new options on the other. The dialectical tension between these underpins a major theme in the rest of the book.

Leydesdorff presents his thinking as “dualist” and not “monist”. In Chapter 3 he criticises Latour’s Actor-Network Theory because “heterogeneous dimensions are homogenized in a pan-semiosis”. Leydesdorff, by contrast, offers a structural, dualistic and dialectical approach based on different orders of system dynamics in communication: specifically, the dialectical relation between communication and expectation. He argues that, unlike in Latour (and indeed, a lot of postmodern theory, and some interpretations of second-order cybernetics), authors, texts and cognitions cannot be reduced to one another. Homogenizing dimensions leads in the direction of theology and the postulation of the single unifying cause for social complexity, or a prime mover. Leydesdorff argues that this cannot be right. While this assertion of irreducibility might lead one to suspect Leydesdorff of realism of the kind that Margaret Archer or Roy Bhaskar subscribe to, Leydesdorff remains true to the Second-order cybernetic roots which he inherited from Luhmann. There is no mind-independent reality, but there are dialectical dynamics of communication and expectation from which consciousness constructs reality.

If Leydesdorff is to defend this position, however, he requires an empirics of expectation which can be distinguished from an empirics of communication. His transformation of Husserl’s “horizon of meanings” into a “horizon of expectations” allows him to operationalize expectations in terms of anticipations. He argues that dualism between expectation and communication was always implicit in Husserl’s Cartesian Meditations, but it was misread even by Husserl’s most faithful followers (particularly Alfred Schutz). If there is a fundamental critique of modern social theory, it is here. Not just Latour is accused of monism: Anthony Giddens’s Structuration Theory, which Leydesdorff builds on, suffers from monism for reasons similar to those indicated by Margaret Archer (who complains that Giddens is “elisionist”). Giddens argued that structures could not be empirically investigated, because they represented “absent differences”. Leydesdorff argues to the contrary - the “second contingeny” of expectation dynamics is empirically investigable: “Why would one not be allowed to formulate hypetheses about a second contingency in social structures?” he asks.

In the chapter “Towards a Calculus of Redundancy” (Chapter 4), Leydesdorff shows how this might be done, by introducing the information theoretical calculations which have been at the centre of his empirical work. In revisiting Shannon’s sender-receiver model, and challenging Shannon’s own antipathy towards using his entropy calculations to measure meaning (Shannon’s co-author, Warren Weaver, famously disagreed with this!), Leydesdorff elaborates higher-order dynamics in human communication wherein the selection mechanisms for meaning are constructed at a supra-individual level. At the centre of his basic idea that human communication depends on a deep level “coordination of expectations” which operate on the basis of codes of communication which themselves have a self-organising dynamic, Leydesdoff points out that what matters is the balance between the capacity to produce variation, or alternative options, and the process of selecting between options. The production of variation he sees as the generation of redundancy, arguing that this is particularly relevant to economic systems because, as he puts it:

“The number of options available to an innovation system may be more decisive for its survival than the historically already-realized innovations.”.  This therefore necessitates an indicator of surplus innovations, for which Leydesdorff deploys Shannon’s understanding of redundancy. The dynamics between redundancies in scientific communication Leydesdorff relates to “synergy”.

The mechanics and exploitation of an indicator of synergy and surplus innovations is the focus for Part 2 of the book. At the heart of Leydesdorff’’s empirical work has been economic analysis which explores the dynamics between different institutional groups, particularly on the “Triple Helix” relations between Universities, Industry and Government. Perhaps the most obvious question to ask with regard to this is, why three? Leydesdorff responds (citing Simmel) that only with three dimensions do complex dynamics emerge. Each dimension establishes its identity through specific “codes of communication”, and innovation can be seen to be the result when different codes of communication interact or “interfere” with one another, producing varying levels of “mutual redundancy” (alternative descriptions of the same thing), and transmission (or “mutual information”). This allows Leydesdorff to decompose the Shannon formulae for mutual information in 3 dimensions (which has puzzled cyberneticians since Ashby), identifying two complementary components: a measure of “transmission”, and a measure of “mutual redundancy”.

Part 2 gives an overview of the empirical application of these techniques, which have been applied to various national and regional economic systems. Throughout this work, Leydesdorff looks for synergies in the activities of researchers or innovators, looking specifically at the dynamics across regional boundaries, using a variety of data sources including the three levels of Eurostat “Nomeclature of Territorial Units” (NUTS) data. The mutual redundancy measure, derived from Shannon formulae, provides the indicator of the generation of multiple options between different organisations. This can be calculated at a basic level by examining co-authorship relations and using the entropies of citation statistics in the Shannon equations against geographical locations of research activities. This highlights certain geographical features which seem to be significant in driving synergies: for example, the highways to Amsterdam Airport, or development in the coastal regions of Norway driven by foreign investment. Chapter 6 presents a fully-worked example of these techniques examining the regional economic differences  in Italy, which leads him to conclude that regional policies risk overlooking the significance of inter-regional development. This leads to a powerful political point: “A political administration that is not reflexively aware of and informed about how the relevant innovation systems are shaped may lack the flexibility required to steer these systems and feel in the longer term constrained by the unintended consequences of its own actions”

More specific details of how Leydesdorff calculates the synergy measure are provided in Chapter 7, where an illustrative “Toy model” is presented. Fundamentally, this is a calculation of the entropy of terms in documents. So, for example, one might have a body of academic papers in which the occurrences of key terms are analysed. Across the whole dataset, levels of occurrence and co-occurrence can be calculated in terms of Shannon’s entropy. With this, mutual information can be calculated between pairs of variables, while mutual redundancy can be calculated between 3 or more variables. If the latter calculation produces a number greater than the sum total of mutual information, this indicates a high level of synergy within the system. This presents ways in which degrees of synergy in co-authorship relations can be calculated across regional and national boundaries and comparisons made in relation to specific policy instruments in each case.

In Part 3, Leydesdorff’s focus turns from empirical work to simulation. To do this, he enriches the biological mechanism of autopoiesis of the social system (from Luhmann’s theory), with the work on “anticipatory systems” by Robert Rosen and Daniel Dubois. Anticipation is a biological dimension that is missing from Maturana’s theory, but as others have argued (including Conant and Ashby and Stafford Beer), the projection of future possible states is essential for the effective management of variety in a system. The domain of biological anticipation provides Leydesdorff with an alternative way of thinking about the dynamics of expectation in addition to his empirical analysis with Shannon equations. With the help of Daniel Dubois, who showed how an anticipatory system can be conceived and computed as a fractal, Leydesdorff shows how such a fractal can be simulated to demonstrate the operational principle of anticipation. It seems, if one is to predict future events on the basis of past events, there must be some way of determining a common pattern which connects past to future, and this common pattern must be fractal.

Like all fractals, the pattern which connects past to future is constructed as an interference pattern between different levels of dynamic process. In Dubois and Leydesdorff’s scheme, there are three levels of dynamic: the lowest level concerns the flow of events in time, and the projection of future events on the basis of past events. Dubois calls this “recursion”. The second level concerns the reflexive construction of a model of the dynamics of events, which is called “incursion” (because the model feeds back on itself). The third level involves generation and selection of possible models of the future, which is called “hyperincursion”. Dubois’s mathematics express this in terms of variations of the logistic equation, and drawing on joint work with Dubois, Leydesdorff shows how the interference between these levels can produce anticipatory behaviour.

The simulations that Leydesdorff presents with these equations underlie the fundamental point he wishes to make about dualism between communication and “horizons of expectation”. The latter can be represented and simulated by the Dubois fractals, meaning that Leydesdorff can present a strong case to counter those like Giddens who argue for the impossibility to empirically investigate expectations, or Latour’s homogenising of dimensions. The horizon of expectations is the domain of unrealised possibilities which can be empirically identified through measuring redundancy, and simulated through the equations of Dubois. The dialectical dynamic between different processes can be made manifest and can be investigated. Furthermore, by focusing on the redundancy in the system, he is able to show how the dynamics of absence as causal force (which Deacon has drawn attention to) can be operationalized.

In one of the most profound sections towards the end of the book, Leydesdorff unpicks Descartes’s intention behind “Cogito Ergo Sum”. Drawing on Heidegger’s comment that Cogito Ergo Sum was simply a statement of a “First Principle” rather than any claim about the ontological status of thinking, Leydesdorff argues that the Cartesian dualism between mind and body can be reinterpreted as a dualism between first and second-order contingencies, between the selections of utterances and the codification of meaning. This is a powerful idea which reorients centuries of debate about mind and body: the essence of a Marxist dialectic can be maintained, without postulating a mind-independent reality, or what Bhaskar calls the “intransitive domain”.

Perhaps the most important intellectual attribute one must develop in an increasingly complex intellectual and technological environment is a good “compass”. Compasses are complex technological objects combining practical material properties, calculations and human expectations. Leydesdorff has made his compass out of a combination of practical empirical work, sophisticated simulation and a close reading of philosophy and sociology. What, for example, are we to make of the “Big Data” trend which threatens to swallow up social theory in a mass of algorithmic calculation? Leydesdorff’s compass detects the problem which seems to escape the majority of discourse around data: it is monistic because behind the big data rationale is a proposition that ultimately everything is data. So again, there is reduction of social dynamics to a single causal “agent”. Where then is the dialectical pulse which drives social life? The same applies for those theories which reduce social complexity to biological causes, or to see society as a kind of meta-biology, as Habermas accused Luhmann of doing (with some justification).

Making the book available for free download is very welcome. The printed book is also very nicely produced with good use of colour diagrams and excellent indices and referencing. I can imagine that it will appeal to sociologists, cyberneticians and economists. Taken in its entirety, or in causal reading, there is something new for everyone. There are many intriguing references and original perspectives on other literature which scholars will find extremely valuable. Taken as a whole, it is a magisterial synthesis, which while it makes demands of the reader, will be guaranteed to provide fresh insights and perspectives on current thinking in sociology, technology and economics. Leydesdorff has pursued his logic relentlessly, backed-up both with detailed readings of his intellectual domain, and practical examples of how to use his equations in a meaningful way. This is refreshing when so much of today’s sociological critique resorts to a kind of intellectual “posturing”, or data analysis is insufficiently grounded in an ontology and epistemology. Leydesdorff has shown how a deeper, coherent and empirical approach is possible.


Mark Wiliam Johnson

University of Liverpool, 2021