Reviewed by
Chris Goldspink
P.O. Box 591, Tanunda, South Australia 5352.
IntroductionWhen the opportunity arose to review this book, I leapt at the chance. The title was inviting - relating to areas of research that have held my attention for many years - in particular the relevance of complexity and autopoiesis to social science. Compiled from a selection of over 100 papers presented at the 1998 World Congress of Sociology, Montreal, this text brings together examples of the application of second order cybernetics (the cybernetics of observing systems) to the social sciences. According to the authors, responding to the growing complexity of societies is the major challenge that sociocybernetics is directed towards. In the introduction, the editors argue that an explosion of points of social differentiation and the development of technology and the knowledge revolution present major challenges for the governability of society. The theme of governability was therefore adopted to link the three sections of the volume. These sections are Growing Societal Complexity, Autopoiesis and Observation of Social Systems. The aim of the first section is to establish and describe the challenge or problematic, the second the theoretical tools to tackle it and the final section to raise methodological issues for the study of the phenomena of interest - complex social interaction.
For those readers interested in social simulation the book offers two opportunities. Firstly, the latter chapters provide examples of approaches to simulation. Of these, the chapter by Kluver and Smhmidt (chapter 11) addresses one of the more challenging aspects of simulation work; how to define and specify the relational dimensions of social actors in some space of interaction. Chapter 12 by Dijkum, Lam and Ganzeboom provides an example that, while interesting, pushes no boundaries in social modelling or simulation method. The remaining chapters provide useful triggers to thinking about alternative approaches to social modelling, covering a range of conceptions within a wider systems discipline. In addition, given the focus of sociocybernetics on second order phenomena, there is a useful emphasis on the importance of the reflexive nature of social organisation. This orientation is important as it is very easy to conceive of simulations where the parameters are set by the omnipotent system designer/researcher and far more difficult to close them informationally - the equivalent of placing the observer inside the system observed.
Some General ObservationsIn an earlier work Geyer (1995) traces the lineage of sociocybernetics to General Systems Theory and first order cybernetics. He notes that the advent of second order cybernetics represented a significant transition, but the overall approach of sociocybernetics discussed reveals that a broad net is cast and many diverse (and sometimes inconsistent) systems concepts are appealed to. Geyer uses the term cybernetics to include almost all systems theories, including autopoiesis and complex systems approaches (Geyer 1995). Sociocybernetics therefore runs the risk of suffering the same criticism earlier levelled at General Systems Theory (GST) that it "...pays for its generality with a lack of content." (Jackson 2000). Similarly, Checkland (1984, p. 93) cites Naughton as suggesting that GST was a "...melange of insights, theorems, tautologies and hunches..." Unfortunately this collection, particularly some of the earlier contributions, suffers in just this way.
I found the quality of the chapters quite varied. Several contributions returned to debates that have haunted General Systems since its inception - holism versus reductionism, naturalism versus anti-naturalism, modernism versus postmodernism. These are recurrent themes and as much as we might hope otherwise, they have not yet been adequately laid to rest. The source of my frustration was that, by and large, rather than developing and extending the debate, the presentations simply echoed well established dichotomies.
Despite the claim that sociocybernetics is founded on second order concepts (and hence a constructivist epistemology) the position within and between some contributions was inconsistent with this claim. The contributors move curiously from modernist assumptions to more post-modern ones and in some cases never quite come to terms with the challenges many of the recent developments of systems concepts (particularly complexity theory) pose to cybernetics. Again, surprisingly, contributors take quite different positions with respect to the ontological status of systems. Some reify systems, others assert that they are distinctions made by observers. Overall these more difficult aspects of systems approaches are not consistently or well handled. It is accepted that this is a collection of papers and so the offerings are made to stimulate debate and cannot be expected to be consistent in all respects. Nevertheless these are significant foundational issues and may point to some underlying inconsistencies within the sub-discipline.
The theme of governability was a potentially interesting one. It is relevant to contemporary debate about post-Fordism in organisation science and disorganised capitalism (Reed 1991) and risk society (Beck et al. 1994) in sociology. The theme is however pursued only loosely and at times the thematic relevance of contributions is not apparent. The issue of governability in the context of systemic uncertainty is attracting considerable interest within the wider systems sciences - in particular the recent rapid development of complex systems theoretical positions applied to sociology (see for example Eve et al. 1997). The approaches presented in this collection, however, seldom stray far from a general systems/cybernetic framework and its attendant focus on self-regulating/homeostatic systems. This may suggest excessive adherence to a particular line of argument that may hinder the development of valuable insights, at least with respect to this chosen governability theme. Perhaps the choice of this theme itself is evidence of a preoccupation with and pre-commitment to predictability and understanding the order producing aspects of social systems common to most social theory (Burrell and Morgan 1994 xx) rather than the disordering processes.
The
Contributions and Overall ThemesGiven my personal interests, I found chapter 11 most interesting and it provided me with a good link to the remainder of the content. Rather than address the content in a linear order, chapter by chapter, and on the assumption that readers of JASSS would have a similar interest in the content of this book as I, this chapter is used as the entry point. It provides a way of structuring my discussion of the perceived relevance of the overall book to the simulation endeavour. I will return to the theme of governability chosen by the editors at the end of the review.
In chapter 11 (Social Differentiation as the unfolding of dimensions of social systems) Jurgen Kluver and Jorn Schmidt address issues relevant to social simulation, and in particular the challenges in modelling social relations topologically. They note that there are many ways to define social experience and hence of specifying the relational dimension of any model. They propose an approach that draws principally on the theory of social differentiation to define the fundamental dimensions and thus to classify different social forms (family, tribe, class). Accordingly they argue that a feature of social evolution is an increase in this dimensionality. They conclude that modern societies can be represented using three dimensions as they are horizontally segmented (as with families), vertically differentiated (as with class) and functionally differentiated (as with role). They contrast this with earlier social forms, which they argue, have fewer dimensions.
They note that dynamic modelling of such systems is difficult, as social systems are adaptive. In particular they note that the notional 'rules' governing interaction are under the control of the system itself. An approach for dealing with this, they suggest, is to approach social systems as sets of social actors whose interactions are determined by specific rules, which generate the system dynamics. The trajectories the system describes in its state space are "nothing else than the intended and unintended consequences of social actions." Here then social behaviour can be modelled as a consequence of both rational actions by individuals (acting in response to the rules) and the product of complex organisation of the system. This captures a concern of interest to complexity researchers - the fact that interactions between the rational and structural aspects of social systems have not hitherto been well dealt with. McKelvey for example (1997, p. 7) identifies four sources of order in the natural and social world, these are:
He argues that order in social systems has tended to be seen as originating exclusively from rational order.
Kluver and Schmidt further note that the reflexive closure or capacity for social systems to self-organise and adapt their response capability as a part of that self-organisation, implies that such systems have not one set of rules of interaction but at least two - the second set comprising meta rules "by which the rules of interaction - the base rules - are changed..." As a consequence of these capabilities and of defining them in this way they argue that "social evolution occurs by varying, eliminating, and enlarging social rules". They consider these rules to be the 'gene' equivalents of social evolution. The authors briefly outline an approach to modelling such a system using Cellular Automata to represent the 'real system' or base rule interactions and a Genetic Algorithm to provide the meta level capability.
This model generates some very interesting behaviour. The authors argue that while the model is very simple, there is a mapping between the behaviour generated and observations of real social systems. These include the fact that more differentiated social structures are more sensitive to perturbation (less intrinsically stable) or, looked at another way, have greater requisite variety and hence greater adaptive and survival capability. They note, however, that getting to this state is "difficult and rare". They argue that "modern societies obviously have to pay for their adaptive efficiency with permanent unrest". Further they observe that in their model, increasing dimensionality - increases in social stratification (class) or functional differentiation (role) - disrupts cohesion at other levels (e.g. family). Hence traditional social forms are disrupted as society increasingly differentiates in the other dimensions. Importantly they suggest that such characteristics are an emergent property of intrinsic social organisation and not the result of first order mechanisms such as human nature (biology) or social interests (politics). They conclude also that any search for fundamental rules (i.e. laws) of social interaction as with natural science will necessarily fail as social rules are constantly changed. However, they consider that the current search for higher order descriptive mechanisms such as rules of self-referentiality may prove more fruitful. Here there is an implicit link to other work in the volume and in particular the work of those concerned with autopoiesis.
There is a significant divergence of opinion about whether social systems constitute third order autopoietic systems or whether they are systems simply comprised of autopoietic unities. Stafford Beer in the preface of Maturana and Varela's Autopoiesis and Cognition: The Realisation of the Living (Maturana and Varela 1980), suggests that they meet the relevant criteria for autopoiesis although in this reference (originally written in 1973) the authors appeared unclear or uncertain. Subsequently there has been a divergence of view, with Varela (1981) firming his opposition to treating social systems as autopoietic in and of themselves. Bednarz (1988, p. 61) summarises the overall problem as follows:
"The attempt to extend autopoiesis to the social domain has failed so far precisely because it has one foot in each camp. The process of an autopoietic system cannot belong to one domain while its components belong to another. But this is what occurs when social systems are regarded as being composed of human beings."
His conclusion is that if we are to resolve the conceptual difficulty, we need to cease considering human beings as the components of social systems and instead view social systems as constituted by interrelations between humans. This leads to the approach to social autopoiesis adopted by Luhmann and it is the Luhmannian approach that is most commonly appealed to by contributors to this volume. For Luhmann (1990), it is the linguistic domain (communicative acts) which gives rise to social relations and which in turn are constituted in and through the social domain. This domain is dependent on the pre-existence of the biological and becomes possible only when sufficient complexity is available at the biological level consistent with Maturana and Varela's views on the emergence of language. The 'meaning structures' which arise from communicative acts are referentially closed and self-producing. All non-communicational things and processes belong to the environment. Note that what is pointed to here is a functional closure rather than a physical one, a point which has a bearing on the concept of boundedness that is central to autopoiesis. Significantly, when it is a functionally defined unity that is self-produced, the distinction between autopoietic and simply viable seems to break down. In other words, all functionally autonomous unities are, by definition, self-producing: the categorical distinction collapses for this class of system. This is what Varela (1981, p. 38) clearly identifies when he says of social systems:
"Such units are autonomous but with an organisational closure that is characterisable in terms of relations such as instructions or linguistic agreement."
It is for this reason also that Hejl (1984) distinguished between self-maintaining systems and self-referential systems. Functionally autonomous unities are abstract (observer defined), they can be approached as self-referential but as they do not 'self-produce' in a physical domain they should not be regarded as autopoietic. Thus both Varela and Hejl identify social systems as belonging to the broader class of autonomous, operationally closed and self-organising/self-referential systems but not as autopoietic. Further, if this argument holds the concept of autopoiesis only offers new insight into systems that do self-produce in a physical domain, i.e. biological systems as per the genesis of the concept. In relation to other classes of system the concept of operational closure and self-organisation are sufficient and equivalent. Along with Robert Kay, I have argued elsewhere that autopoiesis provides a foundation for understanding how human social action is constrained by our intrinsic biological character (Goldspink and Kay 2002).
In chapter 6 (Information, meaning and communication: An autopoietic approach) Mingers further adds to the debate about the possibility for and legitimacy of social autopoiesis. This is a debate to which he has been a significant contributor over some years (Mingers 1991 and 1995). In this chapter, Mingers reaffirms his earlier concerns about treating social systems as autopoietic but argues that the conception has some relevance because it compels us to reconsider the role and nature of communication. In this Mingers appears to be attempting to reconcile alternative approaches to social autopoiesis. Despite pointing out that Luhmann's concept invokes quite different social agents (communicative acts rather than people) he tries to work with this concept. This involves him in working under two alternative (and in my view incompatible) ontologies and leads to a presentation which I find somewhat less clear and satisfying than much of his earlier work. Here his attention is on the relationship between 'information' and 'meaning'. He begins by rearticulating a proposed typology of organisationally closed or self-referential systems. This typology terminates at level seven with a consideration of the relational characteristic of what Mingers calls self-conscious systems or the 'embodied individual'. This distinguishes human agents and hence acts as a starting point for the examination into social systems and communication that follows. The main contribution of this chapter involves introducing three additional levels of social organisational closure, The Social Individual, Social Networks and Society/Organisations. These are discussed in terms of their components, structural relations, mode of closure and emergent properties.
Focusing on the first level (that of the individual) Mingers reinforces the argument presented by Maturana and Varela that information can only act as a trigger for any individual and hence cannot determine the state of their nervous system. He introduces a distinction between 'information' and 'meaning' arguing that a sign constitutes a "complex analogue stimulus" to the nervous system which is then "progressively transformed through a process of digitalisation" generating meaning for a particular individual. It is not clear why the terms "analogue" and "digitalise" are introduced here nor how they are to be interpreted - does the process of digitalisation refer to the neurological process? Information, Mingers argues, is ever present while "human consciousness only ever exists in a domain of meaning". This meaning is embodied i.e. captured in the physical structures of the body and nervous system. It does not exist as "pure thought". While trying to stay clear of the minefield of self-referential systems concepts, Mingers draws on Luhmann's concepts of society as communicative action to discuss the social individual. This discussion is rather convoluted and in my view this level is better dealt with, as Varela does, through the studious avoidance of concepts of communication, information and meaning. Invoking these concepts invariably leads to a potential for confusion between the subject and object of the discussion. To discuss 'meaning' in this context, for example, immediately risks reification or leaves unclear who or what is making the attribution of 'meaning' - meaningful for whom. It seems to me that Mingers does not entirely succeed in disentangling these issues.
At level two - social networks - Mingers observes that with recurrent interaction between individuals, structural coupling occurs and there arises a circular co-determinacy between the emergent structures of interaction and the structures of the individuals which give rise to them through their networks of interaction. Mingers makes the important observation that individuals participate in many such networks and that these networks notionally intersect in and through their common members. This is a theme dealt with by Hejl earlier (1993) and as I have argued elsewhere, has very important implications for thinking about the dynamics of social systems (Goldspink 2000 and Goldspink and Kay 2002). Mingers argues for attending to 'membership' determined by the emotion of acceptance and rejection as a basis for the bounding of social interaction.
The focus of this chapter is an important one, trying to tease apart and understand the relationship between the physical (biological) phenomena and the non-physical (emergent social), and to deal with the linkages between levels of phenomena. For me the treatment offered here is not ultimately successful or lacks sufficient development and refinement to be convincing.
In chapter 5 (On the interpenetration of social subsystems) Michael Rempel examines the social differentiation theories of Parsons and Luhmann. Rempel opens by arguing that "In general, deliberations in one institution [he gives the examples of politics and law] seem increasingly to incorporate social and technical influences rooted historically in others" (p 89). He argues that there is therefore a need for a better theoretical approach that deals with such interpenetration in the context of social differentiation theory. This further develops the arguments about interpenetration of social domains mentioned above. Rempel summarises the difference between Parsons' and Luhmann's approach to understanding structuration as follows: "Whereas Parsons defines a social system to consist of socially structured actions, Luhmann defines it to consist of conceptually structured meanings" (p. 91). These two approaches (Rempel argues) have complementary weaknesses and on this basis he proposes a synthesis of the two. The problem is that the classificatory formalisms of both theories make it difficult to examine the implications when agents of functionally differentiated units increasingly participate in networks of processes which are not functionally differentiated but rather which bring together functional specialisation to produce outputs. The proposed dual focus which analyses both action systems (as with Parsons) and communication systems (as with Luhmann) will increase the likelihood of appreciating processes of interpenetration in any social structuration research.
In chapter 7 Lucio Biggiero asks Are Firms Autopoietic? Biggero opens this chapter by noting that interest in autopoiesis has been greatest outside biology (its field of origin). He argues that the debate about autopoiesis has distracted attention from the relevance of many of the underpinning cybernetic concepts such as self-organisation and autonomy which lie at the core of autopoiesis but are not restricted to application within it. This again reflects the earlier observation by Hejl, that the concept of autopoiesis and operational closure collapse when applied to non-physical phenomena. Biggiero goes on to illustrate, drawing on a diverse range of organisational theory, many areas that lead to a need to question the legitimacy of seeing firms or organisations as autopoietic. This is done, not as Mingers does in his earlier work, by comparing the foundational concepts of autopoiesis with the intrinsic nature of social systems, but rather by a looser comparison between the implications of extant organisation theory and the implications of a notional social autopoiesis. It serves to highlight the contradictions and difficulties of reconciling autopoiesis with most contemporary theory but is less effective than alternative arguments at clearly disqualifying autopoiesis for such an application.
Again, as with Mingers, Biggiero several times notes the significance of social actors (i.e. people) belonging to more than one social system at the same time. This suggests (he argues) that social systems are not closed as suggested by the theory of autopoiesis. He further argues that given that social systems are not substantive (but are brought forth through acts of distinction) their closure and boundaries are notional or matters of degree rather than being absolute. He goes on to argue that second order cybernetics furnishes the necessary analytical tools for dealing with such systems and has no need for social autopoiesis.
In chapter 8 (The autopoiesis of social systems: An Aristotelian interpretation) Colin Dougall further contributes to the autopoiesis theme by engaging with the controversial issue of social autopoiesis. Seeking a resolution he proposes what he calls the M-A model as a basis for generalising the concept of autopoiesis and disconnecting it from its biological roots or tie to life. He notes that social models of autopoiesis commonly fail "because social systems do not meet the formal requirements of the theory". Dougall usefully reconnects autopoiesis to related ideas, both in contemporary sociology and earlier metaphysics. In so doing he surfaces issues and weaknesses in alternative conceptions. His M-A model is a more abstract rendering of the defining characteristics of systems that are self-producing and is constructed drawing on both Aristotlean and Maturanean influences. Dougall then argues that Maturana and Varela's biological theory is an instance of this more abstract model and that what is sought by social scientists is a derivative suited to social systems - a role the organic variant cannot fill. He hints at the form such a model may take but does not go on to describe the specific characteristics nor to compare it with the organic except by way of the common root characteristics captured in the M-A model.
In chapter 9 (Autopoiesis and governance: Societal steering and control in democratic societies) John Little discusses the relevance of autopoiesis concepts to the theme of governability. He notes that we have increasingly had to confront the limited efficacy of government as a basis for social control, observing that "In complex networks governance is a matter of autonomous self control and not top down steering from a central position" (p. 160).
He expresses the opinion that as well as needing to address the influence potential of government on society there is growing concern about the influence potential of citizens over the government which claims to represent them. In developing this theme he explores the alternative implications of Luhmann's theory and that of Peter Hejl. Luhmann's ideas, he concludes, lead to a pessimistic prediction for the possibility of improved democratic governance. He argues that this theory suggests that top down structural change will not make administration more democratic. He argues further that as complexity grows through, for example, greater social plurality, governmental systems must respond by increasing their organisational complexity thus further reducing the possibility for responsiveness and control from and by citizens. By contrast, Little argues that Hejl's theory does suggest the possibility for influence over complex systems and suggest that such influence requires an intimate knowledge of system possibilities, and that this may best come from within. This, he argues, is at odds with recent reforms influenced by the so-called New Public Management (Pollitt and Bouckaert 2000) or managerialism which, drawing on management practices from commercial enterprises, tends to reduce citizens to 'customers' belonging to some general category rather than engaging with them as individuals. He concludes by arguing that the implication of adopting the perspective of either Hejl or Luhmann is that small units of government interacting directly with citizens are the best way forward.
Little suggests that Luhmann's and Hejl's perspectives are potentially complementary focusing on different levels of analysis (macro for Luhmann and micro for Hejl). This micro-macro issue is taken up in chapter 4 by Robert Artigiani as he discusses The emergence of societal information.
Artigiani reminds us that the modernist venture was atomistic - attempting to reduce explanation of macro phenomena to micro order. Here he begins to tap into the longstanding micro-macro problem that bedevils social science (Coleman 1994 and Smith 1997). He argues that "to make progress towards answering Big Questions, a new paradigm respecting the integrity of qualitatively different levels of being is needed." This seems to be an appeal to move from a dichotomous ontology (real/not real) to one based on levels (Broad 1925, Newman 1996, Emmeche et al. 1997 and Schroder 1998) and recognising that emergent phenomena have ontological status. This possible theme is not developed, however. Artigiani reveals that his concern is not so much for the development of an explanatory framework that can deal adequately with level transition but rather but for emergent phenomena "to be treated according to rules appropriate to their level of reality, rather than analysed in terms of material atoms". Sociology already arguably adopts this (inadequate) solution to the micro-macro problem - positing perspectives and concepts which describe some chosen behaviour but failing to develop an adequate response for explaining how phenomena at one level generate those at another.
Artigiani draws on Prigogine's thinking and makes appeals for insights from thermodynamics. This is somewhat problematic as thermodynamic systems are a very limited class that are closed with respect to energy transfer. Social systems are manifestly not of this class. They can and have been argued to be informational closed. Artigiani does not flag this distinction nor address the implications of different forms of closure and this is unfortunate, as it would seem fundamental to thinking about the basis of and mechanisms for self-organisation in social systems. Drawing on natural (thermodynamic) systems can at best be relevant as loose metaphor.
According to Artigiani "In societies, human behavioural choices depend on rules appropriate to their emergent level of reality, and these rules are moral rather than biological" (p. 79). These rules, he argues, are stored in the systems themselves i.e. in the interrelationships between people. This idea is important - that social structure captures 'information' relevant to social viability and thus the survival of the individuals that comprise it. There are some concerning aspects of his application of selectionist ideas here however. The author argues thats selection "acts on social systems" but the mechanism is unspecified. I found the handling of these matters is too simplistic and not sufficiently based in a knowledge of the other disciplines drawn on.
In chapter 2, Walter Buckley takes on the issue of the relationship between mind and brain and sets out to establish a A Dynamic Systems Model. This is another hot topic that is relevant to the micro-macro theme. It is critical to developments in AI and in computer simulation of social processes (Kennedy and Eberhart 2001). In the introduction to this chapter Buckley proposes to set out a model of "mind brain interaction and the continuous real-time generation and maintenance of consciousness and mental events in terms of the organism-environment interaction." He proposes to set out the main features, provide some empirical evidence to support it and deal with some of the philosophical issues. Essentially Buckley is arguing for a view of mind as emergent phenomenon - a phenomenon that arises from the dynamic between the nervous system and the environment. Accordingly he argues against the value of locating mind in the brain or rather of reductionist attempts to argue that the two are the same. This is a reasonably well established line of argument and it is surprising that the author does not locate his own position in the context of others who have mounted the same or similar arguments. Given the focus of the volume, Maturana and Varela's constructivism (Maturana and Varela 1980) and the more recent enactive position of Varela (Varela et al. 1991) spring immediately to mind. Also, connectionist approaches to cognition link strongly to complex systems approaches such as those of Paul Cilliers (1998) and it would have been useful to see these links established and explored. Buckley draws attention to the tendency for many of his contemporaries to assign lesser ontological status to dynamic processes and this insight could also be extended to emergent processes which I am sure was his intention. These two classes of process are not the same - the latter being phenomena of a different logical type which results from non-linear interactions. The former includes linear interactions and results in phenomena at the same logical level (i.e. motion) albeit of a different class.
In the end the author recapitulates some of the observations from recent neurological studies and discusses the veracity of viewing mind as emergent. He notes that "it is not clear why this is problematic for some since it seems intuitively obvious..." The promise of a model is not forthcoming; rather there is a wide-ranging and somewhat superficial recapitulation of fragments of the argument for this perspective. The author often seems to be ranging into the work of disciplines with which he is only marginally familiar - drawing on reports that support his general argument. It is unfortunate that he does not focus on developing and articulating a model with some prospect for testing and advancing the development of this approach rather than another loose recapitulation of the need for it. The reason that there is concern about viewing 'mind' as emergent is again that there is inadequate treatment of how micro-level (neurological) activity gives rise to the range and type of macro phenomena we may classify as 'mind'. In other words, as with the micro-macro problem in social theory generally, claiming something as emergent means that we can describe it but not explain how it is generated from the interaction of the micro agents which give rise to it. That said, the principal conclusion is fair enough - Buckley asserts that adequate analysis (of mind) must focus on the total system of organism and environment as a complex and on-going dynamic whole.
MethodologyIn the final chapter titled Towards a methodology for the empirical testing of complex social cybernetic models van der Zouwen and van Dijkum argue that social applications of cybernetic systems methods have become more sophisticated. As a consequence the complexity of the models has become such that it has become increasingly difficult to test them empirically. It is this that has led to a growing interest in computer simulation as a means for exploring the behaviour of models and as a basis for comparing that behaviour with real world phenomena. In addressing the question of how researchers might approach evaluation of complex sociocybernetic models they set out what they argue are the defining characteristics of social systems.
The working definition proposed is that a social system is "a system in which actors, their actions and/or their communications" form the elements. These elements interact and the resulting system is separated from its environment by a boundary. My first observation about this definition is that it hides a great deal that is important. Real (biological) actors are very different types of things than 'actions' or 'communications'. The latter are distinctions made by observers. In suggesting that a social system can be comprised of one or more of these different types of things, the definition simply papers over some of the most important debates about the constitutive nature of social phenomena. Similarly, the assertion of the necessary existence of a boundary leaves open or fails to address the fundamental question - what form of boundary? As Mingers argues in his critique of those who advocate social autopoiesis, social system boundaries are observer relative and lack the ontological standing of physical boundaries found in, for example, biological autopoietic systems. Hejl also has had quite a bit to say about this notion of boundedness in social systems.
Moving on from the base definition, van der Zouwen and van Dijkum identify the first defining characteristics of social systems as their 'openness'. This reflects the longstanding assertion of general systems theorists, that social systems are open. In this context it is a remarkable assertion given that, in failing to specify open with respect to what, they potentially take a line at odds with autopoietic theory that is clearly embraced within sociocybernetics. A defining characteristic of an autopoietic system is its informational closure.
The second defining characteristic is that delays in input/output transformation may occur. This they note is a source of non-linearity. Strange then that the rapidly advancing field of non-linear systems research, which has a great deal of relevance to understanding how such systems may as a consequence behave is almost entirely absent from all work presented in this book. This is further evidence that many working in sociocybernetics are reluctant to venture far from traditional cybernetic concepts. This is reflected in the third defining characteristic - homeostatic/goal-seeking behaviour. In asserting this defining characteristic, there is no treatment of the important distinction between teleonomic and teleological goal seeking. A failure to adequately distinguish between whether implicit or explicit goal directedness is being implied is all too common in social applications of systems theory and is vital if the issue of the 'observer' is to be adequately handled. The fourth characteristic is the possibility for positive feedback loops and this is also noted as a source of non-linearity but no reference to the wider implications of this assertion is made.
The fifth characteristic is the presence of reflexivity and anticipation via feed forward loops. This is a vital aspect and its recognition is to be commended as it is arguably an aspect of social systems inadequately dealt with by the models being derived from complex systems approaches as well as those being used by many simulations. Simulating the implications of reflexivity in social agents is rare. The sixth and final characteristic is that of 'goal adaptation and morphogenesis' - the capacity of social systems to change theory, goals and/or structure. There is not a great deal of discussion of this characteristic but there is apparently implicit acceptance of a definition of social systems as goal seeking systems without again clarifying if this is teleonomic or teleological or both. The observation (raised by autopoietic approaches and complex systems theoretical approaches) that order may arise without any explicit, rational or goal directed behaviour but as a consequence of recurrent interaction is left unexplored. All in all a great deal that is important is left unexamined both within the definition and in setting out the defining characteristics.
There then ensues a discussion of the problems with empirical validation for social hypotheses using the hypothetico-deductive method. More usefully, the authors propose a methodology (based on what they refer to as a sophisticated interpretation of Popper's falsification principle) for the validation of models. Here it is argued that a useful approach to model testing is to test two models against the real world data and to establish which best explains the data. They argue further that there is a need to consider and test for fit - i.e. where there is a match between the real world data and that generated by the model and the absence of falsifiers - i.e. states that would be forbidden given the theory. Most of the subsequent elaboration is for the use of linear models. Only brief account is taken of issues for non-linear modes and this mainly involves observing that tests for such models are in their infancy. This is true but not particularly helpful. Given that two defining characteristics that the authors present are potent sources of non-linearity and that a premise of the chapter is that the greatest challenge to validation comes from increasing model complexity, the development of methodologies which can deal with this aspect of social system research are particularly important. The theme has been developed elsewhere by McKelvey (McKelvey 1999) and by myself (Goldspink 2002).
Chapter 10 (Implications of autopoiesis and cognitive mapping for a methodology of comparative cross-cultural research) by Bernd Hornung and Charo Hornung addresses the possibility for cross-cultural social research. The authors briefly discuss a range of conceptions that alternatively suggest the impossibility of such research (due for example to cultural relativism and the lack of a common grounding point) or the suggestion that it should be feasible. The problem centres on the possibility for ontological claims that are not culturally specific. While they introduce concepts (including autopoiesis) which they believe have some conceptual relevance to this topic, the nature of the relevance is addressed only sketchily and it is unclear what their central argument is. In some cases it is also unclear why the concepts were introduced as few clear conclusions about their relevance are drawn. In the end, they draw rather eclectically on these alternative conceptions to propose a methodology for cross-cultural research. This is unsatisfactory as it is not presented as a synthesis and the inherent philosophical compatibility issues between the approaches drawn on are not addressed. Their proposed 'hermeneutic circles' are supposed to allow identification of conceptual and functional equivalence between cultural contexts. They argue quite reasonably that only once such equivalence has been established can more conventional research methods be applied to compare evidence from the two or more cultural contexts. Clearly they have derived a general heuristic for attempting to isolate such equivalence from a variety of systems related approaches. These may form practical tools for any researcher wanting to work in cross-cultural research. To call this heuristic a methodology seems, however, to be going too far in that the conceptual tools drawn upon (while derived from systems ideas) are not necessarily theoretically compatible.
Chapter 12 by van Dijkum, Lam and Ganzeboom provides an example of a simulation to investigate the dynamics of educational expansion. Based on an assertion that very few aspects of social evolution demonstrate pattern, with the exception of the upward expansion of educational levels, the authors set out to explain this phenomena. They suggest that existing arguments in which educational expansion is driven by demand for more educated labour do not stand up to critical scrutiny and posit an independent mechanism. They argue that micro decision making (by rational actors) on grounds of job competition and/or status seeking can explain the phenomena. In this sense, educational expansion is an unintended consequence rather than a direct effect. They present a simulation to model this relationship and compare the results with data on actual expansion in the Netherlands over the past century. The simulation uses first order cybernetics concepts to model the relationships and dynamics. The model comprises three sub-systems, a population sub-model, a choice sub-model and an education sub-model.
This model provides a useful example of the way in which simulation can be used for theory testing where there are good sources of data for both calibration and testing. The model allows for active experimentation, allowing modifications of the theory derived rules to test their potential for explaining the real world data. That said the model pushes no technical boundaries and doesn't serve to advance knowledge about the more challenging aspects of social system modelling.
GovernabilityIn chapter 1, Paris Arnopoulos sets out his central thesis which is that some social control is "necessary, possible and desirable". He points in particular to the difficulties of controlling natural and cultural systems given their intrinsic complexity. He suggests as guiding principles the need to act "humbly, carefully and responsibly". Post modern sociocybernetic strategies, he argues, are appropriate as they balance the "libertarian and totalitarian extremes". But this is about as far as he goes with addressing this central issue. Instead of developing a well-focused argument we are confronted with an eclectic mélange - a string of bold assertions almost all of which would warrant some defence but none of which are afforded one. The issue of the desirability, necessity and possibility for social control is not taken up with any rigour - indeed there is little argument presented in any form on any topic. The author claims to take a realist line, asserting natural and cultural systems as 'isometric'. What ensues is an example of what Khalil and Boulding (1996) call identificational slips - associating or seeing as related, disparate phenomena on the grounds of a superficial resemblance. In this case it takes the form of suggesting a homologous relationship between a (long) list of natural science concepts and social phenomena where even metaphorical association would be stretching a point. Hence we have the suggestion of 'sociomass' and 'social inertia' and later even 'socio-sclerosis'. There is no attempt to justify the suggested homology or to argue for it. The concepts spanned include those derived from various positions, both Newtonian and complex systemic, with no recognition that these may not be incremental developments but rather are founded on incompatible assumptions. In short all grist to the anti-naturalists mill.
Arnopoulos shows little awareness of the debates and sensitivities of the many disciplines through whose which he wanders. For example, he appears to take a progressionist position about evolution on several occasions e.g. "man [sic] is the paragon of animals and the highest stage of organic evolution." There are few references in the chapter - suggesting an ignorance of (or a disregard for) the contributions of others - or perhaps a desire not to be confused by the many competing and alternative arguments that populate the areas where he so blithely strolls. Concepts and ideas are thrown almost at random into a pot and stirred resulting in a thick cloud of assertions. For example "Societies are complex self-organising adaptive systems which value creation and propagation. Since they are precariously balanced on the cosmos-chaos boundary, autopoietic systems evolve by selection and mutation, convergence and divergence. Thereby order can emerge spontaneously by homeostatic convergence of various factors." Well I am glad that's settled then! Although that 'various factors' looks a little under specified.
It is not that there are no issues to be addressed here and many interesting ideas are approached but unfortunately only briefly and tangentially. For example, Arnopoulos points to the problem of increasing global complexity (like Beck et al. 1994) remarking on the rapid increase in transnational issues such as pollution and social dislocation and observing that they are beyond the control of national policy. He highlights global moves to establish frameworks of fundamental human values that allow scope for local adaptation but provide a core set of backdrop principles thus hinting at a possible response to the problem. But on the whole opportunities for effective argument are lost or passed up.
Heinrich W. Ahlemeyer approaches the issue of governability from the perspective of organisations rather than wider social systems. The author opens chapter 3 (Management by complexity: Redundancy and variety in organisations) by noting that complexity has generally been seen as a problem to be addressed rather than as a potential solution. He sets himself the task of looking at how organisations can use complexity. To accomplish this he adopts an alternative focuses on organisational redundancy and variety. The author takes a second order cybernetics perspective in asserting that the complexity of systems rests on a distinction made by an observer. He then outlines a set of criteria an observer may use to make this distinction.
According to Ahlemeyer, when approaching an organisation as a system, the observer notes that the system is neither completely ordered or disordered but contains both redundancy (order) and variety (disorder). Ahlemeyer observes that variety in organisations "grows by increasing the range and heterogeneity of decisions". Here he draws primarily on Luhmann in positing social systems as systems of communication. Hence the dimensionality (heterogeneity) refers to communicative acts rather than individuals. He further states that a system can be observed as complex "when it contains more elements than can be connected completely". This may be a pragmatic definition rather than a technical one as it is not consistent with definitions proposed by others. For example, Kauffman (1993) describes systems where K=N (every element is connected to every other) as maximally complex in his experiments with Boolean networks. Using this definition, a incompletely connected network is less complex than a completely connected one with the same value of N. This problem may arise due to a confusion between structural complexity (i.e. a constitutive characteristic of the observed system) and the uncertainty associated with incomplete information about the system on the part of the observer. In other words the author may be arguing (consistent with his stated epistemology) that an observer will perceive a system as complex when he or she has an incomplete description of it. To say something about the constitutive nature of such a system would be to make ontological claims of a type with which Ahlemeyer may be uncomfortable. He argues that "complexity enforces a selective connection" and that "There is always a selection from a range of possibilities and this selection is made by decision". This suggests that the constraint on governability is the bounded rationality of the observer. Incompleteness and patchy connectivity arise as managers make 'boundedly rational' choices in the face of a wide range of possibilities and limited information that constrains their capacity to choose. From this perspective Ahlemeyer seems to be arguing that complexity arises from limits to rationality. This is a very different position than that taken by many complexity researchers who assert intrinsic unknowability as an aspect of ontology rather than epistemology. Given the position that Ahlemeyer takes one can ask whether complexity would cease to be complex from the perspective of a hyper-rational being. If so this position is suggestive of reductionism in disguise. It suggests that the universe is in principal knowable and predictable. In other words this is a restatement of the position that emergence is a product of limited knowledge. (See Gilbert 1995 for a discussion of this point.)
The author sums up as follows and raises some interesting issues:
"If complexity enforces a decision by taking a selection, organised social systems themselves are a solution of the problem of complexity as both their existence and their elementary operations are based on decision making. Only by drastically limiting the range of alternative possibilities, can organisations come into existence. Their continued operation demands the recursive production of decisions with an ongoing reference to former decisions."
This is an argument that organisations operate so as to reduce complexity. This makes sense in that the patterns of recurrent interactions which allow us to distinguish an organisation are a reduction of complexity compared to completely unordered interactions we may encounter with the same individuals in another context - complexity is collapsed as behaviours become correlated. The reduction is however internal i.e. experienced by those who participate. This can also impact on the environment as systems with which the organisation interacts are changed by the encounter. This may disrupt the patterned dynamics of either or both or it may lead to another order of coupling and hence complexity reduction. If the coupling becomes too close (high co-adaptation) then as with over evolved biological systems their vulnerability to contingency increases - it can be absorbed only within limits. When such limits are exceeded the contingent event may trigger failure of the organisation. This is essentially the argument endorsing the need for requisite variety, a debate that complexity research approaches in the form of the edge of chaos argument (Bak 1996). With respect to most organisations, Ahlemeyer notes that the market is the environment that tests the validity of the complexity reduction choices (decisions). Markets involve uncontrollable complexity from the perspective of the firm. The approach to managing this complexity is to treat it as risk. The strategies the author notes as having been adopted to increase organisational complexity include the following:
These are all consistent with the theoretical work of Kauffman and Macready (1995) on the need for 'patching' and link strongly to post-Fordist advocacy of the need for greater organisational flexibility. As a check on the advocacy of such change Ahlemeyer notes that "Many organisation members - management and employees alike - feel overrun by rash and radical changes. They feel they cannot cope and they feel rendered superfluous. They are vulnerable and distraught; many have lost their orientation." In other words he cautions us to attend to the effects of alternative approaches to achieving such changes in configuration as they invariably create winners and losers.
ConclusionWhile this book revisits many recent and long standing themes in social research and there are some chapters which make a useful offering, the overall reach of its contribution as a whole is not great. There is certainly much more substantial and deeper material readily available in the journal literature. The convenience of combining several contributions in one book is somewhat nullified by their variable quality. The book lacks a well-integrated theme and the disparate (and often contradictory) positions taken by contributors (along with some who fail to locate themselves in the wider debate) limit the value of the collection to anyone new to the area and wanting to gain a good overview. The lack of depth and failure to push the boundaries or contribute significant new thinking similarly limits the value of the book to established researchers in these areas. It is therefore difficult to locate a clear audience for it.
ReferencesBAK P. 1996. How nature works: The science of self-organised criticality. Copernicus, New York, NY.
BECK U., A. Giddens and S. Lash 1994. Reflexive Modernisation: Politics, Tradition and Aesthetics in the Modern Social Order. Polity, Cambridge.
BEDNARZ J. 1988. Autopoesis: The organisational closure of social systems. Systems Research, 5:57-64.
BROAD C. D. 1925. The mind and its place in nature . Routledge and Kegan Paul, London.
BURRELL G. & Morgan G. (1994), Sociological Paradigms and Organisational Analysis, Virago, London.
CHECKLAND P. 1984. Systems Thinking, Systems Practice. John Wiley, Chichester.
CILLIERS P. 1998. Complexity and postmodernism: Understanding complex systems. Routledge, London.
COLEMAN J. S. 1994. Foundations of Social Theory. Belknap/Harvard University Press, Cambridge, MA.
EMMECHE C., S. Koppe and F. Stjernfelt 1997. Explaining emergence: Towards an ontology of levels. Journal for General Philosophy of Science, 28:83-119.
EVE R. A., S. Horsfall and M. E. Lee, editors, 1997. Chaos, Complexity and Sociology: Myths, Models and Theories. Sage Publications, Thousand Oaks, CA.
GEYER F. 1994. The challenge of sociocybernetics. Kybernetes 24:6-32.
GILBERT N. 1995. Emergence in social simulation. In N. Gilbert and R. Conte, editors, Artificial Societies: The Computer Simulation of Social Life. UCL Press, London.
GOLDSPINK C. 2000. Social attractors: An examination of the applicability of complexity theory to social and organisational analysis. Social Ecology, Unpublished thesis, Richmond, University Western Sydney: 312.
GOLDSPINK C. 2002. Methodological implications of complex systems approaches to sociality: Simulation as a foundation for knowledge. Journal of Artificial Societies and Social Simulation, 5, <https://www.jasss.org/5/1/3.html>.
GOLDSPINK C. and R. Kay (2002). Bridging the micro-macro divide: a new basis for social science. Unpublished.
HEJL P. M. 1984. Towards a theory of social systems: Self-organisation, self-maintenance, self-reference and syn-reference. In P. Ulrich, editor, Self-Organisation and Management of Social Systems: Insights, Promises, Doubts and Questions. Springer-Verlag, Berlin.
HEJL P. M. 1993. Culture as a network of socially constructed realities. In A. Rigney and D. Fokkema, editors, Cultural Participation: Trends since the Middle Ages. John Benjamins Publishing Company, Amsterdam.
JACKSON M. C. 2000. Systems Approaches to Management. Kluwer Academic Publishers, London.
KAUFFMAN S. A. 1993. The Origins of Order: Self Organisation and Selection in Evolution. Oxford University Press, New York, NY.
KAUFFMAN S. and W. Macready 1995. Technological evolution and adaptive organisations. Complexity, 1:26-43.
KENNEDY J. and R. C. Eberhart 2001. Swarm Intelligence. Academic Press, London.
KHALIL E. L. and K. Boulding 1996. Evolution, Order and Complexity. Routledge, London.
LUHMANN N. 1990. Essays on Self Reference. Columbia University Press, New York, NY.
MATURANA H. and F. Varela 1980. Autopoiesis and Cognition: The Realisation of the Living. D. Reidel, Boston, MA.
MCKELVEY B. 1997. Quasi-natural organisation science. Organisation Science, 8:351-380.
MCKELVEY B. 1999. Complexity theory in organisation science: Seizing the promise or becoming a fad? Emergence, 1:5-32.
MINGERS J. 1991. The cognitive theories of Maturana and Varela. Systems Practice, 4:319-338.
MINGERS J. 1995. Self-producing systems: Implications and applications of autopoiesis. Plenum, New York, NY.
NEWMAN D. V. 1996. Emergence and strange attractors. Philosophy of Science, 63:245-261.
POLLITT C. and G. Bouckaert 2000. Public management reform: A comparative analysis. Oxford University Press, Oxford.
REED M. I. 1991. The end of organised society: A theme in search of a theory? In P. Blyton and J. Morris, editors, A Flexible Future: Prospects for Employment and Organisation. Walter de Gruyter, Berlin.
SCHRODER J. 1998. Emergence: Non-deducibility or downwards causation? The Philosophical Quarterly, 48:434-452.
SMITH T. S. 1997. Non-linear dynamics and the micro-macro bridge. In R. A. Eve, S. Horsfall and M. E. Lee, editors, Chaos, Complexity and Sociology: Myths, Models and Theories. Sage Publications, Thousand Oaks, CA.
VARELA F. 1981. Describing the logic of the living: The adequacy and limitations of the idea of autopoiesis. In M. Zeleny, editor, Autopoiesis: A theory of the living organisation. Elsevier - North Holland, New York, NY.
VARELA F., E. Thompson and E. Rosch 1991. The embodied mind: cognitive science and human experience. The M. I. T. Press, Cambridge, MA.
Return to Contents
of this issue
© Copyright Journal of Artificial Societies and Social Simulation, 2003