Order this book
School of Applied Social Sciences, Durham University
Jörg does not mention simulation at all. I read through the book twice and checked the index. Simulation stands out by its absence. This is significant because the key idea Jörg develops is that social systems must be understood in terms of self-potentiating generative complexity. That is to say their properties are emergent - the fundamental of complexity thinking as a whole. Now, whatever the limitations of simulation - and I believe that they are very considerable . simulation is a method by which emergence can be illustrated. I will come back to the more general significance of this absence.
It is hard to summarize the overall book. Jörg spends a lot of time establishing what he means by complexity and why thinking within a complexity frame of reference is so important for the future of the social sciences and humanities. Along the way he brings in a lot of relevant material, notably realism as developed by Bhaskar with a particular attention to the work of Margaret Archer and her discussion of the significance of agency in the social world. Other names which re-occur are Morin and, repeatedly, Vygotsky. I have to admit that my eyes tend to glaze over whenever I try to engage with arguments based on the ideas of Vygotsky. And that matters because I can see that engagement with Vygotsky is a way into developing a complexity based psychology which gets us beyond the reductionist direction taken by that discipline for most of its recent history. Others for whom this is a central interest will get more out of that aspect of the book than I did.
The engagement with Archer is interesting but superficial. This is because Jörg deals only with one side of the fundamental dichotomy which is central to her work, the idea of structure. Archer has a very sophisticated interesting take on the structure/agency dichotomy which certainly sees structure as constructed by agentic action but also understands that it is real in and of itself and has causal powers of its own. This is the fundamental problem for agent based modelling - it cannot generate structures which are causally powerful through time. Sawyer deals with this by denying the causal powers of structure. Jörg doesn't deal with structure at all, and that is a strange thing in a book addressed to reconfiguring the social sciences in complexity terms.
Which brings me to the most important element in the book, an argument with which I wholly agree that it is essential to establish a new framing for our thinking about causation and that a complexity frame of reference provides us with a way of doing this. Jörg raises this issue throughout the book but addresses it directly in his Chapter Eleven on "Rethinking Causality". He gets one thing here absolutely right. That is he points out that causation in the world of complexity is both self-referential and reflexive. Causes cause themselves and moreover cause effects whose reflexive causal effect is essential for the nature of the causes themselves. His discussion of effects in terms of direct effects - what is conventionally thought of as a cause; self effects - the self-maintenance of the nature of the causes themselves; and reciprocal effects - the role of the effects of causes in reflexively engendering the causes themselves, is persuasive and very useful.
However, having done this he then commits what in the view of this reviewer at least is the cardinal crime in addressing causality in relation to complex systems. He turns to an account constructed in terms of variables and endorses a development of structural equation/latent variable analysis, as the mode for explicating complex causality. This is surprising because Jörg insists throughout the book on the relational character of complex systems and that systems must be understood in terms of networks of relations. One of the major defects of traditional variable based methods derived ultimately, whatever the sophistication of their development, from the General Linear Model, is that they cannot accommodate such relational elements. Indeed Macy and Willer (2002) entitle their useful, if somewhat over-enthusiastic, discussion of "Computational Sociology and Agent Based Modelling: From Factors to Actors", precisely to indicate the turn from a quantitative sociology focused on hidden causal variables to a simulation based programme focusing on relations.
Jörg is absolutely right - understanding causation is the fundamental issue. It is therefore surprising that he takes this turn back to variables when complexity in general has moved beyond them. To be fair, his proposed way of trying to develop the variable based approaches does identify the problems associated with their linearity and failure to address reflexive causation but it seems to me that he is trying to fix the unfixable.
This is an interesting book but it seems to me to go all around the target of his project of developing a new way of doing social science without ever getting anywhere near not just the bullseye but also the inner rings. If you aren't turned off by the hippy dippy elements then there are lots of interesting ideas but in the end it is a frustrating read.
HAYLES, K (1999) How we became Post-human. Chicago: University of Chicago Press
MACY, MW and Willer, R (2002) From Factors to Actors; Computational Sociology and Agent Based Modeling. Annual Review of Sociology 28, pp. 143-66
SAWYER, RK (2005) Social Emergence. Societies as Complex Systems. Cambridge: Cambridge University Press
Return to Contents of this issue
© Copyright JASSS, 2012