Reviewed by
Klaus G. Troitzsch
University of Koblenz-Landau
In this book, published in Oxford’s catalogue under the headline of
"Theoretical and Statistical Physics", Sen and Chakrabarti, physicists
at the Department of Physics, University of Calcutta, and the Saha
Institute of Nuclear Physics, Calcutta, "examine the studies and
analyses of the physical aspects of social systems". This is how they
announce their undertaking in the first sentence of the book preface.
Mainstream social scientists might wonder why physicists are interested
in a "physical analysis of social systems" which are clearly many-body dynamical systems". In reality, this interest goes back to the early
1970s (and thus is not only something that is "little more than a decade
old", as suggested by the authors), when Wolfgang Weidlich first
published some preliminary work in this field (Weidlich 1972) that eventually gave rise to Weidlich and Haag’s famous book on
quantitative sociology, which must be considered the first book that
applies the methods of theoretical and statistical physics"
synergetics" to the analysis of structures or patterns in society (Weidlich and Haag 1983).Unfortunately, Sen and Chakrabarti did not consider these early approaches and started only by quoting Dirk Helbing’s work on pedestrian movement (Helbing 1992), which cited an earlier paper (Henderson 1971). It seems that, in the dawn of sociophysics or synergetics, their protagonists were well aware that "the fundamental laws governing physical systems are completely different from those in social systems" and only tentatively mentioned "analogies between physical model systems and social structures" (Weidlich 1971, 51).
This book insinuates that sociophysics can deal with all social phenomena of major interest. The authors mention agent-based models as "another approach" (p. 23) particularly for modelling "systems that are non-linear or non-Markovian (i.e., systems with memory)" and their advantage of being able to incorporate sources of randomness more precisely" which is connected to "the disadvantage of such models [...] that limiting cases such as t ➝ ∞ or system size N ➝ ∞ cannot be considered". One could add that the basic units of social systems have perhaps more memory than anything else in the world and make heavy use of it, such that social systems "are difficult to handle analytically", and that the cases t ➝ ∞ or N ➝ ∞ need only seldom to be considered, as parameters of social systems rarely remain unchanged for t ➝ ∞ and the most interesting social systems (e.g., from peer groups to parliaments) are manageable in size.
Notwithstanding this criticism and the restrictedness of this book from a social science point of view, I must say that this is a great introduction to the methods that social scientists should use for modelling target systems which can be understood as Markovian and which have a large number of elements and where immergence (and hence second-order emergence) does not occur. This is the case of human social systems which are rather similar to animal social systems. And nobody would deny that sociophysics models have had and still have their merits, e.g., for understanding and avoiding traffic jams or collisions in pedestrian crowds and the resulting panic dynamics (section 7.3 of the book). Nevertheless, the movement of people in crowds can be quite different from the movements of particles in a gas or of birds or fish in a swarm or sheep in a flock, which can be looked at in terms of "social forces" that are similar to the concept of "forces" in mechanics as Sen and Chakrabarti show mainly in their Chapter 7. Imagine three persons walking in a crowded (but not overcrowded) shopping mall; A wants to show B a window to the left, C wants to show B a window to the right, and it is not the parallelogram of forces that describes B’s movement, but the three will first discuss which of the two windows to visit first, a behaviour which is typical for humans and which makes heavy use of memories of the persons involved.
And one can also wonder whether Thomas Schelling would be happy if he were subsumed under the heading of sociophysics. Not all micro-macro links are owed to sociophysics. Indeed, James Coleman and Herbert Simon could have been mentioned here as they proposed something similar, even before Thomas Schelling, trying to pave the way for a mathematical sociology (Coleman 1964 and 1973) long before physicists found the field of the social sciences an attractive area of application for their elaborated methods.
I doubt whether sociophysics likens to the "physique sociale" of Auguste Comte and to the "physique sociale" of Adolphe Quetelet, which are two entirely different concepts, as argued by Turner et al (2012, 41). The modern origins of sociophysics (e.g., Hermann Haken, Wolfgang Weidlich, Günter Haag and Dirk Helbing, to mention the German contributors to what they first named synergetics) come from a physics which was entirely unknown at the time of Auguste Comte and Adolphe Quetelet. This implies that it is questionable whether Comte and Quetelet can count as early witnesses of sociophysics.
In spite of this criticism, my impression is that this book is relevant. Perhaps, the hype of sociophysics is over (and no longer "on the rise" as Dirk Helbing writes on the back flap of the book), and it is now time for a presentation of its merits but also of its shortcomings. I will discuss some parts of the proposal following the book structure, hoping that also the book authors can find something of interest.
The main merit of this book is (in my view) the discussion of networks in Chapters 2 and 6. Social scientists and particularly those who build agent-based models can learn here how to initialise agent networks in a fairly realistic way, as the small worlds serve nicely as models for real networks among real humans.
When sociophysicists deal with opinion formation they often forget that opinions in humans are only rarely binary and even far from being one-dimensional. Furthermore, human opinions are not formed via fields and forces, but via communication, persuasion, message passing, observation and other much more complicated mechanisms. Of course, I admit that in very large crowds of isolated individuals, sociophysics can be a good first approximation. But, in real social contexts, opinion formation processes occur in small groups (e.g. cabinets of ministers or parliaments) such that sociophysics is not applicable (as its analytic solutions often depend on approximation which are only exact enough for N ➝ ∞). Thus, the book should have considered these limitations and confronted the sociophysics approach with this challenge.
Dynamics of popularity (Chapter 4) are also more complicated than what can be understood by sociophysics. For instance, let us consider the mutual influence between the stars or politicians, the fan community or the electorate and the mass media. These do not always simply follow word-of-mouth influence mechanisms. Furthermore, media can act strategically to promote certain stars or politicians who, in a sociophysicist model, would never have a chance to become popular. Sociologists and political scientists modelled similar scenarios in the 1960s, much earlier than the interest of physicists towards understanding social phenomena took place.
All these critical remarks do not devaluate the merits of this book. But they were written to make readers aware that social scientists know their fields and have developed useful methods before physicists in the late 1960s and early 1970s suggested their methods as a superior endeavour. It must be said that one of the most prestigious disciplines in the realm of social sciences, i.e., economics, which mostly followed a methodological apparatus similar to physics, at least in its faith on simplification and generalization, is correctly contested for its weak predictive power.
Once having said this, I must admit that I personally learnt a lot from sociophysicists, but I also learnt from their shortcomings, too. Had this book considered a little more seriously the challenge of the social and economic sciences, here intended as the real "hard sciences", to mention Herbert Simon’s claim (Simon 1987), and discussed restrictions and approximations that sociophysics is forced to make, an even more relevant contribution to the unification of science and the improvement of the methodological apparatus of economics and the social sciences could have been achieved by the book authors. So far, it seems that the dialogue between sociophysics and computational sociology is still missing (e.g., Squazzoni 2012) although it would help to fecundate them.
ReferencesCOLEMAN, J.S. (1973) The Mathematics of Collective Action. London: Heinemann
HELBING, D. (1992) A fluid-dynamic model for the movement of pedestrians. Complex Systems 6, 391-415
HENDERSON, L. F. (1971) The statistics of crowd fluids. Nature 229, 381-383
SIMON, H. (1987) Giving the social science a hard sell. The Boston Sunday Globe, May 3 1987, accessible at: http://digitalcollections.library.cmu.edu/awweb/awarchive?type=file&item=34441
SQUAZZONI, F. (2012) Agent-Based Computational Sociology. Chichester: Wiley and Sons
TURNER, J.H., Beeghley, L. and Powers, C. A. (2012) The Emergence of Sociological Theory. Seventh Edition. Los Angeles: Sage
WEIDLICH, W. (1972) The use of statistical models in sociology. Collective Phenomena, 1, pp. 51-59
WEIDLICH, W. and Haag, G. (1983) Concepts and Models of a Quantitative Sociology. The Dynamics of Interacting Populations. Springer: Berlin (Springer Series in Synergetics 14)
Return to Contents
of this issue