Computational Approaches to Studying the Co-Evolution of Networks and Behavior in Social Dilemmas by Rense Corten

Corten’s book addresses an interesting topic: how can we model how social network and actors’ traits co-evolve? This research question is relevant from a theoretical and an empirical point of view: theoretical models offered to explain stylized facts are rarely empirically tested; empirical research, on the other side, often lacks theoretical foundations. Following the stream of literature on computational sociology (Fararo and Hummon 1995), Corten argues that computational approaches could represent the appropriate theory-building tool needed to fill the gap between theoretical models and empirical research. The book structure includes a first chapter that introduces social dilemmas and social networks and outlines the theoretical approach that will be used through the rest of the book. Namely, Corten follows and extends Granovetter’s general research program: actors are goal-directed but nevertheless embedded in social context. The interesting extension, in this case, is that social context is not fixed and exogenously imposed to actors but co-evolves as a consequence of (and as a constraint to) actors’ actions. Chapter 2 presents a theoretical model for coordination problems in dynamic networks. It is somewhat surprising that Chapter 2 does not review opinions dynamic models, even if the core of the chapter is to explore the role of social networks in mediating the diffusion of behaviours and opinions through society. This said, the author’s perspective on the problem at stake is a clear motivation: coordination games are the conceptual tool used to tackle diffusion, thus a rational choice settings (links have benefits and costs) and a game-theory machinery is used to predict equilibrium states in a population that has to decide between a binary behaviour (or opinion, attitudes etc.). Subsequently, the next step is to derive implications about the dynamics of polarization and network structural outcomes. Hummon (2000) showed how rational choice mechanisms cannot always generate analytically predicted outcomes when particular combinations of parameters are chosen. Since Corten’s scheduling in this case looks very similar to Hummon’s (at each tick actors are asked to add or drop a link based on an utility maximization function), it would have been interesting to see a comparison with Hummon’s findings. Chapter 3 presents a theoretical model for cooperation problems in dynamic networks, where bounded rational agents plays a Prisoner’s Dilemma while they have to build their own network. One of the most interesting findings is that the effect of the spread of reputation tends to be stronger if the network is less dense. Unfortunately, no empirical testing for the cooperation model is provided. In my opinion, this jeopardizes the theory-model-test cycle suggested in the book.