Index and Stability in Bimatrix Games Book

A set of clear and compelling demonstrations of the milage you can get out of applying evolutionary dynamics to classic games. Your favorite demonstration will probably vary with your prior interests (e.g. some may appreciate the applications to theories of justice, while others may gravitate toward the discussion of altruism). For me, it was the final chapter on the evolution of meaning. This book is often cited in discussions of social conventions, and it probably makes the biggest theoretical jump since David Lewis set up the foundations in 1969. Lewis formulates conventions as arbitrary, but self-sustaining, solutions to coordination problems and showed exactly how they're self-sustaining once established, but more-or-less left open the question of how they're established in the first place. The usual explanation appeals to cognitive constructs like salience (i.e. focal points in Shelling games), but this just pushes back the question to how common knowledge of salience is established. Skyrms shows that random populations of non-cognitive agents will necessarily converge on conventions simply due to replicator dynamics (with a minuscule value of a correlation parameter controlling how often "like meets like"), and that these conventions are evolutionarily stable. This obviously isn't to say that conventions among humans are formed without social reasoning, especially for the substantially more complex behaviors we display. But if we're comfortable modeling sequences of social interactions as "generations" in replicator dynamics, this is an encouraging proof of concept for how we might hope to model & explain the formation of conventions. In the 2 decades since Evolution of the Social Contract was published, we've been able to elaborate on it.