Self-Organization in Communication Networks

We develop a dynamic model to study the formation of communication networks. In this model, individuals periodically make decisions concerning the continuation of existing information links and the formation of new information links, with their cohorts. These decisions trade off the costs of forming and maintaining links against the potential rewards from doing so. We analyze the long run behavior of this process of link formation and dissolution. Our results establish that this process always self-organizes, i.e., irrespective of the number of agents, and the initial network, the dynamic process converges to a limit social communication network with probability one. Furthermore, we prove that the limiting network is invariably either a wheel network or the empty network. We show in the (corresponding) static network formation game that, while a variety of architectures can be sustained in equilibrium, the wheel is the unique efficient architecture for the interesting class of parameters. Thus, our results imply that the dynamics have strong equilibrium selection properties.networks;learning;coordination;self-organization;path-dependence