Dynamics of Social Interactions in a Network Game

We characterize the social dynamics of human players in a transactional online game. We introduce two new approaches for understanding temporal behavior: the introduction of entropy dynamics, where we measuring the change in the entropy of certain descriptive distributions over the course of a game, and the use of clustering to discover temporal dynamics in subpopulations of experimental subjects. Experiments were conducted in the context of the Social Ultimatum Game, a multi-agent multi-round extension of a classical game-theoretic domain, using two distinct populations. We show that we are able to extract temporal behavior from the use of entropy dynamics and identify unique subpopulation behavior with respect to generosity, reciprocity and fairness.