Physics Colloquium: "Contributions of statistical physics to data analysis in the social sciences"
Abstract: What are the micro-level behavioral mechanisms of macro-level social outcomes? With data on complex social systems coming in at unprecedented volume and granularity, social scientists have a chance to dramatically increase the accuracy, simplicity, and testability of their theories. I present two projects that use data analysis techniques from physics to illuminate the intersections of the social and cognitive sciences. First, with a large corpus of online poker play, and developments in information theory, I will show how poker players use synergistic information processing to encrypt their strategic reasoning. Then, with high-granularity tracking data from hundreds of NBA games, and methods from statistical mechanics, I will estimate the upper bound that human cognitive limits place on the ability of professional basketball players to act as a team. Engineered social environments like games and sports are very simple, highly structured, and incredibly data-rich. These qualities make them ideal for data-driven approaches to micromotives and macrobehavior.