Emergent collective behavior in humans playing online games

Collective behavior is a key component of humanity's success story. Through interaction and collaboration, groups of humans have built edifices and bridges, discovered vaccines, developed complex languages, and accumulated over generations a collective knowledge stored in the structure of society. While human collective behavior has undergone scrutiny across various contexts in the past decades, a comprehensive theory elucidating the fundamental principles leading to the emergence of this behavior has yet to be developed. We present work in the service of that goal that examines collaboration in groups of humans as they play a well-controlled and tunable online game. We examine collaboration through the lens of statistical physics (in particular network theory and information theory), and find that collaboration among humans reliably emerges and can be enhanced or suppressed by manipulating game environment geometry and player capability mismatch. Our results contribute to a broader understanding of emergent collective intelligence during active team collaboration, and point toward causal mechanisms for that emergence in a model environment.