A Virtual World Experiment on Emergent Behavior in Collective Action Problems

Abstract:     
Social scientists are interested in how organizations manage group decisions to more efficiently solve collective action problems. In studying this question scientists have relied on a combination of empirical field work, game theoretic modeling, and experiments. What has emerged is a successful common practice for institutional design which can be characterized as follows: One, use field studies, to characterize the incentives and informational problems that exist, and to hypothesize how institutional rules work to efficiently organize collective action. Two, use non-cooperative game theory to formalize these hypotheses, and then study the comparative equilibrium properties of different institutional rules in theory and in the laboratory. Three, run randomized experiments in the field to study how policy changes can improve collective action outcomes. In this paper we investigate the potential for virtual worlds in taking intermediate steps between one and two, and between two and three. A key element of empirical field research (in one and three) is natural context and natural language. These two elements help to situate and discuss the problem and yet they are usually absent in the translation to theory and in laboratory experiments.

In our experiments in Second Life™ we study a collective action problem of protecting capital from natural disasters. These experiments are motivated in large part by Elinor Ostrom’s analysis of self organized and self-governed solutions to common pool resource problems. To do this the participants must (1) supply the rules of the game, (2) commit to following the rules, (3) and monitor that the rules are indeed followed. Ostrom demonstrates that in the field enduring solutions to common pool resource problems satisfy design principles which can be implemented in our experiments. Besides the obvious questions of how successful are subjects at solving their collective action problem, we are also interested in the properties of these emergent solutions. We hypothesize that game theory will be useful in elucidating the issues that will emerge in finding congruent solutions between the local conditions (modifiable by experimental design) and the appropriate governance rules. We also hypothesize that Construal Level Theory, Liberman and Trope, can be used to analyze the psychological distance from objects and events and subjects mental construal. Psychological distance can be measured in the natural language message space of subjects and will have an impact on the solution to the collective action problem.

working papers series