Empirical Evaluation of Collective Rationality for Quota Rules in Judgment Aggregation
Sharon Gieske, Elise van der Pol, Ulle Endriss

Abstract:
A major challenge in multiagent systems research is to design good aggregation rules for combining the judgments---or beliefs, or opinions---of different autonomous software agents collaborating with each other. If the chosen rule is too sophisticated, we may encounter algorithmic difficulties. But if it is too simple, we may encounter some of the paradoxes of social choice theory and end up with inconsistent information at the system level. We report on an empirical study aimed at improving our understanding of how frequently these paradoxes strike in practice for a class of simple aggregation rules, the uniform quota rules. Our results indicate that quota rules can be expected to work significantly better in practice than the still relatively scarce theoretical results may suggest.