Characterization of All Strategyproof Mechanisms for Expected-Utility Agents

Characterize all strategyproof mechanisms for expected-utility agents, including in the classical setting of lotteries over finitely many alternatives, beyond the implicit menu representation.

Background

The paper reformulates alignment as social choice over a convex impact space and shows that strategyproof mechanisms admit an option-set or menu representation: each agent receives a menu determined by the other agents’ reports and the mechanism selects the agent’s most-preferred feasible impact from that menu. This representation is exact but does not enumerate which menus, and therefore which mechanisms, are possible.

The authors identify voting-by-issues rules and random dictatorships as structured subfamilies of the general menu-based class. A complete characterization would extend the paper’s partial classification of strategyproof alignment mechanisms and would clarify the full range of mechanisms compatible with linear expected-utility preferences.