Designing auctions for deliberative agents

In many settings, bidding agents for auctions do not know their preferences a priori. Instead, they must actively determine them through deliberation (e.g., information processing or information gathering). Agents are faced not only with the problem of deciding how to reveal their preferences to the mechanism but also how to deliberate in order to determine their preferences. For such settings, we have introduced the deliberation equilibrium as the game-theoretic solution concept where the agents' deliberation actions are modeled as part of their strategies. In this paper, we lay out mechanism design principles for such deliberative agents. We also derive the first impossibility results for such settings – specifically for private-value auctions where the agents' utility functions are quasilinear, but the agents can only determine their valuations through deliberation. We propose a set of intuitive properties which are desirable in mechanisms used among deliberative agents. First, mechanisms should be non-deliberative: the mechanism should not be solving the deliberation problems for the agents. Secondly, mechanisms should be deliberation-proof: agents should not deliberate on others' valuations in equilibrium. Third, the mechanism should be non-deceiving: agents do not strategically misrepresent. Finally, the mechanism should be sensitive: the agents' actions should affect the outcome. We show that no direct-revelation mechanism satisfies these four properties. Moving beyond direct-revelation mechanisms, we show that no value-based mechanism (that is, mechanism where the agents are only asked to report valuations – either partially or fully determined ones) satisfies these four properties.