Non-Obviously Manipulable Mechanism Design

Non-obvious manipulability (NOM) is a new notion of incentive compatibility motivated by imperfect rationality that allows mechanisms to tolerate certain manipulations as long as they are not "obvious" to cognitively limited agents. Contrasted to strategy proofness, which stipulates that truthtelling is a dominant strategy, NOM instead only requires an agent to compare the extremes, taken over all possible bids of the other players, of their utility function. First and foremost we want to study the conditions under which we can design mechanisms that implement certain social choice functions by enhancing them with payments, and when this is possible we want to understand what can be gained -- with regards to notions such as fairness, efficiency, and computational tractability -- as well as what is impossible. In addition we look to apply our framework to a number of different and well-studied settings in the literature, including auctions, two-sided markets, and fair division.