Ronald Fadel’s research while affiliated with Stanford University and other places

We consider how many bits need to be exchanged to implement a given decision rule when the mechanism must be ex post or Bayesian incentive compatible. For ex post incentive compatibility, the communication protocol must reveal enough information to calculate monetary transfers to the agents to motivate them to be truthful (agents' payoffs are assumed to be quasilinear in such transfers). For Bayesian incentive compatibility, the protocol may need to hide some information from the agents to prevent deviations contingent on the information. In both settings with selfish agents, the communication cost can be higher than in the case in which the agents are honest and can be relied upon to report truthfully. The increase is the "communication cost of selfishness." We provide an exponential upper bound on the increase. We show that the bound is tight in the Bayesian setting, but we do not know this in the ex post setting. We describe some cases where the communication cost of selfishness proves to be very low.

We propose a model that formalizes the beliefs of agents in strategic environments and restricts their possible behaviors, without the typical epistemic assumptions used in game theory. We formalize the beliefs of an agent using outsmarting belief systems (OBS) and then propose the notion of belief stability to explain why some OBSs, in particular some that should occur in equilibrium, are more sensitive to perturbations than others. Also, we propose the concept of belief complexity as a criteria to restrict the possible OBSs. This allows us to formalize the notion of strategic communication as belief engineering, in which agents act in order to have other agents believe some low-complexity OBS. These concepts provide a new approach to understand why some equilibrium and non-equilibrium strategies are seen in practice, with applications to the centipede game.

We consider the communication complexity of implementing a given decision rule when the protocol must also calculate payments to motivate the agents to be honest in an ex post equilibrium (agents' payoffs are assumed to be quasi-linear in such payments). We find that the communication cost of selfishness when measured with the average-case communication complexity may be arbitrarily large. For the worst-case communication complexity measure, we provide an exponential upper bound on the communication cost of selfishness. Whether this bound is ever achieved remains an open question. We examine several special cases in which the communication cost of selfishness proves to be very low. These include cases where we want to implement efficiency or where we have only two agents, and the precision of agents' utilities is fixed.