UC Irvine

In many real-world scenarios, individual agents' interests are often not fully aligned, in fact, they can even be conflicting with a principal's objectives. The principal needs to take measures to influence agents' decisions or behavior to achieve desirable system-wide outcomes. A powerful tool for motivating self-interested agents to cooperate is to offer incentives for their efforts (e.g. cooperation or sacrifices) by committing to some allocation and payment rules. This approach to implementing the incentive rules is called mechanism design. Mechanism design has many promising applications in a variety of critical domains that include spectrum allocation, online marketplace, transportation management, power grids, education, and health care. Despite its promising prospects in addressing some of the most challenging societal issues, mechanism design has not leveraged its full potential due to assumptions such as full rationality, direct preference revelation, and no group manipulation.

In this work, I introduce a unified framework called mechanism design with thresholding agents (MDTA) to relax some of those unrealistic

assumptions. The proposed approach integrates a series of new techniques that include modeling agents' decision-making with cutoff policies, indirect preference revelation, and using contests to increase competition among agents to counter group manipulations. I

demonstrate the power of the proposed framework by applying it to real-world problems that arise in crowdfunding, transportation systems, information diffusion, and utility sharing. My work extends traditional mechanism design by providing a systematic approach to influencing agents' behavior for desirable objectives.