This project investigates efficient protocols for complex contracts with more challenging utility functions, including higher order dependencies, and for multilateral negotiations. Constraint-satisfaction techniques, as well as nonlinear optimization techniques such as simulated annealing, Tabu search, and evolutionary algorithms, are the basis for negotiation mechanisms for socially beneficial truth-telling behavior and are pareto-optimal. The project uses computer simulations as well as real-world test cases to develop novel analytic techniques suited to nonlinear utility functions.
Techniques to enable fast and effective complex contract negotiations have the potential for impact on social welfare in domains that range from electronic commerce to collaborative design to the creation of laws and international treaties. While the focus here is on at least partially automating the negotiation process, the development of optimal incentive compatible protocols should also often be applicable in contexts with human agents. The proposed work will provide opportunities for students at MIT, which include many minorities, to participate as research assistants in a challenging multi-disciplinary research project.