Search

Incentive Engineering: Getting to the Right Inputs

William Mong Distinguished Lecture by Professor David C. Parkes
Jul 22, 2016

Professor David C. Parkes, Area Dean for Computer Science, Harvard College Professor, George F. Colony Professor, Harvard University, gave a lecture on July 22, 2016 titled “Incentive Engineering: Getting to the Right Inputs”.

It is typical in computer science and operations research to take inputs to a problem as given, and develop scalable computational methods to determine an optimal output, which might correspond to a way to allocate resources, a plan of actions, or a social choice. But there are many systems where the output affects the input. Consider meeting scheduling, where the design of the rules for selecting a time affects the reports of availability and preferences; or, Internet advertising systems, where the design of auctions affects the bids that are received; or crowd-sourcing where the design of social incentives affects participation; or “bake-offs” between competing firms where the design of the test (or experiment) induces a game that each firm wants to win. In all these settings, the way in which the output is selected from the inputs affects the inputs, and thus the performance of the system. Systems must be engineered not just to be scalable, but also to have good performance in equilibrium, considering the self-interest of participants. In this talk, Professor Parkes showed current work on incentive engineering for the design of effective, multi-agent systems and describe some of the challenges in this new field.