Wed May 1, 2019
Newton Room, 730 Sutardja Dai Hall, 4–5 PM
|Center for Human-Compatible AI Interdisciplinary Insights Seminar
Wes Holliday (UC Berkeley)
New Barriers to Strategic Voting
Theorems in social choice theory show that any reasonable voting method is susceptible to strategic voting. One potential barrier against strategic voting, investigated mostly in the AI literature, is the computational complexity of determining a profitable strategic vote for a given voting method (see “AI’s War on Manipulation: Are We Winning?” by P. Faliszewski and A. Procaccia, AI Magazine, 2010). In this talk, based on joint work with Eric Pacuit (University of Maryland), I will discuss another potential barrier against strategic voting: adding uncertainty about the voting method that will be used to aggregate votes.