May 11
CS Tea: "A Computational Look at Ranked-Choice Voting"
Thu, May 11, 2023
• 3:30pm
- 4:30pm (1h) • Anderson 329
Visiting CS instructor Kiran Tomlinson ('19) will speak about his research:
"A Computational Look at Ranked-Choice Voting"
Most elections in the United States work like this: each voter picks their favorite candidate and whoever gets the most votes wins. While this system (plurality voting) is simple, it suffers from a variety of issues—for instance, spoiler candidates can prevent popular options from winning and voters are incentivized to be strategic in who they pick. To address these shortcomings, many cities (among them, Minneapolis and St. Paul)—and even some states—have been turning to another approach: instant-runoff voting (IRV, a.k.a. ranked-choice voting). In IRV, voters rank candidates in order of preference and the winner is then decided by repeatedly eliminating the candidate with the fewest first-place votes. The adoption of IRV has become a contentious political issue, with many arguments for and against. Its supporters claim it reduces polarization, promotes civility, increases voter choice, and produces more fair outcomes; meanwhile, its opponents say it’s too complex, too expensive, hurts third parties, and leads to wasted votes.
However, many of these claims are made based on little more than a vague intuitive understanding of the voting system. My research aims to provide a solid theoretical backing for understanding IRV. In this talk, I’ll present two projects analyzing IRV from a computational angle. First, I’ll discuss the importance of ballot length (how many candidates voters can rank) on determining election outcomes. Then, I’ll explore a way of formalizing the notion that IRV can help address polarization by electing more moderate candidates than plurality.
This talk is based on joint work with Jon Kleinberg and Johan Ugander (https://arxiv.org/abs/2207.08958, https://arxiv.org/abs/2303.09734).
Event Contact: pstowe@carleton.edu