Computational Social Choice in the Wild: Lessons Learned from Two Voting Experiments in Political Elections
Abstract: The problem of electing a set of representatives that stand for a given society is at the heart of any reasonable democracy. This problem has been discussed for ages, and social choice theorists have proposed dozens of voting methods, that all have their own merits and properties (but all suffer from the same paradoxes). There exist many voting rules. Yet, mainly the most simple ones like plurality or first-past-the-post are commonly used for crucial political elections (except in a few countries). Suppose that we change the voting rule for such high-stake political elections. What impact would it have on the voters’ behaviour and on the result of the election itself? This has been the main topic of two different experiments that have been carried out in France in 2017. The first one was run during the French presidential election and involved more than 40,000 voters that were asked to test alternative voting rules to elect the French president. The second one was run after the legislative election and aimed at using computer simulations to evaluate the impact of a reform of this election. In his talk, Sylvain Bouveret will use these two experiments to illustrate what happens when computational social choice theorists go out of the labs in the wild world of political elections.
ShortBio: Sylvain Bouveret is an Associate Professor in Computer Science at
Grenoble INP, University Grenoble-Alpes. He obtained an engineering degree in 2004 and a PhD in Computer Science from the University of Toulouse in 2007. From 2007 to 2011 he worked as a Research Engineer at Onera Toulouse. His current research mainly concerns various aspects of Computational Social Choice, with a special focus on fair division of indivisible goods and voting. Although his main technical background is on algorithmics and complexity, he recently extended his research activities to data visualization and experiments in voting.
The myths and science of negotiation and other conversational practices
Abstract: Why do we think we know so much about negotiation, communication, and conversation when social scientists of all disciplines so often fail to capture and study actual talk in the wild? In this talk, Professor Elizabeth Stokoe will describe her work in conversation analysis, a field rooted in sociology and one which has been producing some of the most cited – yet relatively unknown – research in academia. Drawing on a variety of workplace and organizational settings, including crisis negotiation, medical communication, police interviews and commercial sales encounters, she will show how the analysis of conversation reveals its highly organized and systematic nature, and how research findings about what is (in)effective can underpin evidence-based communication training with integrity and impact.
Short Bio: Elizabeth Stokoe is Professor of Social Interaction in the School of Social Sciences at Loughborough University. She uses conversation analysis to understand how talk works – from first dates to medical communication and from sales encounters to hostage negotiation. Outside the university, she runs workshops with doctors, mediators, salespeople, police and other professionals using her research-based communication training method called the “Conversation Analytic Role-play Method”. She is a WIRED Innovation Fellow and her research and biography were featured on the BBC Radio 4’s The Life Scientific. In addition to publishing over 100 scientific papers and books, she is passionate about science communication, translating the world of conversation analysis for audiences of all kinds. She has given TED, New Scientist, Google and Royal Institution lectures, and performed at Latitude Festival (in the Wellcome Trust/British Psychological Society stage) and Cheltenham Science Festivals. Her book, Talk: The Science of Conversation, is published by Little, Brown (2018).