Beyond plurality: Truth-bias in binary scoring rules

Svetlana Obraztsova, Omer Lev, Evangelos Markakis, Zinovi Rabinovich, Jeffrey S. Rosenschein

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Scopus citations

Abstract

It is well known that standard game-theoretic approaches to voting mechanisms lead to a multitude of Nash Equilibria (NE), many of which are counter-intuitive. We focus on truth-biased voters, a model recently proposed to avoid such issues. The model introduces an incentive for voters to be truthful when their vote is not pivotal. This is a powerful refinement, and recent simulations reveal that the surviving equilibria tend to have desirable properties. However, truth-bias has been studied only within the context of plurality, which is an extreme example of k-approval rules with k = 1. We undertake an equilibrium analysis of the complete range of k-approval. Our analysis begins with the veto rule, the other extreme point of k-approval, where each ballot approves all candidates but one. We identify several crucial properties of pure NE for truth-biased veto. These properties show a clear distinction from the setting of truth-biased plurality. We proceed by establishing that deciding on the existence of NE in truth biased veto is an NP-hard problem. We also characterise a tight (in a certain sense) subclass of instances for which the existence of a NE can be decided in poly-time. Finally, we study analogous questions for general k-approval rules.

Original languageEnglish
Title of host publicationAlgorithmic Decision Theory - 4th International Conference, ADT 2015, Proceedings
EditorsToby Walsh
PublisherSpringer Verlag
Pages451-468
Number of pages18
ISBN (Print)9783319231136
DOIs
StatePublished - 1 Jan 2015
Externally publishedYes
Event4th International Conference on Algorithmic Decision Theory, ADT 2015 - Lexington, United States
Duration: 27 Sep 201530 Sep 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9346
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference4th International Conference on Algorithmic Decision Theory, ADT 2015
Country/TerritoryUnited States
CityLexington
Period27/09/1530/09/15

Fingerprint

Dive into the research topics of 'Beyond plurality: Truth-bias in binary scoring rules'. Together they form a unique fingerprint.

Cite this