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.
CITATION STYLE
Obraztsova, S., Lev, O., Markakis, E., Rabinovich, Z., & Rosenschein, J. S. (2015). Beyond plurality: Truth-bias in binary scoring rules. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9346, pp. 451–468). Springer Verlag. https://doi.org/10.1007/978-3-319-23114-3_27
Mendeley helps you to discover research relevant for your work.