Justifications for Generalizations of Approval Voting

TL;DR

This paper explores a generalized framework for approval voting by considering votes as vectors under arbitrary p-norms, and examines how existing justification methods extend to these new voting systems.

Contribution

It introduces a general p-norm framework for voting and analyzes how traditional justification methods apply to these generalized voting rules.

Findings

Approval voting, satisfaction approval voting, and quadratic voting are special cases of p-norm voting.

The paper investigates the applicability of existing justification methods to the generalized p-norm voting systems.

Abstract

Approval voting is a common method of preference aggregation where voters vote by ``approving'' of a subset of candidates and the winner(s) are those who are approved of by the largest number of voters. In approval voting, the degree to which a vote impacts a candidate's score depends only on if that voter approved of the candidate or not, i.e., it is independent of which, or how many, other candidates they approved of. Recently, there has been interest in satisfaction approval voting and quadratic voting both of which include a trade-off between approving of more candidates and how much support each selected candidate gets. Approval voting, satisfaction approval voting, and quadratic voting, can all be viewed as voting where a vote is viewed as analogous to a vector with a different unit norm ($\mathcal{L}^{\infty}$, $\mathcal{L}^{1}$, and $\mathcal{L}^2$ respectively). This suggests a generalization where one can view a vote as analogous to a normalized unit vector under an arbitrary $\mathcal{L}^p$-norm. In this paper, we look at various general methods for justifying voting methods and investigate the degree to which these serve as justifications for these generalizations of approval voting.