arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2205.14838cs.GTcs.AIecon.TH
收藏

最公平的投票规则

Most Equitable Voting Rules

RPI(伦斯勒理工学院)

Lirong Xia

AI总结:

针对匿名性、中立性与可决定性不可兼得的ANR难题,本文提出最公平细化概念与MFP平局打破方法,并给出不可能性条件及多项式时间算法。

摘要:

在社会选择理论中,匿名性(所有主体被平等对待)和中立性(所有备选方案被平等对待)被广泛视为公平与公正的“最低要求”和“无争议”公理。然而,ANR不可能性——不存在同时满足匿名性、中立性和可决定性(始终选出一名获胜者)的投票规则——即使在两个备选方案和两名主体的简单情形下也成立。如何设计以最优方式满足匿名性、中立性和可决定性的投票规则,仍是一个开放问题。我们针对包括排序名单和委员会在内的广泛偏好与决策类型,研究这一最优设计问题。我们的概念贡献是提出一种新颖且强的“最公平细化”概念:对于任何满足匿名性和中立性这两条公理的非决定性规则,该细化都能最优地保留这两项性质。技术贡献有两方面:第一,我们刻画了一般情形下ANR不可能性成立的条件,尤其是主体数量很大时;第二,我们提出最有利排列(MFP)平局打破方法,用于计算最公平细化,并在主体偏好为完整排序时设计了计算MFP的多项式时间算法。

英文摘要:

In social choice theory, anonymity (all agents being treated equally) and neutrality (all alternatives being treated equally) are widely regarded as ``minimal demands'' and ``uncontroversial'' axioms of equity and fairness. However, the ANR impossibility -- there is no voting rule that satisfies anonymity, neutrality, and resolvability (always choosing one winner) -- holds even in the simple setting of two alternatives and two agents. How to design voting rules that optimally satisfy anonymity, neutrality, and resolvability remains an open question. We address the optimal design question for a wide range of preferences and decisions that include ranked lists and committees. Our conceptual contribution is a novel and strong notion of most equitable refinements that optimally preserves anonymity and neutrality for any irresolute rule that satisfies the two axioms. Our technical contributions are twofold. First, we characterize the conditions for the ANR impossibility to hold under general settings, especially when the number of agents is large. Second, we propose the most-favorable-permutation (MFP) tie-breaking to compute a most equitable refinement and design a polynomial-time algorithm to compute MFP when agents' preferences are full rankings.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑