Abstract
This paper introduces formal proportionality criteria for collective decision-making, ensuring that any group of voters—not just those who agree on everything—receives a share of influence proportional to its size. The authors prove that adapted versions of committee election rules, specifically Proportional Approval Voting (PAV) and the Method of Equal Shares (MES), provide strong fairness guarantees even when voters abstain from certain issues. The introduction of the Method of Coordinated Auctions (MeCorA) further extends these guarantees to complex scenarios involving varying voter interest.
The Problem: The Winner-Takes-All Flaw
In typical democratic settings, groups often face a series of independent binary (YES/NO) decisions—such as a housing cooperative deciding on various renovations or legislators negotiating a coalition agreement. Usually, these are decided by simple majority voting. However, majority voting suffers from a significant "winner-takes-all" flaw: if 51% of voters agree on every issue, the 49% minority is completely ignored across all decisions.
Ideally, a group representing 49% of the population should influence roughly 49% of the outcomes. While "proportionality" is a well-studied concept in committee elections (choosing $k$ candidates), this paper tackles the challenge of applying these fairness principles to public decisions where the number of "winners" (YES outcomes) is not fixed and voters might only care about a subset of the issues.
Proposed Methods: PAV, MES, and MeCorA
The authors adapt two prominent rules from the social choice literature and introduce a novel variant to handle abstentions:
- Proportional Approval Voting (PAV): An optimization-based rule that picks outcomes to maximize a harmonic "social welfare" function ($1 + 1/2 + 1/3...$). This system rewards satisfying a voter multiple times with diminishing returns, which naturally pushes the collective toward diversity and proportionality.
- Method of Equal Shares (MES): A budget-based rule where each voter starts with a fixed amount of "currency." Decisions are "bought" sequentially; if a group wants a specific outcome, they pool their money to pay for it. Once a voter's budget is exhausted, they can no longer influence further decisions.
- MeCorA (Method of Coordinated Auctions with Equal Shares): A novel variant designed to handle cases where voters abstain. In MeCorA, voters bid on decisions, and prices can be raised if a different group wants to "outbid" and flip a decision, ensuring that the outcome remains proportional even with varying levels of voter interest.
Theoretical Framework and Proportionality Axioms
The paper’s core contribution is providing guarantees for any group of sufficient size, moving beyond previous work that only focused on "cohesive" groups (voters who agree on everything). This is crucial because real-world voters rarely have perfectly aligned preferences.
- The "Price" of Fairness: The authors show that to achieve proportionality, we must relax the "ideal" requirement by a small factor. They define proportionality such that a group of size $\alpha$ should influence at least a fraction of decisions roughly equal to $\alpha/2$.
- Optimality of PAV: The paper proves that PAV provides the strongest possible proportionality guarantees in the standard model without abstentions.
- Addressing Abstentions: The authors discover that the standard MES fails when voters can abstain. Their proposed MeCorA fix uses a coordinated auction mechanism to restore proportionality in these more complex scenarios, as illustrated in the comparative outcomes of Figure 1.
Key Results and Performance
- Theorems 1 & 5: PAV satisfies the "Proportional Average Representation" axiom both with and without abstentions.
- Theorem 3: For the Method of Equal Shares (MES), the authors prove a lower bound on average utility: $d_{MES}(\alpha) > \frac{\alpha \cdot (m+1)}{3} - 1$.
- Theorems 6 & 7: The newly introduced MeCorA rule is proven to be $\epsilon$-proportional and satisfies proportionality for cohesive groups, making it a robust choice for settings with varying voter interest.
Conclusion and Future Work
While providing a strong theoretical foundation for proportional public decisions, the authors acknowledge several open challenges. Future research is needed to determine if the mathematical bounds for MES and PAV can be further tightened. Additionally, the current model assumes all decisions are independent; in reality, a "YES" on one issue might change a voter's preference on another. Finally, scaling these rules—particularly the auction dynamics of MeCorA—to massive numbers of issues and voters requires further algorithmic optimization.
Summary updated