Explanation Systems for Approval-Based Multiwinner Voting

Abstract: In approval-based multiwinner voting, voters express approval preferences over a set of candidates, and the goal is to return a winning committee. This model captures a broad range of subset selection problems under preferences. Prior work has focused on the study of binary proportionality axioms that certify whether a given committee is proportionally representative or not. We take a more fine-grained perspective and initiate the study of explanation systems that quantify how a committee represents the electorate, i.e., how much influence each voter exerts, how this influence is allocated across selected candidates, how each candidate is backed by the voters, and why certain candidates were not chosen. Building on the notion of priceability, we propose price systems as a framework for such explanations. A price system assigns each voter an individual budget, which they can spend on selected candidates they approve, and each candidate needs to be purchased at a unit price. Since many price systems can exist for a given outcome, selecting among them requires care. We initiate an axiomatic study of price systems and propose several axioms capturing structural coherence, faithful attribution of influence, and alignment with proportionality. On the algorithmic side, we introduce a polynomial-time computable rule in which voters continuously gain and exercise influence and show that it satisfies all jointly satisfiable axioms. Experiments on synthetic and real-world instances indicate that our explanations correlate with established proportionality notions and can recover unequal influence when it is present.

• The paper introduces price systems, a quantitative framework for explaining voter influence and auditing representation in approval-based multiwinner voting.
• It details three explanation rules, including the novel Continuous Phragmén, which accurately reflects voter budgets and aligns with strong proportionality axioms.
• Empirical results demonstrate that Continuous Phragmén effectively recovers latent voting inequalities and provides actionable insights for detecting underrepresented coalitions.

Explanation Systems for Approval-Based Multiwinner Voting: An Expert Analysis

Introduction and Motivation

Approval-based multiwinner voting (ABMV) is a foundational framework in computational social choice for collective decision-making, capturing various real-world subset selection problems such as participatory budgeting, product review curation, and group deliberation. Traditionally, ABMV research has concentrated on binary proportionality axioms that certify whether a given committee is representative. This paper (2604.24307) advances the field by shifting from coarse, binary assessments towards fine-grained, quantitative explanations of how selected committees represent and are influenced by the electorate. The study is predicated on the need for granular auditing tools that elucidate the distribution of voter influence, support attributions, and justifications for inclusion or exclusion of candidates.

Price Systems as a Framework for Explanations

The authors introduce price systems as the backbone of their explanation framework. In a price system, each voter is assigned a (potentially non-uniform) budget, which must be spent—subject to approval constraints—on purchasing selected candidates, each at unit cost. Any leftover budget (residual) remains with the voter and is critical for justifying why certain unselected candidates were not included. The central technical notion is residual stability: for any unselected candidate, the total residual budget of its supporters must be insufficient to buy that candidate outright.

This construct builds upon but significantly generalizes prior notions such as priceability (Peters & Skowron 2020), which is typically binary and only certifies the existence of a budget-uniform residual-stable price system. The current work, however, leverages the full richness of admissible price systems to quantify inequities in influence and to explain representation at the level of both voters and candidates.

Axiomatic Foundations

Given the non-uniqueness of price systems for a committee, the authors embark on an axiomatic analysis to distinguish desirable explanation rules. Their axioms are organized in three classes:

• Structural Coherence: Requires explanations to respect automorphisms and voter/candidate symmetries, enforcing "equal treatment of equals" and invariance under profile automorphisms.
• Faithfulness of Influence Attribution: Develops technical strengthenings of residual stability (most notably, 1-stability) ensuring that budgets, payments, and residuals truly reflect criticality and the constraints imposed by the profile for representation. Additional axioms enforce responsive payments in degenerate (e.g., single-winner) settings and regulate how unused budget is averaged among voters.
• Alignment with Proportionality: Enforces that, whenever committees satisfy strong forms of proportionality (e.g., in laminar or perfectly symmetric instances), all voters' budgets should be equal. Conversely, explainers should output non-uniform budgets when the committee is patently unproportional.

The authors comprehensively catalog the logical relationships between these axioms and present both impossibility and possibility results. Notably, they prove that certain combinations (e.g., monotonicity with 1-stability and budget-averaging) are provably incompatible.

Algorithmic Contributions

Three principal explanation rules are developed and compared:

1. Approximate Priceability: This baseline computes a residual-stable price system minimizing absolute pairwise budget deviations via linear programming. Although it ensures budget uniformity in highly proportional settings, it can assign equal budgets and obscure influence inequalities even in unproportional profiles, thereby failing to satisfy many structural and influence attribution axioms.
2. Equal Split: This rule divides each candidate's cost equally among its supporters, then allocates residuals to maximize equality without violating 1-stability. Equal Split satisfies strong symmetry and proportionality axioms but does not differentiate critical from non-critical voters in payments.
3. Continuous Phragmén: The main technical advance is an efficient polynomial-time explainer inspired by sequential Phragmén's method but generalized to work over arbitrary committees. Here, voters continuously accrue and allocate budget to current candidates they can afford, while ensuring no violation of residual or 1-stability. The method is carefully designed to avoid arbitrary tie-breaking and respects all structural symmetries by construction. Crucially, authors demonstrate that all jointly satisfiable axioms are met by Continuous Phragmén, except for monotonicity (proven to be incompatible).

Strong Empirical Findings

The empirical evaluation spans synthetic (Euclidean-VCR, Resampling) and real-world participatory budgeting datasets (Pabulib), across varying committee sizes. The computed price systems—especially via Continuous Phragmén—consistently align granularity of voter budgets with established proportionality benchmarks such as EJR+ (Brill & Peters 2023).

Key numerical findings include:

• In every instance where the committee fails EJR+, Continuous Phragmén and Equal Split assign distinctly non-uniform (i.e., low) minimum voter budgets, precisely identifying underrepresented coalitions.
• The price systems can recover latent unequal influence in weighted voting scenarios: budgets assigned to voters in explanations are highly correlated (often $r > 0.7$) with underlying ground-truth weights.
• Approximate Priceability fails to disclose influence inequalities, particularly in structured or pathological instances.

Theoretical and Practical Implications

This paper establishes a formal and computable framework for post-hoc auditing of committee representation in ABMV. The key practical implication is the ability to not only check proportionality ex post, but also to diagnose which voters, subgroups, or regions are over- or under-represented, and to trace the flow of influence to individual selected candidates.

The theoretical implications are substantial: the axiomatic framework provides a nuanced generalization of proportional representation to arbitrary settings, encompassing and refining prior concepts such as priceability, PJR+, and core stability. Moreover, the work lays the groundwork for extending explanation systems to more complex settings, including ranked ballots, projects with heterogeneous costs, and broader subset selection problems such as text summarization and hiring.

Contradictory Findings and Limitations

A bold technical claim established is that budget uniformity in explanations directly witnesses strong proportionality (i.e., EJR+), while minimum assigned budgets can precisely quantify deficits in proportionality. Furthermore, the impossibility results reveal fundamental tensions between axioms that appear desirable in isolation.

One limitation is that well-designed axiom systems for payments (beyond what is pursued here) are proven tricky; strengthening payment requirements often induces perverse or counterintuitive behavior elsewhere. This suggests open complexity in judgment aggregation and influence allocation mechanisms.

Future Directions

The research opens several promising directions:

• Extending axiomatic frameworks to ranked or cardinal preferences.
• Incorporating non-uniform candidate prices to diagnose and explain more subtle forms of misrepresentation.
• Applying price system explanations as auditing tools in real-world applications such as participatory budgeting, content curation, and algorithmic hiring to detect unfairness in the absence of sensitive attribute labeling.
• Exploring optimization-based explainers and their theoretical properties.

Conclusion

This paper systematically develops an axiomatic, fine-grained, and computationally efficient framework for explanation and auditing in approval-based multiwinner voting. By leveraging the flexibility of price systems and centering residual and 1-stability as touchstones for justified representation, the work sets a new standard for transparency and fairness analysis in collective subset selection. Continuous Phragmén stands out as an explainer satisfying all jointly realizable structural, influence-attribution, and proportionality-alignment axioms, demonstrating both theoretical rigor and practical efficacy in discerning latent influence and justifying selection outcomes.