Introduction
In the field of multiwinner voting, selecting a committee often involves a fundamental choice between competing philosophies. Should the committee consist of the "best" individual candidates (excellence), or should it represent the diverse preferences of the electorate (representation)? Traditional approaches require a decision-maker to select a single voting rule, such as $k$-Borda for excellence or Chamberlin–Courant ($\beta$-CC) for diversity, and accept the inherent trade-offs. This paper introduces a framework for Multigoal Committee Selection, enabling the discovery of "compromise" committees that perform well across multiple criteria.
The Multigoal Framework
The authors formalize the $\mathcal{R}$-Multigoal Committee problem: given an election and a set of scoring rules $\mathcal{R}$, find a committee that meets specific lower-bound scores for every rule in $\mathcal{R}$. This approach allows for the identification of committees on the Pareto frontier, where no single goal's performance can be improved without degrading another.
The research identifies a "boundary of tractability" for this problem. While finding optimal committees is generally complex, the problem is solvable in polynomial time using dynamic programming if the number of rules is fixed and scores are polynomially bounded. Furthermore, for rules that are NP-hard to compute (like $\beta$-CC), the authors provide polynomial-time approximation algorithms that achieve $(1 - \epsilon)$-approximate results for both excellence and diversity.
Experimental Methodology
To evaluate these trade-offs in practice, the study utilizes synthetic election data, primarily the 2D Euclidean Model. In this model, candidates and voters are represented as points in a two-dimensional plane, with voter preferences determined by proximity to candidates. This model effectively captures the spatial nature of political and social preferences. Robustness was further verified using Impartial Culture and Mallow’s Model distributions.
Key Findings: The Overstated Tension
A central insight of this work is that the perceived conflict between excellence and diversity is often less severe than theoretical models suggest.
As shown in the figure above, it is frequently possible to find committees that achieve over 90% of the optimal score for both $k$-Borda and $\beta$-CC simultaneously. In many realistic scenarios, the Pareto frontier includes committees that are 95%-approximate for both rules, suggesting that high-quality compromises are readily available.
Rule Compatibility and the SNTV Outlier
While many rules harmonize well, the Single Non-Transferable Vote (SNTV) proved to be an exception.
As illustrated in the figure above, SNTV is significantly less compatible with other rules. Because SNTV only considers a voter's top-ranked candidate, it discards the broader preference information utilized by Borda or $\beta$-CC. Consequently, optimizing for SNTV often leads to a sharper decline in performance for other metrics compared to compromises between Borda and $\beta$-CC.
Conclusion and Future Directions
This research demonstrates that multigoal selection is computationally feasible and practically rewarding. By framing committee selection as a multi-objective optimization problem, organizers can move beyond binary choices between excellence and diversity. Future work remains to extend this framework to other classes of rules, such as those based on Condorcet winners or iterative elimination, and to explore the axiomatic properties of these compromise committees.
Summary updated