Minimizing Metric Distortion in Social Choice

This direction is merged into another direction: ordinal-cardinal information gap. Its last version stays readable and citable here.

Version 1 of 120 papers22 conceptsfrozen 2026-10-04

In a perfect world, a voting system would select the candidate who minimizes the total "misery" of the electorate. If we could measure the exact distance between every voter’s ideal position and every candidate’s platform, we could simply pick the candidate at the geometric center of the crowd.

In reality, we rarely have access to these precise cardinal distances. Instead, we have ordinal rankings: "I prefer Candidate A over B, and B over C." The research into metric distortion asks a fundamental question: how much social welfare do we lose by relying on these rankings instead of the underlying distances?

The Geometry of Disagreement

The core assumption in this field is that voters and candidates exist within a metric space. This means the "distance" between a voter and a candidate satisfies the triangle inequality. If a voter is close to Candidate A and Candidate A is close to Candidate B, the voter cannot be arbitrarily far from Candidate B.

As Piotr Skowron and Edith Elkind explore in Social Choice Under Metric Preferences: Scoring Rules and STV, this metric structure is what allows us to bound the "distortion"—the ratio between the social cost of the candidate we pick and the cost of the true (but hidden) optimal candidate. Without the metric assumption, the distortion could be infinite; with it, we can prove that some voting rules are mathematically "safer" than others.

The Limits of Standard Voting Rules

Not all voting rules handle this information gap equally. Common "scoring rules," which assign fixed points to rank positions, often struggle. Yu Cheng, Shaddin Dughmi, and David Kempe 0001 demonstrate in On the Distortion of Voting With Multiple Representative Candidates that while the Borda Count performs reasonably well when candidates are representative of the population, Plurality and Veto voting remain inefficient.

A famous result in this area is the "distortion of 3." For any single-winner election using only ordinal rankings, the best possible worst-case distortion is 3. This means that in the worst possible geometric configuration, the chosen winner might be three times as "far" from the electorate as the ideal candidate. While many scoring rules fail to reach this bound, the Single Transferable Vote (STV) has been shown to achieve a respectable logarithmic distortion.

Breaking the Distortion Barrier

Recent research has focused on how to break past the theoretical limits of ordinal data by introducing small amounts of extra information.

  1. Learning-Augmented Predictions: Vasilis Gkatzelis, Michal Feldman, and their colleagues in Optimal Metric Distortion with Predictions suggest using machine learning to predict the optimal candidate. They prove that even a noisy prediction can help a voting rule achieve a distortion lower than 3, provided the rule is designed to be "robust" (safe if the prediction is wrong) and "consistent" (perfect if the prediction is right).
  2. Strategic Queries: If rankings aren't enough, perhaps a few specific distance questions are. Dimitris Fotakis and Laurent Gourvès show in On the Distortion of Committee Election with 1-Euclidean Preferences and Few Distance Queries that asking just a logarithmic number of distance queries can bring the distortion of multi-winner elections down to a constant factor.

Dimensionality and Multi-Winner Elections

The geometry of the space itself—whether voters sit on a 1D line or a 2D plane—drastically changes the difficulty of the problem. Diptarka Chakraborty, MohammadTaghi Hajiaghayi, and others explore this in Bi-Criteria Metric Distortion.

They uncover a "dimensionality gap": on a simple 1D political spectrum (a line metric), selecting a committee of just two candidates can perfectly approximate the social cost of the best single winner. However, in higher dimensions or more complex tree metrics, achieving that same level of efficiency becomes impossible without selecting a much larger portion of the candidate pool.

This perspective shifts the goal of social choice. Instead of just picking the "best" person, researchers like Elliot Anshelevich are looking at Line-up Elections, where candidates are matched to specific roles. Even in these complex scenarios, the underlying metric structure provides a mathematical anchor, ensuring that ordinal preferences can still lead to outcomes that are provably close to the cardinal optimum.

Papers in this version

History & citationVersion 1 · 2026-10-04

Merged into another direction · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_minimizing_metric_distortion_in_social_c_85a6fe8e,
  title = {Minimizing Metric Distortion in Social Choice},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/minimizing-metric-distortion-in-social-choice/rdr_85a6fe8e5a6643899ef61102550299fa},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_minimizing_metric_distortion_in_social_c_85a6fe8e_v1,
  title = {Minimizing Metric Distortion in Social Choice},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/minimizing-metric-distortion-in-social-choice/rdr_85a6fe8e5a6643899ef61102550299fa/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}