How can systems limit metric distortion?

A formal-mechanism perspective focuses on settings where distortion can be bounded mathematically rather than eliminated through governance alone. It asks what information is available, how decisions are aggregated, and what worst-case guarantees a rule provides. This lens emphasizes richer preference elicitation, carefully designed voting rules, and explicit bounds on the cost of incomplete or distributed information.

• In metric social choice, voting rules can be judged by the ratio between the chosen option’s social cost and the unknown optimum, making distortion an explicit worst-case objective.
• Under limited ordinal information, some pairwise-comparison mechanisms achieve logarithmic distortion with relatively few comparisons, while using fewer than a threshold number can yield unbounded distortion.
• Distributed aggregation can itself create efficiency loss, even when the full metric space is known, because group representatives are aggregated into a final choice.
• Adding preference intensities can improve distortion relative to relying only on ranked ballots, although the appropriate rule depends on the objective and information available.
• In metric learning, bounding distortion can improve stability and generalization by constraining overfitting and numerical inaccuracy.

Featured sourceBreaking the Metric Voting Distortion Barrierarxiv.org7 sourcesView sources