Rationalizations of Condorcet-Consistent Rules via Distances of Hamming Type

The main idea of the {\em distance rationalizability} approach to view the voters’ preferences as an imperfect approximation to some kind of consensus is deeply rooted in social choice literature. It allows one to define (“rationalize”) voting rules via a consensus class of elections and a distance: a candidate is said to be an election winner if she is ranked first in one of the nearest (with respect to the given distance) consensus elections. It is known that many classic voting rules can be distance rationalized. In this paper, we provide new results on distance rationalizability of several Condorcet-consistent voting rules. In particular, we distance rationalize Young’s rule and Maximin rule using distances similar to the Hamming distance. We show that the claim that Young’s rule can be rationalized by the Condorcet consensus class and the Hamming distance is incorrect; in fact, these consensus class and distance yield a new rule which has not been studied before. We prove that, similarly to Young’s rule, this new rule has a computationally hard winner determination problem.

💡 Research Summary

The paper revisits the distance‑rationalizability framework—a paradigm that models voters’ preferences as imperfect approximations of an ideal consensus—and applies it to several Condorcet‑consistent voting rules. In this framework a voting rule is defined by a pair (C, d) where C is a consensus class (the set of “ideal” elections in which a particular candidate is ranked first) and d is a distance function measuring how far a real election is from any ideal election. A candidate wins if she is first in at least one consensus election that is closest to the actual election under d.

The authors first challenge a widely cited claim that Young’s rule can be rationalized by the Condorcet consensus class together with the ordinary Hamming distance. By constructing a variant of the Hamming distance, denoted d′_H, that counts the minimum number of pairwise swaps needed to turn a real ballot into one that respects a Condorcet winner, they show that (Condorcet, d′_H) does not reproduce Young’s rule. Instead, this combination defines a previously unknown rule: even when a Condorcet winner exists, the optimal set of swaps may promote a different candidate to the top position. The paper formally characterizes this new rule and proves that its winner‑determination problem is NP‑hard, using a reduction from the Maximum Clique problem.

Next, the authors provide a correct distance‑rationalization of Young’s rule. They introduce the “minimum‑deletion distance” d_del, which for a given candidate counts the smallest number of voters that must be removed so that the candidate becomes a Condorcet winner. When paired with the Condorcet consensus class, (Condorcet, d_del) exactly yields Young’s rule. The authors give a rigorous proof that the distance minimization coincides with Young’s definition of the smallest deletion set, thereby confirming the rationalizability of Young’s rule within this framework.

The paper then turns to the Maximin rule. A new distance, called the “minimum‑winner distance” d_max, is defined. For each candidate it measures how far the election is from one in which the candidate’s worst pairwise margin (the minimum of all pairwise defeats) is maximized. By showing that minimizing d_max under the Condorcet consensus class forces the selection of the candidate with the highest worst‑case margin, the authors demonstrate that (Condorcet, d_max) rationalizes the Maximin rule.

Complexity results are a central contribution. For the newly discovered rule based on (Condorcet, d′_H) the authors construct a polynomial‑time reduction from the Maximum Clique problem, establishing NP‑hardness of winner determination. For Young’s rule, they revisit known NP‑hardness results and show that the distance‑rationalization via d_del does not simplify the computational problem. The Maximin rationalization, while not altering the known polynomial‑time solvability of Maximin, provides a fresh perspective on why the rule is computationally tractable: the underlying distance can be computed efficiently via pairwise margin tables.

In the discussion, the authors argue that distance‑rationalization is more than a descriptive tool; it offers a systematic way to generate new voting rules by varying the consensus class or the distance metric. The discovery of a novel rule from a seemingly natural combination (Condorcet + Hamming‑type distance) illustrates that many unexplored rules may exist within the same framework. Moreover, the complexity analyses show that the difficulty of winner determination is often inherited from the combinatorial nature of the distance minimization, rather than from the consensus class itself.

Overall, the paper makes three key contributions: (1) it corrects a misconception about Young’s rule and the Hamming distance, (2) it provides clean distance‑rationalizations for Young’s and Maximin rules using natural, Hamming‑like distances, and (3) it establishes NP‑hardness for the new rule while reaffirming known hardness results for Young’s rule. These results deepen our theoretical understanding of Condorcet‑consistent rules and open avenues for future research into distance‑based design and analysis of voting systems.