[Paper Review] Coalitional manipulation for Schulze's rule

Serge Gaspers, Thomas Kalinowski+2 more|arXiv (Cornell University)|May 6, 2013

Game Theory and Voting SystemsEconomics, Econometrics and Finance17 references9 citations

TL;DR

This paper demonstrates that Schulze's rule, despite its strong axiomatic properties, is computationally vulnerable to coalitional manipulation. It proves that unweighted coalitional manipulation (UCM) is polynomial-time solvable for any number of manipulators, resolving an open question, and shows that weighted coalitional manipulation (WCM) is also polynomial when the number of candidates is bounded.

ABSTRACT

Schulze's rule is used in the elections of a large number of organizations including Wikimedia and Debian. Part of the reason for its popularity is the large number of axiomatic properties, like monotonicity and Condorcet consistency, which it satisfies. We identify a potential shortcoming of Schulze's rule: it is computationally vulnerable to manipulation. In particular, we prove that computing an unweighted coalitional manipulation (UCM) is polynomial for any number of manipulators. This result holds for both the unique winner and the co-winner versions of UCM. This resolves an open question in [14]. We also prove that computing a weighted coalitional manipulation (WCM) is polynomial for a bounded number of candidates. Finally, we discuss the relation between the unique winner UCM problem and the co-winner UCM problem and argue that they have substantially different necessary and sufficient conditions for the existence of a successful manipulation.

Motivation & Objective

• To investigate the computational complexity of manipulating Schulze's rule through coalitional voting strategies.
• To resolve the open question on the complexity of unweighted coalitional manipulation (UCM) under Schulze's rule.
• To analyze the tractability of weighted coalitional manipulation (WCM) when the number of candidates is bounded.
• To compare the necessary and sufficient conditions for successful manipulation in the unique winner and co-winner versions of UCM.

Proposed method

• The authors model the manipulation problem as a computational decision problem under Schulze's rule, focusing on strategic voting by a coalition.
• They prove polynomial-time solvability of UCM by constructing an efficient algorithm that determines whether a coalition can ensure a preferred candidate wins or ties for victory.
• For WCM, they analyze the problem under bounded candidate sets and show it remains solvable in polynomial time using combinatorial optimization techniques.
• They distinguish between unique winner and co-winner UCM by analyzing the structural differences in manipulation feasibility conditions.
• The proofs rely on properties of Schulze's rule, such as its Condorcet consistency and path-based ranking mechanism, to derive algorithmic efficiency.

Experimental results

Research questions

• RQ1 Is unweighted coalitional manipulation (UCM) under Schulze's rule computationally tractable for any number of manipulators?
• RQ2 Does the complexity of weighted coalitional manipulation (WCM) remain polynomial when the number of candidates is bounded?
• RQ3 What are the necessary and sufficient conditions for successful manipulation in the unique winner version of UCM?
• RQ4 What are the necessary and sufficient conditions for successful manipulation in the co-winner version of UCM?
• RQ5 How do the conditions for successful manipulation differ between the unique winner and co-winner variants of UCM?

Key findings

• Unweighted coalitional manipulation (UCM) under Schulze's rule is solvable in polynomial time for any number of manipulators, resolving an open question in the literature.
• The polynomial-time solvability holds for both the unique winner and co-winner versions of UCM, indicating high tractability of manipulation under this rule.
• Weighted coalitional manipulation (WCM) is also polynomial-time solvable when the number of candidates is bounded, showing limited resistance to manipulation under weight constraints.
• The necessary and sufficient conditions for successful manipulation differ significantly between the unique winner and co-winner versions of UCM, indicating distinct structural challenges.
• The results reveal a paradox: despite Schulze's rule satisfying strong axiomatic properties, it remains vulnerable to efficient manipulation by coalitions.