STV Voting Highly Vulnerable to No-Show Paradox in 1D Preferences, Study Finds

Computer scientist Farhad Mohsin has published a paper on arXiv (2606.12785) that upends conventional wisdom about the robustness of ranked-choice voting systems. The group no-show paradox (GNSP) occurs when a set of voters abstains from an election, and the new winner is actually more preferred by that group than the original winner. While GNSP was previously considered rare in real-world elections, Mohsin shows that under one-dimensional preference models—such as 1D-Euclidean, single-peaked, or single-crossing preferences—the Single Transferable Vote (STV) rule is alarmingly vulnerable.

In stark contrast, Condorcet-consistent voting rules, which also exhibit GNSP under general preferences, never produce the paradox when preferences are one-dimensional. Mohsin theoretically identifies tractable and prevalent sufficient conditions for GNSP to occur under STV in these domains. Through both theoretical analysis and experiments with synthetic preference profiles, he demonstrates that the likelihood of GNSP increases substantially as the number of alternatives grows, and that voters at the extremes of the one-dimensional spectrum are the most likely to trigger it by strategically (or accidentally) abstaining. The findings have direct implications for electoral system design in organizations, political bodies, and AI-based multi-agent decision systems that use STV-like mechanisms.