Game theory paper unveils Vote-Left strategy that triples Faithful win rate
Forget random voting—deterministic coordination crushes Traitor collusion, according to new research.
Vince Knight’s new paper on arXiv (cs.GT, May 2026) tackles the foundational problem in social deduction games: how an uninformed majority (the Faithful) can vote without being exploited by a coordinated minority (the Traitors). The author proposes a deterministic “Vote-Left” protocol: every player votes for the next surviving player in a fixed cyclic ordering. Because the votes are deterministic functions of public information, any deviation from the protocol is immediately identifiable. Combined with a simple punishment rule, this strategy constitutes a Perfect Bayesian Equilibrium for any game state where n_t > 2m_t + 2 (where n_t is surviving Faithful and m_t surviving Traitors). This region covers every configuration seen in televised seasons of The Traitors.
In late-game phases (n_t ≤ 2m_t + 2) the Faithful lose the ability to guarantee punishment, so Traitors can deviate via collusion and break the equilibrium. Crucially, across all televised configurations, Knight shows that Vote-Left raises the Faithful’s winning probability by a factor of approximately three over random voting when Traitors collude. The result bridges game theory and practical strategy: a simple, enforceable coordination protocol that outperforms random voting without requiring any private signals. The work has implications beyond the show, including decentralized consensus protocols and adversarial voting systems.
- Each Faithful votes for the next surviving player in a fixed cyclic order—deterministic and publicly verifiable.
- Protocol is a Perfect Bayesian Equilibrium for all states with n_t > 2m_t + 2, covering every televised game configuration.
- Faithful winning probability triples compared to random voting when Traitors collude (empirical factor ≈3).
Why It Matters
Provides a rigorous, enforceable voting strategy that could reshape real-world social-deduction tactics and consensus mechanisms.