pith. machine review for the scientific record. sign in

arxiv: 2605.10233 · v1 · submitted 2026-05-11 · 💻 cs.GT · math.PR

Recognition: no theorem link

The Vote-Left Equilibrium: A Deterministic Coordination Strategy for the Faithful in The Traitors

Vince Knight
Authors on Pith no claims yet

Pith reviewed 2026-05-12 05:06 UTC · model grok-4.3

classification 💻 cs.GT math.PR
keywords social deduction gameThe Traitorsvoting strategyPerfect Bayesian Equilibriumdeterministic protocolcoordinationgame theory
0
0 comments X

The pith

The Vote-Left protocol establishes a Perfect Bayesian Equilibrium for the Faithful in The Traitors, tripling their winning probability over random voting when Traitors collude.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces the Vote-Left protocol as a deterministic coordination strategy for the Faithful in the social deduction game The Traitors. Each player votes for the next surviving player in a pre-agreed cyclic order, producing exactly one vote per player under full compliance and matching the banishment distribution of random voting. Any deviation is immediately detectable because votes are deterministic functions of public information. Combined with a punishment rule targeting deviations, Vote-Left forms a Perfect Bayesian Equilibrium in every state where the total number of players exceeds twice the number of Traitors plus two, a condition met by all televised configurations. This approach raises the Faithful's winning probability by a factor of approximately three compared to random voting when the Traitors collude.

Core claim

Vote-Left is the deterministic rule where every surviving player votes for the next surviving player in a fixed cyclic ordering. Full compliance yields a uniform vote distribution equivalent to random voting, but deviations are instantly identifiable from public information alone. With a simple punishment rule for detected deviations, this constitutes a Perfect Bayesian Equilibrium for all states with n_t greater than 2m_t plus 2. The region includes every configuration played on television. The Traitors' best response in the late-game phase when n_t is less than or equal to 2m_t plus 2 is to collude and deviate, as the Faithful lack sufficient votes to guarantee punishment. Across televised

What carries the argument

The Vote-Left protocol: each player votes for the next player in a pre-agreed cyclic ordering of survivors. This mechanism produces uniform votes under compliance while allowing immediate detection of any non-compliance.

If this is right

  • Full compliance produces the same banishment distribution as random voting.
  • Any deviation is immediately identifiable and punishable.
  • The strategy is a Perfect Bayesian Equilibrium for states with n_t > 2m_t + 2.
  • Traitors optimally collude only in late-game states where n_t ≤ 2m_t + 2.
  • The Faithful's win probability increases by a factor of approximately three in televised configurations.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If players can publicly agree on a cyclic order at the start, the protocol could apply to similar voting games with hidden roles.
  • Simulations of human play could test whether the required strict adherence to the protocol is realistic.
  • The late-game characterization suggests a phase transition in strategy that might be observable in actual episodes.

Load-bearing premise

The Faithful players will all adopt and strictly follow the Vote-Left protocol along with the associated punishment strategy for detected deviations, and that the initial cyclic ordering is publicly agreed upon and fixed.

What would settle it

A direct comparison of the Faithful's winning probability in simulated games matching televised configurations, using Vote-Left with punishment versus random voting under collusion; failure to observe an approximate threefold increase would falsify the performance claim.

Figures

Figures reproduced from arXiv: 2605.10233 by Vince Knight.

Figure 1
Figure 1. Figure 1: Outcomes by country, 2021–2026. More than 80 seasons of the show have been played across approximately 30 countries. The overall win rate shows a clear advantage to the Traitors. A Roles F F F F F F F F T T 1 2 3 4 5 6 7 8 B Night: Murder F F F F F F F F T T × Traitors coordinate to murder player 5 C Day: Banishment F F F F F T T × banished All players vote publicly Most votes → eliminated D Random Voting … view at source ↗
Figure 2
Figure 2. Figure 2: The Traitors game and voting strategies. A, Eight players (F for Faithful in blue, T for Traitor in red) with two Traitors. B, Night phase: the Traitors coordinate to murder a Faithful (player 5). C, Day phase: the public vote leads to banishment. D, Under random voting, Traitor collusion (the two red arrows converging on player 1) is indistinguishable from random noise. E, Under Vote-Left, each player vot… view at source ↗
Figure 3
Figure 3. Figure 3: Recursion relationship for w(n, m). A, Initial state for w(8, 2), given random voting the next step could either be B, or D. This leads immediately to C, or E respectively. 3 Equilibrium Analysis 3.1 Strategy profile Definition 2 (Vote-Left with Punishment). Vote-Left with Punishment is the strategy profile σ ∗ , governed by a public state s ∈ {comply, punish(j)} common to all players: • In state comply, e… view at source ↗
Figure 4
Figure 4. Figure 4: Monte Carlo simulation loop. One pass of the inner loop simulates a single game of TG(n, m). The parity win condition is checked before the day phase; if Traitors already hold parity the game ends immediately. During the day phase, votes are cast and any deviation from the prescribed protocol is detected: if a deviation is found the deviator is punished (all surviving players vote for them); otherwise the … view at source ↗
Figure 5
Figure 5. Figure 5: Strategic game progression. Top row: voting patterns under RV (random, uncoordinated), σ ‡ (RV+C, random with Traitor collusion), and VL (cyclic Vote-Left). Bottom row: the Traitors’ optimal response σ † (VL+Opt) branches on the credibility of Faithful punishment: when n > 2m + 2 (left), Traitors comply with the cyclic pattern; when n ≤ 2m + 2 (right), the Faithful lack the numbers to punish, so Traitors c… view at source ↗
Figure 6
Figure 6. Figure 6: Strategy comparison: Faithful win rate by strategy. Each panel shows the Faithful win rate 1 − w as a function of n for traitor counts m ∈ {2, 3, 4, 5}. Where a closed-form recurrence exists, exact values are shown as crosses; simulated values (lines, with 95% confidence intervals) are shown alongside for verification. A, Random Voting (RV): Faithful win rate under mutual random play. B, Random Voting with… view at source ↗
Figure 7
Figure 7. Figure 7: Traitor gain from σ † (VL+Opt). Each panel shows the ratio of the Faithful win rate under VL+Opt to that under an alternative strategy. Exact closed-form values are used where available; simulated values are used otherwise. Points are omitted where the simulated denominator records no Faithful wins. Values below one favour the Traitors; the dashed line marks equality. A, VL+Opt / RV: near but below one, co… view at source ↗
read the original abstract

The Traitors is a social deduction game in which an informed minority of Traitors face an uninformed majority of Faithful, and the recurring question facing the Faithful is how to vote. Random voting is known to be optimal for the uninformed majority under simultaneous-signal protocols [Braverman, Etesami and Mossel, 2008], but when votes are cast individually, random votes are indistinguishable from strategic ones and the Faithful remain exposed to coordinated Traitor collusion. We introduce the Vote-Left protocol, a deterministic rule under which every player votes for the next surviving player in a fixed cyclic ordering. Under full compliance every surviving player receives exactly one vote, so the banishment distribution coincides with random voting; since prescribed votes are deterministic functions of public information, any deviation is immediately identifiable. Combined with a simple punishment rule, Vote-Left constitutes a Perfect Bayesian Equilibrium for every state with $n_t > 2m_t + 2$, a region that contains every televised configuration. We characterise the Traitors' best response in the late-game phase ($n_t \leq 2m_t + 2$): deviate via collusion once the Faithful no longer have enough votes to guarantee punishment. Across the configurations played on television, Vote-Left raises the Faithful's winning probability by a factor of approximately three over random voting under collusion.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces the 'Vote-Left' protocol, a deterministic voting strategy for the Faithful in the game The Traitors, where players vote for the next in a fixed cyclic order. It claims that, paired with a punishment rule for deviations, this forms a Perfect Bayesian Equilibrium whenever the number of surviving players n_t exceeds 2 times the number of traitors m_t plus 2. This condition is said to hold for all televised game configurations. The paper further characterizes the Traitors' best-response collusion strategy in the late game (when n_t ≤ 2m_t + 2) and reports that Vote-Left approximately triples the Faithful's winning probability compared to random voting under collusion, based on enumeration of televised configurations.

Significance. If the equilibrium proof and the probability calculations are rigorous, the result provides a concrete, implementable strategy for coordination in social deduction games with asymmetric information. The deterministic and observable nature of the voting rule allows for immediate detection of deviations, addressing a key limitation of random voting. The application to real televised episodes adds empirical relevance, and the threshold condition offers a clear boundary for when the strategy is sustainable. This could inspire similar deterministic protocols in other multi-agent systems with hidden adversaries.

major comments (2)
  1. Abstract: The central claim that Vote-Left constitutes a PBE for every state with n_t > 2m_t + 2 relies on the punishment rule deterring all deviations, but the abstract provides no derivation or argument for why this inequality is the precise threshold; the full proof in the main text is necessary to substantiate this load-bearing condition for the equilibrium result.
  2. Abstract: The approximate threefold improvement in winning probability is stated as resulting from enumeration over televised configurations, but without details on the specific configurations, the exact probability calculations, or the baseline random voting under collusion, it is not possible to assess the accuracy of the factor of three.
minor comments (2)
  1. Abstract: The notation n_t and m_t is used without prior definition; it should be introduced explicitly as the number of surviving players and traitors at stage t.
  2. Abstract: The paper mentions 'a simple punishment rule' but does not specify its details; a brief description would improve clarity for readers.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful comments, which highlight opportunities to improve the clarity of the abstract. We address each major comment below and will revise the manuscript accordingly to incorporate brief references to the supporting analysis and data.

read point-by-point responses

  1. Referee: Abstract: The central claim that Vote-Left constitutes a PBE for every state with n_t > 2m_t + 2 relies on the punishment rule deterring all deviations, but the abstract provides no derivation or argument for why this inequality is the precise threshold; the full proof in the main text is necessary to substantiate this load-bearing condition for the equilibrium result.

    Authors: We agree the abstract is concise and omits the derivation. The threshold n_t > 2m_t + 2 is derived in Section 3 of the main text: it is the minimal condition under which the Faithful retain a strict voting majority sufficient to credibly punish any unilateral deviation (even when the remaining Traitors vote strategically to protect a defector). With this majority, the punishment strategy is incentive-compatible and deters all deviations, establishing the PBE. We will revise the abstract to include a short parenthetical reference to this justification and the relevant section. revision: yes

  2. Referee: Abstract: The approximate threefold improvement in winning probability is stated as resulting from enumeration over televised configurations, but without details on the specific configurations, the exact probability calculations, or the baseline random voting under collusion, it is not possible to assess the accuracy of the factor of three.

    Authors: The full manuscript (Section 5 and Appendix B) enumerates all televised configurations from the UK and US series, computing exact win probabilities for the Faithful under Vote-Left (with punishment) versus the Traitors' optimal collusion strategy against random voting, via exhaustive state-by-state enumeration. The reported factor of approximately three is the average ratio across these episodes. We will revise the abstract to reference the section containing the table of configurations and probabilities, and we will make the raw enumeration data available as supplementary material. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation is self-contained

full rationale

The paper explicitly defines the Vote-Left rule as a deterministic function of observable public state (surviving players and a pre-agreed cyclic order), specifies an explicit punishment strategy for any deviation, and verifies that the resulting strategy profile meets the standard definition of Perfect Bayesian Equilibrium precisely when n_t > 2m_t + 2. The claimed factor-of-three improvement in Faithful win probability is obtained by direct enumeration over the finite set of televised configurations rather than any parameter fitting or self-referential construction. The only citation is to external prior work on random voting; no self-citation is load-bearing, no uniqueness theorem is imported from the authors' own prior results, and no ansatz or known empirical pattern is renamed as a new derivation. The central claims therefore reduce to the paper's own definitions and standard equilibrium verification rather than to their own inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The analysis relies on standard game theory axioms for incomplete information games and the specific mechanics of voting in The Traitors; no new entities are postulated and no free parameters are introduced in the abstract description.

axioms (2)
  • domain assumption Players are rational Bayesian agents who maximize their expected probability of winning the game.
    Standard assumption in game-theoretic modeling of strategic interactions in social deduction games.
  • domain assumption The game proceeds with observable individual votes and public information about surviving players.
    Based on the rules of The Traitors as described.

pith-pipeline@v0.9.0 · 5555 in / 1432 out tokens · 43691 ms · 2026-05-12T05:06:44.852356+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

7 extracted references · 7 canonical work pages

  1. [1]

    Mafia: A theoretical study of players and coalitions in a partial information environment.The Annals of Applied Probability, 18(3):825–846, 2008

    Mark Braverman, Omid Etesami, and Elchanan Mossel. Mafia: A theoretical study of players and coalitions in a partial information environment.The Annals of Applied Probability, 18(3):825–846, 2008. 13

    work page 2008

  2. [2]

    Mafia, 1986

    Dimitry Davidoff. Mafia, 1986. Party game

    work page 1986

  3. [3]

    Data for the vote-left equilibrium: A deterministic coordination strategy for the faithful in the traitors paper, May 2026

    Vincent Knight. Data for the vote-left equilibrium: A deterministic coordination strategy for the faithful in the traitors paper, May 2026

    work page 2026

  4. [4]

    A mathematical model of the mafia game

    Piotr Migda l. A mathematical model of the mafia game. arXiv:1009.1031, 2010

    work page arXiv 2010

  5. [5]

    Ten simple rules for making research software more robust

    Morgan Taschuk and Greg Wilson. Ten simple rules for making research software more robust. PLOS Computational Biology, 13(4):e1005412, 2017

    work page 2017

  6. [6]

    Optimal strategy in the werewolf game: A theoretical study.Games: Research and Practice, 2024

    Sizhe Wang. Optimal strategy in the werewolf game: A theoretical study.Games: Research and Practice, 2024. arXiv:2408.17177

    work page arXiv 2024

  7. [7]

    Greg Wilson, D. A. Aruliah, C. Titus Brown, Neil P. Chue Hong, Matt Davis, Richard T. Guy, Steven H. D. Haddock, Kathryn D. Huff, Ian M. Mitchell, Mark D. Plumbley, Ben Waugh, Ethan P. White, and Paul Wilson. Best practices for scientific computing.PLOS Biology, 12(1):e1003745, 2014. 14

    work page 2014