Why it is a question
Handfield, Garcia, Hilbe and Yeo (LIT-966, read in full in NOTE-795) model deductive debate as a game whose payoff trades true beliefs against changes of mind. In their two-debater, three-proposition model, the most truth-conducive equilibrium is often a snowdrift: one debater argues boldly and the other free-rides with a compromise strategy, and in a narrow band of mixed equilibria a higher weight on truth lowers equilibrium accuracy (THEORY-290). Every exact result is for two players and n = 3, and the strategies are fixed rules over current positions. The paper’s own “ideal cooperative debater”, who targets the partner’s false beliefs, lies outside that space. NOTE-795 lists both extensions among its open questions. Until 2026-10-19 they were THEORY-290’s promote_when. THEORY-290’s body disclaimed them, so the clause moved here when the title was narrowed to the model.
What would answer it
An analysis or experiment that measures accuracy and revision costs separately and finds the same free-riding structure among more than two debaters, or with richer strategies than fixed, memoryless premise and conclusion rules. In such a study, a participant facing a bold partner must argue less boldly, collective accuracy must fall short of what mutual boldness would give, and the shortfall must vanish when the cost of revision is removed. It would answer no if, with strategies that can target the partner’s false beliefs (the paper’s “ideal cooperative debater”), the truth-optimal profile is an equilibrium wherever the truth weight is positive.
What would not answer it: more runs of the two-person, three-proposition model, and observational data on debates, which cannot measure revision costs.