Deliberation and the Wisdom of Crowds

Does group deliberation increase group competence, measured by the correctness probability of majority decisions? We present a model of opinion formation based on sources, and a non-game-theoretic model of deliberation as sharing and absorbing. Two jury theorems, one pre-deliberation and one post-deliberation, suggest that deliberation tends to improve group decisions. Three major voting failures are: (1) overcounting widespread evidence — evidence with wider spread gets more impact; (2) neglecting evidential inequality — voters get equal weight despite having unequally strong evidence; (3) neglecting evidential complementarity — information that follows by combining evidences dispersed across voters remains undercounted because few or no voters access the full combination. Simulations and theoretic arguments suggest that deliberation normally reduces all three failures. But there exist systematic exceptions where deliberation increases Failure 1, sometimes even to the extent that group competence overall falls. Our analysis recommends ‘even’ deliberation, which privileges neither certain evidences nor certain persons.

Links to the paper, the seminar recording, and the slides.