A model of debates: Moderation vs free speech

This paper provides a framework to study communication conflicts, such as political debates, using a novel model of competition in Bayesian persuasion. Debating parties can "frame" their arguments for maximal impact. They also can "spam" the discussion to distract the audience from the opponent's arguments. We find that spamming is more detrimental to truth discovery than framing. When parties are allowed to speak freely, spamming can kill truth discovery and make communication uninformative. By contrast, framing is disciplined by competition. If the conflict between parties is strong and the number of arguments is restricted, the parties reveal the truth.