Strategic timing of disclosure in belief formation

Abstract This paper studies the optimal timing of disclosure in a belief formation framework where agents are influenced by two persuaders: a proponent and an opponent. Opinions evolve according to a modified De Groot learning process with time-varying weights. The proponent chooses when to disclose her narrative to maximise belief alignment before a deadline, while the opponent chooses when to disclose a refutation. We show that once the proponent discloses her narrative, it is optimal for the opponent to immediately disclose her refutation. We determine the optimal disclosure timings in both one agent and n -agents settings. We analyze settings where agents are connected on specific network topologies, including symmetric networks with homophily, complete and star networks, while distinguishing supporters and non-supporters. We find that stronger homophily amplifies polarisation and leads to earlier disclosure when the proponent places greater weight on supporters.