Optimal Signaling of Content Accuracy: Likes vs. Fake News

We study the tradeoff between user engagement and misinformation faced by an online social networking platform. The content available on the platform is possibly erroneous. Agents decide whether to engage with the content based on their direct satisfaction from engaging, the disutility from engaging with erroneous content, and the positive externality that they derive from engaging with the same content as their friends in the underlying social network. The platform knows the error associated with the content and can signal it to the agents so as to induce a desired engagement behavior. We show that the optimal (in terms of engagement/misinformation) signaling mechanisms have a simple threshold structure: the platform recommends that the agents “Engage” with the content if its error is below a threshold and recommends “Do not engage” otherwise. For the mechanism that maximizes engagement, these thresholds depend on agents’ network positions, which we capture through a novel centrality measure that we introduce. Surprisingly, in the case where the platform seeks to only minimize misinformation (regardless of the induced engagement), public signal mechanisms with identical thresholds across agents are optimal. This is in contrast with the engagement maximization setting, where when agents are heterogeneous in terms of their network positions, public signal mechanisms induce substantially lower engagement than the optimal mechanisms. We also study the Pareto frontier of the engagement/misinformation levels that can be achieved via different mechanisms and characterize when public signal mechanisms can achieve Paretoefficient performance levels. Finally, we supplement our theoretical findings with numerical simulations on a Facebook subgraph.