The spread of misinformation in networks with individual and social learning

Abstract

The paper studies the spread of misinformation when agents have both individual and social learning. In the presence of sources of misinformation, the most vulnerable and harmful agents are those with poor individual learning and high centrality in the network. However, if opinions are polarized, providing some agents with better private information can backfire on other agents. Moreover, the incentives to spread misinformation have an inverted U-shaped relationship with the spreader’s influence; and, under certain conditions, an unaware spreader can be more harmful than a purposeful one. Overall, this paper warns that any policy that fails to consider both the precision of agents’ private information and the network structure runs the risk of backfiring.