Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
This article argues that, in the presence of dispersed information, individual‐level idiosyncratic noise may propagate at the aggregate level when agents are connected through a social network. When information about a common fundamental is incomplete and heterogeneous across agents, it is beneficial to consider the actions of other agents because of the additional information conveyed by these actions. We refer to the act of using other agents' actions in the individual decision process as social learning. This article shows that social learning aimed at reducing the error of individual actions with respect to the fundamental may increase the error of the aggregate action depending on the network topology. Moreover, if the network is very asymmetric, the error of the aggregate action does not decay as predicted by the law of large numbers.