Big Data and Democracy

Freek van Gils (Tilburg University)
Wieland Müller (University of Vienna and Tilburg University)
Jens Prüfer (Tilburg University)

Abstract: Despite numerous reports in the public press, there is little evidence for successful manipulation of democratic elections by political interest groups' disinformation campaigns. We ask whether such structural disinformation and election manipulation is theoretically possible in equilibrium. We construct a model, where ideologically heterogeneous news outlets send messages about a politically relevant subject via a news platform (e.g. social media or news aggregator) to ideologically heterogeneous voters, who then make election decisions. We vary the communication technology news outlets have access to: they are either restricted to sending the same message to all voters or they can microtarget voters with individualized messages; voters can be aware of the political position of a news sender, or not. We rank the resulting four games in terms of voter welfare and efficient information utilization and show when an election can be flipped in equilibrium. We discuss policy proposals and conclude that two policies appear favorable: to require news platforms to signal the political position of a message's originator clearly and to support local news outlets.