Signal When does the public become suspicious of bots? Demand-side evidence from Botometer query logs

Summary

The paper studies, from the demand side, when people come to suspect that an account is automated. It looks specifically at whom people suspect and what actions follow that suspicion. The researchers used server-side query logs from Botometer, the most widely used bot-detection service. In these logs, each query is treated as a behavioral trace showing that suspicion was acted upon. The authors analyze a large number of these traces to characterize patterns of public suspicion toward bots. This approach makes it possible to examine private, individual-level bot-checking behavior at scale.