Agreeing to be fooled: Optimal ignorance about information sources

Abstract

Should a decision-maker learn whether an information source is reliable? I consider a persuasion game in which the sender is sometimes unreliable— i.e., can covertly manipulate the signal used to persuade the receiver—and the receiver can costlessly investigate the sender’s reliability. I show that the receiver benefits from committing to investigations that do not always reveal reliability. Even without the ability to commit to ignorance, I demonstrate that the receiver benefits from delegating investigations to someone partially adversarial to the sender and partially aligned with the receiver. My results shed light on the efficacy of cross-examination, audits, and ad hominem arguments.