Abstract
Citizens often trust experts, but their trust is contingent. Unfortunately, it is more likely to be contingent on fidelity (correspondence to citizens’ preferences) than on validity (correspondence to empirical truth). Still, citizens are far from impervious to validity, and they appear to want experts to be accurate. Citizens appear to assume that experts are unbiased unless their testimony is unexpected (naïve realism), and they seem to assume that an expert’s confidence is warranted unless the evidence shows otherwise (the presumption of calibration). To foster an appropriate level of trust in experts, we need better systems for promoting and assessing expert accuracy and calibration, and I discuss a number of traditional mechanisms as well as emerging innovations.