Events

Abstract: The search for effective misinformation interventions has produced widespread disappointment — but the disappointment is misplaced. I argue that it stems from two interconnected problems: unrealistic expectations about what behavioral science can deliver, and a competitive framing that pits interventions against each other rather than exploring how they might work together. The replication crisis revealed that the dramatic effect sizes celebrated in the pre-replication era were artifacts of publication bias, yet these inflated benchmarks continue to shape how we judge intervention efficacy. Meanwhile, researchers — myself included — have pursued a horserace mentality, asking which single intervention wins rather than how multiple approaches might combine. I propose three correctives. First, recalibrate expectations: small, reliable effects deployed at scale represent genuine progress, not failure. Second, adopt a "Swiss cheese model" of information defense, bundling imperfect interventions that operate through different mechanisms to provide layered protection. Third, evaluate interventions using truth discernment — the ability to distinguish true from false content — rather than measuring only reductions in false belief.

The Political Science Seminar Series is convened by Desmond King and David Rueda For more information on this or any of the seminars in the series, please contact politics.secretary@nuffield.ox.ac.uk.