Crowd Predictions Exposed Wrong Majorities

In 2017, Princeton University and MIT researchers reported that a "surprisingly popular" answer could overcome an incorrect majority. Their algorithm asks people for both their answer and a prediction of how popular each possible answer will be, then selects the response that exceeds expectations. In one test, most respondents incorrectly said Philadelphia was Pennsylvania's capital, but more people rejected that answer than the crowd had predicted. Across several topics, the method reduced errors by 21.3 percent compared with simple majority voting and by 48 percent across 50 state-capital questions.

Previous Fact Next Fact