People are most likely to trust "what everyone else says" on the exact topics where the crowd is most likely to be wrong.
SSRN · March 18, 2026 · 6429404
The Takeaway
We are taught that groups are smarter than individuals, but this study found that people only seek group consensus on the hardest problems—where the crowd actually performs no better than random chance. We ignore the 'wisdom of crowds' on medium-difficulty tasks where the group is actually most effective, leading to a systematic failure in how we use collective intelligence.
From the abstract
Crowd judgments can be far more reliable than those of individuals – but only if their members are sufficiently competent. Across three studies (N = 306), we show that people are systematically mistaken about when crowds are reliable. Participants answered 12 questions and had four opportunities to view a crowd’s responses. They consulted the crowd on the hardest questions—where the crowd performed at chance—while neglecting the crowd where it greatly outperformed individuals. In Study 2, partic