Social information narrowed a crowd’s diversity.
Lorenz and colleagues gave 144 participants factual numerical estimation tasks. Some participants later saw aggregate or fuller information about other estimates. In the reported conditions, social information narrowed response diversity without improving collective error, while confidence increased.
The finding matters because collective estimates can depend on the diversity and independence of their inputs. Convergence can look reassuring without creating a better collective answer.
Read the evidence literally.
Numerical estimates with objectively assessable answers.
Reduced diversity after social information, without improved collective error in the reported experiment.
That preference groups have correct answers, or that every social cue harms every decision.
Let each participant commit before exposing the room’s aggregate split.
The interpretation received a published challenge.
Farrell’s reanalysis argued that social information benefited individual estimates. The original authors replied that individual improvement can coexist with reduced collective diversity. Showing both publications matters: “better individual estimate” and “healthier collective information” are not always the same claim.
ViewsMeet preferences have no correct group answer.
Group Picks asks low-stakes either-or preferences, not factual estimation questions. It cannot claim to make a group wiser, representative, democratic, or scientifically valid. The defensible translation is narrower: hidden first responses prevent the displayed split from becoming the first prompt.
ASK FIRST · REVEAL SECOND
Open the split at three.
Group Picks lets everyone answer the same frozen set independently. Question-level aggregates appear only after three retained responses; individual answer sheets never appear.
Create a five-question room →Original sources and published exchange.
- Lorenz et al. (2011), DOI 10.1073/pnas.1008636108
Original social-influence experiment. - Farrell (2011), DOI 10.1073/pnas.1109947108
Published critique and reanalysis. - Lorenz et al. reply, DOI 10.1073/pnas.1111007108
Response distinguishing individual improvement from collective diversity.