Groups do not outperform experts on specialised questions
Aliases: expertise · wisdom of crowds · collective intelligence
What it is
The limits of crowd wisdom appear when a question requires scarce expertise, training, equipment, or accountable judgment: ordinary averages do not inherently beat qualified experts. Aggregation reduces independent random error; it does not create missing knowledge.
Why it happens
Specialised answers can depend on unseen mechanisms, training, or verifiable evidence. Nonexperts can share common-sense bias, misuse terms, or follow visible signals; more people amplify correlated error. Experts can interpret evidence, identify anomaly, and bear method responsibility.
Studying it
- Paradigms: compare experts, crowds, and mixed flows against known truth or later outcomes.
- Variables: expertise, independence, truth, aggregation, confidence, error type, and cost.
- Methodological caution: expert labels do not guarantee accuracy; compare task-specific performance and uncertainty.
Where it stops holding
Experts have blind spots, conflicts, and local knowledge; crowds can surface anomaly, context, and oversight. Combine evidence, expertise, and public experience by problem rather than choosing one.
Applying it
- State evidence level, accountable expert, and public-input route for consequential specialist work.
- Collect independent observations first, then expert interpretation and review.
- Provide challenge and conflict disclosure for expert views.
- Verification: trace error type, consequence, and who could detect it across decision flows.
Related
- Same group: V9.07.1 Group estimates outperform individuals only when judgments are independent · V9.07.2 Visible choices break independence and induce conformity · V9.07.3 Collaborative filtering reinforces popular items and buries the long tail · V9.07.5 A few early ratings can lock an item's long-term visibility
- Nearby: V6.01 Task allocation · V6.04 Approval and review
- Search terms:
wisdom of crowds·expertise·collective intelligence