Use structured team forecasting to aggregate diverse views

Combine independent estimates from multiple people before the group talks — aggregation beats any single expert.

Why it works

Group discussion tends to amplify the views of the most confident or senior member and suppress minority information. Aggregating independent estimates before discussion preserves information that discussion would wash out. The wisdom of crowds effect depends on independence; the mechanism breaks down the moment estimates are made in the presence of others’ views.

How to do it

1. Have each team member write down a probability estimate independently before the group discussion.
2. Collect and average the estimates (or use a simple aggregate like the median).
3. Present the aggregate alongside the individual estimates, then open the discussion.
4. After discussion, allow individual updates, but re-aggregate rather than moving to the group’s modal view.

Evidence

The wisdom of crowds effect — that aggregated independent estimates outperform most individuals, including experts — is well documented (Galton, Surowiecki). Tetlock’s tournament used aggregated team scores and found structured teams outperformed individuals; the superiority depended on preserving independence before aggregation. (observational)

The wisdom of crowds requires genuine independence; if people anchor on a shared prior or a visible anchor before giving their estimates, the aggregation adds little to the best single view.

Sources

• Galton (1907), "Vox Populi," Nature
• Tetlock & Gardner (2015), Superforecasting, on team performance

Common mistake

Running a group discussion and then averaging the post-discussion estimates — by that point, independence is already lost and the "aggregate" reflects the most persuasive voice, not diverse perspectives.