On 4 October 2019, the Forum de l’économie vaudoise met at Lausanne’s Salle Métropole under a theme about the growing power of collective intelligence. The date and venue are recorded in Prométerre’s annual report. Its official programme included Émile Servan-Schreiber’s presentation of Supercollectif, alongside contributions from Pierre Dillenbourg, Olivier Viret, Alisée de Tonnac and Laurent Horvath.
Servan-Schreiber had announced his Lausanne keynote on 19 September. On 7 October he shared Olivier Doléans’s recap, endorsing its four practical lessons as worth sharing. The response shows how an audience concerned with planning translated a discussion of group intelligence into questions for organisations.
Four lessons, with different kinds of evidence
- Composition: Doléans argued for diversity rather than simply assembling a team of experts. He linked this to research on women and group intelligence.
- Dissent: he urged organisations to let people express independent, differing opinions, so managers could hear challenges to their assumptions.
- Aggregation: he described reductions in error as more judgements were combined, treating the size of the group as relevant to planning.
- Participation: he connected a collective forecast to what subsequently happens, and urged care in implementing participatory approaches. This is his interpretation of the discussion.
Research does not promise perfect prediction
The original 2010 study by Anita Woolley and colleagues reported a collective-intelligence factor across group tasks. In its two studies, involving 699 participants working in groups of two to five, this factor correlated with social sensitivity, more equal conversational turn-taking and the proportion of women in a group. These were findings in studied groups, without a rule that adding women guarantees better forecasts.
Doléans’s numerical statements about error reduction remain attendee claims rather than verified performance results. The constructive point is the distinction between increasing a group’s size and improving how it handles information. His call for independent dissent asks managers to hear different judgements before deciding what the collective believes, instead of equating participation with agreement.
For forecasters, the next step is to check whether the combined estimate improves on alternatives. Our review of Hypermind’s long-term calibration describes a measurable standard: compare many stated probabilities with the frequency of the corresponding outcomes. The Lausanne discussion brought that organisational question to an audience interested in planning. Its case for diversity and dissent can guide how a group is assembled, while accuracy still needs to be demonstrated through results.
