On 8 May 2025, Émile Servan-Schreiber met the Frenchfounders community in Madrid for a discussion of collective intelligence. His next-day update described a rewarding exchange. In the account he reshared, Florence de Vesvrotte thanked the Institut français Madrid for hosting the gathering in its media library.
The recap described diversity as a potential source of better performance, provided organisations know how to use it. The event addressed business leaders, but the underlying question also applies to forecasting: how can different judgements become a useful collective answer? Bringing several people into a room creates an opportunity to compare information; it still leaves the method of comparison to be designed.
Three conditions in the recap
De Vesvrotte singled out three principles from Servan-Schreiber’s presentation: diversity of opinions, independence of thought and objective aggregation of ideas. In plain language, a group needs different perspectives, room for people to form their own view, and a process for combining those views.
She also presented humility as a leadership requirement and warned that ego can obstruct performance. In that account, a leader’s contribution includes making room for information that challenges the leader’s own view. Those are the recap’s management conclusions, rather than measured effects on the participating companies.
Why independence matters
Research offers a concrete reason to take the independence point seriously. In a 2011 PNAS experiment, Jan Lorenz and colleagues let participants revise factual estimates after seeing other people’s answers. Social information narrowed the spread of estimates without a corresponding improvement in collective accuracy, while increasing confidence. The study concerned estimation tasks, so its results cannot simply be treated as a test of every executive meeting.
The implication for the Madrid discussion is practical: apparent agreement may conceal shared error. Gathering an initial judgement from each participant before exposing them to the group’s view is one way to preserve information for later comparison. That is an editorial application of the research, not a procedure documented at this event.
Our account of Servan-Schreiber’s discussion of expertise and diversity explores the same distinction. The Madrid discussion links group composition with a process for hearing and combining different views. For forecasting, the next step is to turn those judgements into estimates that can be compared with outcomes.
