On 26 January 2020, Émile Servan-Schreiber returned to an argument he had made at BlendWebMix in Lyon: technology should help people organise their collective intelligence to solve shared problems. He shared Dorothée Oké’s recap of his talk, which framed the message as a move from “Tech for good” toward “Tech for smart”.
A November conference, a January reprise
The talk belongs to the 2019 conference, rather than to a new January event. OnlySo, which participated in the conference’s live-tweet coverage, dates BlendWebMix to 13–14 November 2019 and identifies its theme as Tech for Good. Its recap includes Servan-Schreiber’s collective-intelligence talk among attendees’ favourites.
Servan-Schreiber’s own post on 15 November says his Lyon conference had taken place the previous day. He thanked Oké for summarising it in under two minutes. An attendee’s account shared on 14 November described a talk about differing opinions and the composition of groups; Servan-Schreiber’s response credited Scott Page and Michel Ferrary among the researchers whose findings he was presenting.
Parce que c’est en n'étant pas d’accord qu’on arrive à être ensemble intelligent.
Dorothée Oké’s recap of Servan-Schreiber’s BlendWebMix talk, shared 15 November 2019
In our translation: It is by disagreeing that we become intelligent together. The short line captures the design problem behind the talk: useful differences have to reach the group rather than disappear under pressure to agree.
The platform has to organise the contributions
A Press Relations Lyon interview excerpt reproduced by Lyon Éco & Culture develops that point. Servan-Schreiber discussed listening and the flow of information among individuals, then asked how to organise the same process at the scale of a government or nation. He proposed digital spaces that encourage independent views and mechanisms that combine what participants contribute.
For forecasting, that argument becomes a practical question about aggregation: how should distinct judgments be turned into a common estimate? The research on minority rewards and prediction markets examines one version of that problem. It offers evidence about specific mechanisms, rather than treating disagreement itself as a guaranteed improvement.
The talk’s design ambition is that technology can help a group share and combine its information. Listening lets an individual understand what another person knows; an aggregation mechanism gives the wider group a way to use those contributions. The forecasting question is whether the resulting estimate performs well when compared with outcomes. Disagreement preserves different possibilities for consideration, while evaluation determines whether the process produced useful judgement.
