On 8 June 2021, Émile Servan-Schreiber shared a press release, dated the day before in Paris, announcing that the French newsweekly Le Point and Hypermind were creating a joint forecasting body, the Institut Supercollectif. The idea was to ask the magazine’s readers what will happen, rather than what they think, and to turn their answers into probabilities that could be published and, eventually, sold. The trade site The Media Leader FR reported the launch the following day.
The idea behind the institute
The release rests on a familiar premise from the science of collective intelligence: there is a wisdom of crowds, and diversity of viewpoints serves reason better than isolation. The institute would build an international online community of news enthusiasts, professionals and specialists. Their forecasts would then be combined using prediction markets and other aggregation algorithms.
Le Point’s director, Étienne Gernelle, presented it as an extension of the magazine’s founding identity:
L’intelligence de nos lecteurs est notre bien le plus précieux.
Étienne Gernelle, Le Point–Hypermind press release, 7 June 2021
In English: our readers’ intelligence is our most precious asset. Gernelle linked the project to the “équipe de pensée”, or thinking team, the phrase Le Point’s founder Claude Imbert used for the magazine. Servan-Schreiber, who leads Hypermind, described the readership in similar terms:
Les lecteurs et lectrices du « Point » constituent un formidable vivier d’intelligence collective.
Émile Servan-Schreiber, Le Point–Hypermind press release, 7 June 2021
In English: Le Point’s readers form a remarkable pool of collective intelligence. He added that the partnership would bring their forecasts into public debate and make them available to decision makers.
How it was meant to work
- Forecasting contests on a dedicated site, lepoint.hypermind.com, covering political, geopolitical, economic, scientific and industrial questions in France and abroad.
- Further questions sponsored by public or private organisations facing a particular uncertainty.
- Forecasts consolidated into probabilities that update in real time as the news changes.
- For participants, rewards scaled to how accurate their forecasts proved, personal performance tracking and recognition, and a community for discussion.
The release described Hypermind as a prediction-market pioneer since 2000, a reference to its origins as NewsFutures. It also said Hypermind had contributed to several large US government research programmes on the wisdom of crowds.
The first contests
A presidential market opened within days, ten months before the 2022 election. On 16 June 2021, Servan-Schreiber wrote that Macron was crushing the match for now, but that the campaign had barely begun. Further contests followed:
- In July 2021, Le Point invited readers to become news forecasters. Among the contests it described was one on progress in artificial intelligence, designed with Jacob Steinhardt of UC Berkeley and backed by Open Philanthropy, with €25,000 in prizes spread over four years.
- In September 2021, an international contest on worldwide COVID-19 mortality, with €12,600 in prizes, according to the post announcing it.
- In February 2022, a contest on the Africa of the future, which opened with forecasts on Côte d’Ivoire.
The election panel produced the partnership’s best-documented result. In May 2022, Servan-Schreiber reported in Le Point that the readers’ collective forecast of the Macron–Le Pen run-off had been closer to the result than the daily rolling polls in 128 of 143 comparisons.
A year on: from contest to service
By July 2022 the venture had a new name and a commercial offer. On 13 July, Servan-Schreiber shared a photo of a full-page advertisement in Le Point for the Institut prédictif Hypermind–Le Point, headlined “Les lecteurs du «Point» sont d’excellents prévisionnistes” (Le Point’s readers are excellent forecasters). The advertisement said the institute consolidated the informed forecasts of thousands of readers worldwide for businesses, governments, think tanks and NGOs. It offered three services: quantified forecasts with horizons of a few weeks to several years, real-time dashboards with analysis, and promotion of the contests to Le Point’s readers with editorial follow-up of the results.
Servan-Schreiber wrote that, with the partnership, Le Point had become the first magazine in the world to bet durably on its readers’ collective intelligence, and that this was an innovation both in business model and in the relationship between a newsmagazine and its readers. He thanked Étienne Gernelle, Sébastien Le Fol, Renaud Grand-Clément and Thomas Le Mer.
The partnership was a later chapter in a longer story. Two decades earlier, NewsFutures, the company that became Hypermind, had supplied probabilities to US newspapers. Le Point’s variation was to recruit the forecasters from the publication’s own readership.
