On 19 September 2023, International Day of Democracy, the European Citizen Action Service (ECAS) held a Democracy Lab in Brussels with Dreamocracy and Smarter Together. Its title was “How to harness Smart Crowdsourcing for stronger EU policies and democracy”. For a day, members of the EU policy community discussed how large numbers of citizens can contribute knowledge to public decisions. One of the stated aims was producing policy-relevant forecasts from many individual guesses.
According to the programme, Émile Servan-Schreiber, Hypermind’s CEO, led the 10:30 session titled “What can crowd forecasting do for policy makers?”. He would present Hypermind’s results and run a test of the platform with the policymakers in the room. Other sessions covered citizen science and digital participation. Speakers came from ECAS, Dreamocracy, The GovLab and Ashoka, and topics included citizen monitoring of air quality and environmental compliance.
Five points
The next morning, Servan-Schreiber summarised his message in five points:
- Governing is predicting.
- Collective intelligence improves forecasting.
- Crowd forecasting sheds light on noisy policy-relevant topics.
- Skilled citizen forecasters can augment domain experts.
- Crowd forecasting is accurate and explainable.
The first point is the premise. Every policy is a bet on what will happen if it is adopted, and on what will happen if it is not. The second and third explain why a crowd helps. When individual errors are independent, they partly cancel out. Topics that are noisy, with conflicting signals and contested expert views, are where aggregating many judgements adds the most.
The fourth point is about complementarity, not replacement. Citizens who have shown forecasting skill can work alongside specialists. Hypermind’s aggregation gives more weight to forecasters with a strong track record. The fifth answers a common worry about algorithmic advice. A forecast that policymakers cannot explain to colleagues or citizens is hard to act on. His point was that a crowd forecast does not have to be a black box.
What organisers took away
In its write-up, Smarter Together summarised his message in one line: to govern is to predict, and crowds outperform experts at forecasting. It also noted that very few EU institutions or member states used crowd forecasting at the time. The organisers argued that crowdsourcing should be used at every stage of the policy cycle, as a source of both better information and greater legitimacy, and that this required a change of mindset inside administrations.
Smarter Together also published a video of Servan-Schreiber’s talk. Its main example came from public health: a crowd-forecasting project on infectious-disease outbreaks with Johns Hopkins. In that study, according to the published chapter, most experts forecast poorly, while the crowd beat even the most accurate individual. The study is covered in a separate article.
