Prediction Markets
A prediction market is a market where participants trade contracts that pay out based on whether a specified future event occurs. The market price of a contract represents the aggregate probability estimate of the crowd — the collective belief, backed by real stakes, about what will happen.
Prediction markets consistently outperform conventional polling and expert forecasts on factual questions because they aggregate dispersed private information and reward accuracy: participants who know something the market doesn't can profit by betting accordingly, which moves the price toward the true probability.
Application to democracy
The democratic application of prediction markets goes beyond forecasting election outcomes. The key idea, developed by Prediki and others, is that asking people to predict policy outcomes engages a different cognitive mode than asking what they prefer or believe:
- Predictions require committing to a specific claim about what will happen
- Stakes (even small or symbolic) activate careful reasoning rather than expressive or tribal responses
- Accuracy over time is measurable, enabling reputation systems that weight better-informed views more heavily
This makes prediction markets a potential tool for surfacing considered collective intelligence rather than raw preference — relevant to the deliberative democracy project.
In practice
- Prediki — built a platform combining prediction markets with argument capture and sentiment analysis, explicitly framed as a democratic tool. Presented at the 2017 DOD event.
- Futarchy — a governance model proposed by Robin Hanson in which policy is determined by prediction markets: "vote on values, bet on beliefs." Society decides what outcomes it wants (e.g. GDP per capita); prediction markets determine which policies are most likely to produce those outcomes.
- Iowa Electronic Markets — long-running academic prediction market for US elections, consistently competitive with major polls.
Limitations
- Thin markets can be manipulated or may not aggregate enough information to be useful
- Prediction markets work best on factual, resolvable questions — harder to apply to value questions ("which policy is better?") than to factual ones ("will this policy reduce unemployment?")
- Futarchy has attracted philosophical objections: reducing governance to welfare metrics may exclude important values that resist quantification