Running dOrg taught me how inefficient collective decisions can be: in an organization where nobody can simply decide, every choice turns into a proposal, a discussion and a vote, and most members are guessing at the outcome as much as anyone. Prediction markets are the most elegant answer I've found to that problem, a way to pool what people actually know by letting them profit from being right. The version I care most about is the decision market, where an organization bets on the consequences of each option before it chooses.
Prediction Prophet
In 2024 I led the Prediction Prophet project with Gnosis, an onchain agent made to trade on their prediction market, Omen. It forecasts the way an analyst would: it researches the question across the web, writes up the evidence it finds, and only then commits to a probability and a bet.
GitHubGnosis ran benchmarks that set prediction agents against each other on live markets, and Prophet stayed at the top of the leaderboard for years.
Mechanisms for prediction markets
At the Mechanism Institute we wrote up how onchain prediction markets work, taking Polymarket apart as the case, where their designs fall short on liquidity, and what they could do beyond betting.
Mechanism Institute