Forecasting is Way Overrated, and We Should Stop Funding It

In a provocative post on LessWrong, a former #1 forecaster on Manifold (who remains #8 despite quitting two years ago) and profitable Polymarket bettor argues that the Effective Altruism (EA) community should stop funding forecasting and prediction markets. The author claims that nearly $100M in EA grants has gone to platforms like Metaculus, the Forecasting Research Institute, and Manifold, as well as tooling like the Squiggle programming language, without producing concrete decision-making gains. For-profit markets like Kalshi and Polymarket see billions in monthly volume, but the author dismisses this as dominated by trivial bets on sports and crypto, not high-impact questions.

The author critiques forecasting as a "solution seeking a problem," comparing it to cryptocurrency hype. They argue that instead of starting with a tool and looking for applications, problem-solvers should begin with the problem and then select appropriate tools. While acknowledging that forecasting could theoretically improve decisions, the author contends that after years of substantial investment, the lack of measurable impact means funding should be redirected to more proven interventions. The post has sparked debate, with some commenters defending forecasting's role in calibration and long-term thinking, but the author remains firm that the opportunity cost is too high.