A scoring system using four judges rates AI research results on a shared list of solved and open problems, with 100 points set as the median solved result and no upper ceiling. Open results are scored by what they would be worth if achieved and checked, and every 10 net votes doubles or halves a result's points, up to 8 times in either direction. The method is presented as a way to measure the exponential pace of AI-driven research.
Four judges scored every result in one shuffled list, solved and open together, so the points compare directly. 100 points is the median solved result, and there is no ceiling. An open result is scored by what it would be worth if achieved and checked. The thin line on a bar runs from the lowest to the highest judge. Every 10 net votes doubles or halves a result's points, up to 8 times either way.
EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.