
a lot of prediction market resolution fights are lost the moment the market is created. if the question is ambiguous, no oracle can save it. people argue. models disagree. someone loses money on a technicality. so the fix isn't only better resolution. it's refusing to create the ambiguous market in the first place. at kash, that starts in chat: describe the idea, and the AI council turns it into explicit outcomes, resolution criteria and sources for you to review before the market goes live. the check that decides how a market resolves should be the same check that decides whether it gets to exist.
