This is a nice idea in theory, but in practice you have to really trust the judges. Suppose that 99.9% of people thought that X is going to happen, but it didn't. If the judges are a representative sample, then 99.9% of them were wrong and are about to lose money. Unless they cheat.