> The findings challenge the idea that prediction markets work because of crowds. They appear to work because of who is informed.

That sounds about right. A small number of people with inside info are gaming the system itself and everybody else is just gambling.

People who work in mainstream financial markets (e.g. stocks) believe pretty much the same thing: the vast majority of people are uninformed and contribute Brownian motion whereas just a few people know something and they move markets.

https://ideas.repec.org/p/pra/mprapa/24487.html

https://arxiv.org/abs/2501.03658

Notably market makers can make money off the noise injected by uninformed traders but would not trade with informed traders if they could help it. (Look how “sharps” are kicked out of online sport books.)

That tracks. People as a whole are terrible at predicting. There are a few people who can put aside preconceptions and see a situation clearly.

I bet there are fields where any given super predictor consistently fails.