I called prediction markets being unreliable until they’ve been around for longer a while ago (I think somewhere on Paul‘s wall, with some replies by EY, if anyone wants to try digging it up). After they failed to predict brexit badly, I thought about it and concluded that they don’t start being reliable until the people who just go with their gut lose enough times that they drop out, or until the rich+self-trusting people realize how irrational the markets are and are able/willing to overwhelm the small betters (which never happens if there is a betting cap).
My claim is not “there is a better source on an individual level”, but “everyone seems to be taking prediction markets as their personal prediction, so our errors are all correlated, reducing the community’s overall truth-seeking ability to the level of a fairly-bad-by-absolute-standards source”.
Yes, you do better by just trusting the markets. But if everyone in a community tried to make their own models and felt the affordance to post independent predictions then we’d aggregate many more datapoints and we’d be collectively much better at figuring out where the future is headed than if everyone had total epistemic humility and just pointed to the markets as their credence.
I dislike the feeling that it’s wholesome and good for everyone to signal their agreement with prediction markets (the implication of regularly posting them), and kinda non-virtuous to make claims that your personal models contradict the prediction markets (met with “go bet”, which is inconvenient, and maybe the belief is based on only a few datapoints rather than deep enough understanding that you’d want to put money on it). Markets are a good source, but amplifying that signal beyond a certain point weakens our collective predictive ability, especially while the markets are fresh and constrained.
Deferring to the market is fine if you have nothing else to offer, but not deferring to the market means you’re providing information to the community, and should be celebrated much more than we seem to.