[pcchristie]

I thought the reason prediction markets were useful was not that people with inside information participate in them which can provide an indicator to the rest of us, but that by providing a sense of consensus at scale we can more accurately predict things.

e.g. two sports teams participating in a fair match tomorrow, someone runs a book and after 200,000 punters bet, the odds are 90/10 in favour of Team A indicates that 90% of the time Team A is going to win that game.

This assumes perfect and fully available information with punters availing themselves of this info (or at least an equal split of passionate/casual/informed/wreckless or even slightly "inside" punters on each side).