I've got a theory that for new platforms like this, the best way to get useful information from a prediction market is to ask questions that it would take me a very long time to answer, but another person can answer trivially (and make a profit doing so).

... So what kinds of questions fit those criteria?

Currently, any question I can articulate, I can answer (thx google). And if I can't, then I can't really make a prediction market about it...