Prediction markets?

Prediction markets are a hot topic among economist (good choice, Katka).From "what is it?" to "everybody knows it works" in a couple of years is impressive. The latest prediction market is Predictify and you can, of course, try for free how good you are in predicting things. It's a cool idea, but there are couple of reason why I think (predict;-) it fail.
First, predicting is free. The only thing you actually risk is your reputation, which is not much. Of course, there is a liquidity problem (you need to attract a lot of people, hopefully plenty of those who have an idea what they are actually answering), so this is probably necessary and it is not the main problem.
Other problem is the lack of interesting questions. After a couple of weeks from the take-off, the trivial ones (who will be the next president of the USA) were already asked and often are buried deep in the "trash" of question like "What will be Rotten Tomato Scores of movie X", where X goes through all premiers plan for this fall. Since it is hard to distinguish good (=interesting) and bad (boring) questions, and since there is much more of the latter ones, the relevant question are unlikely to get large audience. It does not help much that by asking (for free) questions you get points for your reputation. It is good for attracting new questions, bad for attracting good ones.
Also, the number of people who can answer is limited to 100 for free question and there is a fixed price for answer in paid questions ($1 per answer). Regardless where these fees end up, you have a bunch of people who play a lottery with your money. They get the tickets for free and these tickets cost YOU dollar a piece.
I don't know what is the best way to fix these problems (or if there is any), so I'm not suggesting any. It is certainly a nice attempt, unfortunately I cannot believe that it will be successful.