I'm coorganzing OPTIC, a intercollegiate forecasting tournament. We are currently planning to run 3-4ish tournaments in the fall, and to have about 20 questions per tournament. The following are some of the criteria for what makes a good question:
• Unambiguous resolution
• Participants are likely to make different forecasts
• Doing well on the question is a strong signal (so that people care who won the tournament)
• Resolves within a few months (so participants don't have to wait too long)
The tournaments will most likely be run on Metaculus, and so (for example) continuous questions are possible, but not anything that Metaculus cannot technically support.
I'll do my best to keep track of who inspired what questions, and when in doubt I'll lean strongly towards giving out bounties. I may give out additional bounties if I particularly like an idea.