The PoW algorithm needs to effectively pick winners more or less at random, with a consistent number of winners (and gaps between), and the results be self-evid

The PoW algorithm needs to effectively pick winners more or less at random, with a consistent number of winners (and gaps between), and the results be self-evidently correct. There are few problems which fit those criteria and are useful...