edit: and it's not things I've heard, instead it is direct experiences, i.e. people stole my work, and things like that. As a graduate student to watch professors come to you with problem X, take what you've said (an actual solution) and publish a paper without attribute, that sort of thing; to report it and have nothing be done about it, et cetera, and on it goes, it's just instance after instance of such behavior, or the million ways in which they are careful to trick you into working on their problems without receiving attribute. One such trick for example, that again happened to me, is that after a conference talk I got into an e-mail discussion where I explained my approach; I was told that "they already have these results" (the trick here was to divulge less in the talk than what was currently known in order to be able to avoid "significant progress by another person" in the case another person does share new progress that they have already established, and hence not having to share attribution.) It turned out that our discussion was enough for them to go from n=3,4 to a general formula involving primes, because I pointed out a certain property they had not noticed. This is just a single example of the sorts of tricks, aside from total fraud, that happen, and one of the milder incidents I had happen to me.