alt Hacker NewsSurprisingly many things seem to be spoiled by the "too many of the people being studied were actually other researchers trying to study the same thing" (or even more commonly, students being taught about the thing).
I suspect the ability to post/apply for jobs with AI "to study ___" has played a part in getting us into our present predicament. If only one researcher did it, the results would be negligible, but if a significant number try it, all those negligibles add up.
Replies
I don't know about the others referenced in the article or what else you might be referring to, but that wasn't the case with When Prophesy Fails or the Stanford Prison Experiment though. That was more or less fraud. The researchers put their thumb on the scale significantly.
There's a common joke that defines psychology as the scientific discipline studying the undergraduate psychology students. That is obviously due to the fact that a lot of research subjects are found where it's easiest to find them - right on campus, and a lot of people who have time and desire to participate in studies (instead of, you know, working) are the students themselves.
Interestingly banning a far right party in Germany failed at the constitutional court a few years back with one of the arguments being too many party members were actually paid informants of the authorities and thus it wasn't reliably possible to distinguish what the party did on its own vs what the authorities caused. I think many people concluded that the party finances were mostly kept stable by these informants, so the easiest way to get rid of them would be to drastically reduce that.