Almost every discussion here on HN about those "deaths and disappearances of scientists linked to secret military research" has been about how, when you actually dig into the data, they have been very statistically likely.