A more relevant post by Scott is The Control Group is Out of Control [1]. The starting off point is that parapsychology has high-quality studies with positive results, but we're pretty sure parapsychology is wrong, and this bodes poorly for "high-quality" studies in other fields. But then it just keeps going and going... by the end I always feel like lowering the acceptable p-value ceiling by a factor of 1000, and also like that would be a joke laughed off by the nature of the problem.
1: http://slatestarcodex.com/2014/04/28/the-control-group-is-ou...