Within its statistical boundaries, the scientific method is absolutely trustworthy. Humans though? Not so much.