File under "it's not enough to be smart; you have to actually know things."

Reason is an excellent tool for generating hypotheses. But it is in the world of hard, gritty practicality that honest folks test their favorite ideas, modify them under the helpful heat of criticism, carve away errors, and join others in developing systems that work.

David Brin, The Transparent Society

No matter what your hypothesis says, no matter how elegant your theory, no matter how reasonable, common-sense, and obviously true it may seem, real-world observations and measurements trump everything else. If the observations do not match the theory, if the measurable predictions of the hypothesis are not borne out, then it is not the observations and measurements that are wrong.