Proof should determine confidence, not confidence determine what counts as proof.

There is a direction things are supposed to run: the evidence arrives and your confidence adjusts to match. What often happens instead is that the confidence is fixed and the standard of proof moves to protect it, so that a claim you like is accepted on a fragment while a claim you dislike would require an impossible amount. The standard becomes a servant of the conclusion.

Watch for the moment you change what would satisfy you depending on the answer. Set the bar before you know which way something points, and then hold to it whichever way it falls.