I explained why you can't - Because you don't know what you don't know. An error margin or error rate or error probability or whatever error measure you choose to use, relies on your knowledge of the problem set. This is already known to most researchers. Establishing probabilities such as "life on planet X" or "existence of aliens" etc are problematic for the same reasons. Certainly these are not every-day decisions, but merely to illustrate the point. I suspect we might agree somewhere in the middle - maybe our disagreement is on what constitutes every day decisions?