dexter, "Is that any way to run a life?" is not much of an argument for knowledge; obviously the same argument has been made many times by religious people for faith. The question is, do we have any reason to believe that evidence is reliable?
What evidence to I have that order could disappear tomorrow? Well, what evidence do I have that it won't, is a better question, and none, is the answer. Please see Hume's argument on induction. You say we should focus on what is "known and in front of me and reasonable to expect." The problem is, order is LESS reasonable to expect than disorder.
Why would it be reasonable to expect order? Because it's always been that way? Why does that mean that it will be that way tomorrow? Because it always has meant that? Well, why does that mean it will continue to mean it tomorrow?
Many have looked for a way out of that regress, but none exists.
You may ask why I say it's less reasonable to expect order than disorder. Simple: there are many, many (possible) universes which act ordered up to tomorrow and then misbehave in all kinds of crazy ways. Infinitely many, actually. There is only one that keeps acting ordered up through tomorrow. (This is not quite true, of course, but it works modulo some unimportant details).
Given that everything we've ever seen is consistent with ANY of these possible universes (since they were all, by definition, ordered up till tomorrow), nothing we've seen can act as evidence for preferring one to the other. So why do you believe in that one? How do you count that belief as knowledge?
As for managing by exception -- sure, if you ASSUME the universe will be regular tomorrow, then managing by exception is a poor idea. But it does remain an assumption, an a completely unjustified one.