There is an alternate interpretation, that the existence of an explicit exception proves (confirms) the existence of a rule to which an exception can be made.
So the (existence of an) exception proves (the existence of) the rule.
I agree in that the meaning of prove in that context is "put it to the test" but for me it doesn't go as far as finding the rule incorrect, because it's a general rule, not an absolute rule. A lot more exceptions would be necessary to make the rule incorrect for the general case.
I always thought that the "rule" referred to is that "all rules have exceptions" (R1). i.e. there's a rule (in this case "only men with big beards can tame a unix system" (R2)) which, however, has an exception ("girls in dinosaur themeparks can also do so" (E1)). Therefore, the R1 is, once again, shown to be true.
Hmmm... self-referential vibes coming here. It might be that the only exception to the R1 is itself, but then... etc.
The phrase bothers me because it's often used to set up a cousin to the no true scotsman fallacy. If you can't find an exception, then it proves the nay-sayer right. If you CAN find an exception.... it still proves the naysayer right?!?
I wouldn't use the phrase outside of silly internet jokes about 90s popcorn flicks.
The implication is that the exceptions are outliers.
If your rule is that carpenters are usually men and someone goes into a carpentry conference with 1000 carpenters and points out that 50 of them are women, you get to say "exception that proves the rule" and be right.
If your rule is that carpenters are usually women and someone goes into a carpentry conference with 1000 carpenters and points out that 950 of them are men, that's not the exception that proves the rule because 95% of the target population doesn't qualify as an exception.
