This is more about logic than language, though the two are entwined.
Felix Salmon, the financial writer for Reuters, is a treasure so I quote him only to suggest that even writers of his talent and accomplishment often make this baffling language (and logical) error. Here’s the quote (he’s talking about NYC skyscrapers):
In general, if the public is asked whether they want any new skyscraper, the answer will always be “no” — even as they love the iconic tall buildings they’ve lived with for years. (There was a general consensus that something should restore the skyline after the World Trade Center was demolished, but that’s the exception that proves the rule.)
What does that mean: “the exception that proves the rule?” In this situation – and countless others – it is used to suggest something like the following: a) I am positing that Rule A exists; b) I am also identifying Exception X; c) I say “it’s the exception that proves the rule”; d) I now proceed as if Rule A is still a rule, even though I have just shown that it is not a rule.
If I feel particularly cocky that day, I might even imply that the rule is somehow stronger, what with me having found this exception.
Which is bonkers.
Exceptions are evidence that a purported rule is, in fact, not a rule at all. All Exception X proves is that we shouldn’t be calling Rule A a rule.
Is this extremely common mistake just the product of confusion about the multiple uses of the word “proves?” Of course there is the common definition of prove – something along the lines of “shows to be true.” But there is the older sense of the word, which is simply “tests.” For those in the Washington DC area, think Aberdeen Proving Grounds where weapons were subjected to a testing process to see whether or not they worked.
It is logical that this was the original meaning of the phrase – that a seeming exception tests a rule. If the exception can be shown to be what it appears to be – a true exception – then we know that the rule is not a rule. Isn’t that what this phrase must have initially meant before people started misusing it to suggest the opposite of what the phrase literally says?
Why do even the smartest among us persist in using not just a phrase that makes no sense, but the illogical leap that follows it – that somehow the rule is validated by us having marshaled evidence that it should not be regarded as a rule at all?