I am sure the Perl internals don't optimize a lot of things, specially in examples like Abigail showed us how to reduce the number of % operations by a careful use of user shortcuts (instead of depending in Perls shortcut for logical operators)
Well, actually, if I recall well, that's exactly the truth table for XOR:
Ie. Either A is true and B is false, or B is true and it's A which is false. This form has the advantage over the other one than in the first part is true (ie. A and not B) there's no need to compute the second one (2 calculations, 1 per proposition). The first proposed form, which is a description of 'exclusive or' (ie. either A or B but not both at once) forces the evaluation of both parts (ie. Once proposition A and again A and proposition B to test the second one, which makes it 3. In the case it's neither A or B it shortcircuits sooner, that's right, and if A is false, then propositions A and B must be tested twice in both cases: it's a matter of taste, I guess).
And obviously I agree with you that XOR is pretty much more specialized than, say AND or PLUS...
OTOH, I'd love for Perl to implement a fast NAND (which by itself creates a complete Boolean algebra). BTW, nand doesn't make expressions easier to read, but much faster as it would shorcircuit much sooner in most cases...