The bit is still a bit, but if the bit is generated at the beginning of a cycle it might have more voltage than at the end of the cycle and this difference could be discernable in the electromagnetic environment around the processor, which might effect the actual movement of the electro-acoustical element after the decoding by the DAC!
I would bet money on a blind taste test on a well pulled shot of espresso vs just dumping some water on some ground beans would show a substantially bigger difference than these people using an audiophile grade network switch or a different version of memcpy. The two aren't even in the same league and it requires some real magical thinking to come to any other conclusion.
It's at least as likely as the coffee scenario.