you guys are so predictable.
Obviously, the differences are measurable at the analog output of the DAC; that's by analyzing the audio signal at the analog output that jitter is most commonly measured today (as it doesn't require to open and hack the DAC unit under test).
Paul Miller's tests indeed don't establish audibility by themselves (did anyone say that, certainly not I ?). They deal with whether or not USB cables have various levels of performances and up to what point. It's however enough to prove wrong all the guys who seem happy to just repeat ad nauseum "it's all digital, cables can't possibly make any difference, blah blah".
However, by establishing the actual differences in between cables, those tests either allow to dismiss the audibility of those variations in technical performance or not, in relation with jitter audibility baselines established best through DBT.
Which brings me back to my original statement: usb cables do make differences, but those differences are likely too small to be audible. I'm just waiting to get the figures from the article to be confirmed in my opinion.