Because it is not just a nice idea in theory, it is factual in practice. With asynchronous USB, there is absolutely no timing data incorporated in the data stream. None, zip, nada, 0. It is not a matter of resolving arrival timing errors .... there are none to resolve. This statement shows gross ignorance about how the stuff he is supposed to be an expert on actually works.
Then comes the argument that the DAC will resolve all data
arrival-timing errors by first buffering them and then re-clocking them
before sending them onto the decoder chip. That’s a nice idea in theory
but experience tells us otherwise.
You asked for one. I jumped through your hoop. There are more errors in this document. But this is so gross an error to make all others pointless to point out.