I would like to add my thoughts on this. Digital transmission is on/off. unless you are using a sub par cable, it will work as intended. Else this post woul have subtle changes in type. A part of an "A" might be missing, or a comma would appear to be a period even though I typed a comma.
That said, I bet many USB cables are actually just sub par standard. This will lead to miniscule error correction being applied by the two devices talking. You will still get all digital information through the cable. Now, depending on the way the two devices communicate to each other what info needs to be sent again, and how fast this eec is being applied might determine how smooth the "idle" of the dac is.
Car analogies are fun. So a working and to spec usb will result in a smooth idle, where as one with very sub par copper might rsult in a loping idle? One doesn't know how this affects the dac performance in terms of timing.
The company I work for designs high speed military computer based on Intel and PowerPC chips. They say that every time signals are sent per voltage through copper or even traces on a pcb, the signal will degrade. They explained that one cannot place a cpu more than 12" inches away from a bridge or high speed pci express slot without extra boosters to reboost the signal. This introduces latency into the signal. Its all very miniscule, but it can add up in a path from input device to cpu to game to cpu to audio engine to graphics card to cpu to monitor jadda jadda.
He also said that USB 2.0 and below is not very suspectible to cable problems, because speeds are quite slow. Even USB3 or HDMI isn't affected very much by this. Cables like Thunderbolt cables need to be "pretty good" built so they work. I asked about my car analogy about ecc and he smiled and said: "Then it must be a very very crappy and primitive dac from 1980" but also said: "Well, oscillations and clock problems in standard PC spread spectrum clocking methods are very complicated. Basically the entire system bus varies in speed constantly. This method was introduced so EMI measurements are "better" and cheap PCB and system components and methods can be used to pass emi inspections and standards. Some bios can turn off ssc and this makes for a "faster" system." Then he started ranting about Intel and SSC but I didn't understand a word of it and started this post. He is still shaking his head over there
The same schematics and layout guy soldered nichi caps and other totl components into his headfone amp, while uprading the op amps.
Conclusion: Everything makes a difference in some way. But it is never as simple as "its digital either it works or it doesnt". I mean it does, but it doesn't show the complete picture.
Personally I think usb cables, if properly built will "just work". The sonic differences will not be there if its a good cable. YMMV, and I hope it does so we have enjoyment in these threads to discuss this stuff. As long as you listen to the music once in a while instead of to the system, the world will continue to revolve.
edit: ok i didn't look at the date ofnthe first post.