There are factual inaccuracies in this.
It’s not mile off but.
The big difference between USB and the other inputs on a DAC, is that there is no clock in a USB signal. The USB circuit in the DAC provides one.
All the other digital inputs on a DAC do provide a clock from the source, with I2S being considered the best, because the clock signal is separate from the actual data, on everything but I2S, the Input has to extract the clock from the data.
Even on AES/SPDif etc, DAC’s will commonly either correct the clock with a PLL or even regenerate the clock, some usually higher end DAC’s will allow you to disable this correction.
The quality of a USB signal on the data lines is unlikely to affect the sound of a DAC, it is almost certainly decoded perfectly by the DAC and clocked accordingly. It can have errors since there is no error correction, but in practice they will be extremely rare.
The USB signal entering the DAC on better DAC’s will be galvanically isolated from the signal to the DAC. This isolation is likely not perfect, and the the thing reading the USB signal itself and converting it is probably putting out it’s own noise.
What USB to USB DDC’s can do is add an additional layer of galvanic isolation.
One that converts Network to USB can go further by simply introducing less noise that needs to be isolated than your PC does.
A DDC that converts USB or Network input to say SPDIF or AES or I2S can go a step further because it generates the clock and if it can do a better job than the USB device in the DAC it’s a win too.
I also suspect that most Network transports are more complex than just a DDC, i.e. they apply their own filters to the input signal, I know for example that the Aries transports skipped some of the processing when using Roon at one point (since fixed) and RAAT was considered inferior on the native Lightning protocol and even DLNA.