As you know I’ve agreed with many of your posts in the past, that are often very informative, and I always view them with respect even when I disagree. In this case we’ll have to agree to disagree.
When highly experienced audiophiles such as DGarretson, Grannyring, and Bryoncunningham, all of whom have excellent systems as well as exceptionally high credibility in my book, report that changing from one inexpensive ethernet cable to another inexpensive ethernet cable results in significant sonic improvement, IMO a suggestion that they should get rid of one (and perhaps both) of the components that the cable is connecting amounts to an extreme position. And even more so given that there would seem to be no particular basis for confidence that whatever replacement is chosen would be any better in that respect, and might be worse in that or any number of other respects.
Also, regarding ...
... we should be completely justified to expect that good audio gear should be immune from small changes in Ethernet wiring.
... from a design standpoint I’m not sure that what you are saying we should be able to expect is as easily accomplished as you are envisioning. For example, in a post in this thread dated 3-28-2017 I suggested the following experiment to some of the others:
Tune a portable battery powered AM radio to an unused frequency, with the volume control set at a position that you would normally use. Bring it close to an unshielded ethernet cable on your LAN, while the cable is conducting traffic. You may be surprised at what you hear.
When I do that with the unshielded Cat5e cable I have on the LAN in my house, while the cable is **not** conducting any large amount of traffic, I hear increases in static from the radio when it is as far as 2 feet from the cable. Keep in mind that an AM radio is designed to just be sensitive to a narrow (~10 kHz) range of frequencies in the lower part of the RF region (nowhere close to frequencies corresponding to the bit rate of ethernet traffic, much less to the frequency components that constitute the risetimes and falltimes of the signals), and to have a sensitivity measured in microvolts. And for audio we’re dealing with microvolts as well, but without the benefit of the radio’s narrow band filtering. For digital audio if 2 volts corresponds to full scale the least significant bit of a 16 bit word corresponds to about 30 microvolts. And the least significant bit of a 24 bit word corresponds to about 0.1 microvolts! And perhaps more significantly there are jitter effects that will arise as a result of noise whenever D/A conversion is performed, of course. And this experiment just involves radiation of RFI through the air. Not through what would seem likely to potentially be much more significant unintended pathways for digital noise, such as grounds, other wiring, and parasitic capacitances within the components.
Regards,
-- Al