We can leave religion out of this.
One point I never made. Ethernet is a transport layer used to move IP traffic. One can transport IP over other fabrics. Fiddi, old "thicknet" or "thinnet" on top of ATM ( the world backbone) etc. One can use Ethernet fabric for other protocols. But for our discussion, we are talking IP. That is what comes out of our router and switches. TCP is guaranteed reliable. UDP is best effort. Shame on anyone using UDP.
Exactly, what we hear is the results of our brains processing sounds picked up by our ears. Sound is real. Hearing is our interpretation. It is subjective.
The point I am making is no one should trust their brain to be objectively accurate. It lies. Mine is just as biased as anyone. The difference may be that I understand and accept mine is biased where some believe theirs is not. That is one reason to use objective measurements as a starting point. ( not an end point) Seeing may be believing, but it does not mean it was real. I do love magic. Just I limit it to the good old "Blackstone" in your face skill kind.
If you ask a person "which sounds better", you have biased them to determine they are different. As our subconscious ego won't admit we are not good enough to determine a difference, we will make one up. We WILL believe it!
If we fall for something, and I assure we all have, again our internal bias does not want to admit we were fooled, so it will make up the difference and reinforce the decision. This is called "being human"
I see so many claimed "double blind" tests, but never with a sample size large enough to meet any scientific correlation. 5 out of 5 is just as likely to be a statistical guess fluke as valid. To boot, the very nature of a test is again biasing the listeners that there is a difference. Try running a test with no differences, but an obvious "click" to signify a change. You will get not only selections, but attributes ascribed to them. This has been done. Pretty funny actually. And yes, many people are insulted when they understand they were duped. Our ego hates that!
I hope you are aware the lowest confidence evidence in a court of law is witness identification. The law recognizes we can't trust our brains.
We need a lot more work on what clues our brain uses to decide something is musical. Not the same for everybody. I started studying psychoacoustics in college when I noticed music majors had speed controls on crappy turntables, but engineers had expensive speakers and fixed speed tables. Different experiences required different queues. Music majors hate when the pitch is off. I would not have a clue. We need a lot more work on things like our ability to extract information below the noise level. We need more work to quantify which measurements correlate to hearing preferences. Understand why a "perfect" measuring amp like a Benchmark leaves me cold, but my Mosfet sounds more dynamic. Why does my Vidar sit in the middle? I actually know some of the technical reasons and they are not included in classical SINAD type measurements. Some attempts to simulate "musicality" has been included in CHI-FI DACs. "Tube" mode with increased DSP generated even order harmonics. "BJT" mode with higher overall, but odd order closer to even. I heard one. I'll take the clean setting thank you. A clue, but not correct or complete.
Anyway, a layer one or two switch is not going to change the sound. There is no amplitude, timing, phase, or magical parameters involved. They are just not there. It just moves packets and if you read my link below, you will see most of the packet is not even your data. MTU size dependent. Again I challenge anyone, if they can come up with a parameter we do not know of, please please bring it forward. You-tube rants are not evidence. Neither is my hearing nor yours. If you do hear a difference, it is not a difference is sound buy because your brain made it up. Still, if you think it is better, then for you it is better and keep on keeping on. The good news is a "boutique" switch works just as well as my TP-link or Netgear $35 switch, so no harm done other than your bank account.
https://www.w3.org/People/Frystyk/thesis/TcpIp.html
Ethernet is not like PCM where the bits are just the signal and timing does matters. Ah, except on some newer DACs that buffer and re-clock the PCM. Classic case of bias from old very real faults that have been fixed.