I have no reason to believe that the views expressed above regarding changes to networking equipment and cables leading to sonic improvements are not true if that is the listener’s experience, and we’re all experienced listeners here. But ascribing those improvements to network operation are not true, because it cannot be so, and I suspect that that is where the divergence and misunderstanding may be coming in.
Changing networking gear will not change jitter, amplitude, timing etc. in a way that matters, it cannot - the fundamental operation of the streaming player, TCP, IP and Ethernet make that a fact. By contrast, the streamer and DAC absolutely can introduce (or not) all of those, and do so irrespective of how the bits got to the streamer (Ethernet, WiFi, PowerLAN, other). And assuming the player is appropriately implemented for the network type and scale in question - in this case/my experience, all evidence indicates that Roon is appropriately implemented - we can start to look to the streamer to DAC interface and all the other analogue inputs, RF, power etc. So assuming the network is working correctly - which is fairly easy and cheap to do with Ethernet - when you hear a difference between switches, it isn’t due to the network operation per se, but something else.
An example, for the sake of speculation, a cheap switch may add the right electrical harmonic to the domestic household power circuit that then leads to the analogue section of the DAC being negatively impacted - that I can envision happening. I could imagine such a scenario would (correctly) imply the switch was the problem, and it was (incorrectly) due to what it primarily does, switch packets. But in reality, that has nothing to do with networking, it is just power circuits impacting analogue signals, and I think most of us accept that that is real. A real example of that was a friend’s Pink Triangle turntable used to sound terrible some days and fantastic others - experimentation proved it was the CD player. When the CD player was on, it created sufficient noise over the power circuits to impact the turntable. Basically the digital noise from the CD player did something to the PSU on the turntable - we didn’t bother to investigate what, as electrical isolators provided the same solution as unplugging the CD player, solved. It wasn’t that the CD player sounded bad or good, it was the turntable being unable to reject the electrical noise - CD players were new at the time, who knew?
Similarly, I never measured the RF emitted by a switch, I’m sure metal cases (Cisco, other brand names designed mostly for racks in wiring closest and data centers) emit less than plastic cases designed for desktops. But lots of noisy RF may create problems with analogue signals that are in close proximity - such would be the case of a switch and DAC in the same room, on the same shelf, or worse, stacked.
I suspect, as you hear a difference in your system, and others cannot hear a difference in their system, it is specific to your system. But it isn’t the network, it cannot be the network - but it might be interaction between components, and that may be more expensive to fix than an audiophile switch, only you would be able to assess that.
In my setup, my switches are in the basement, next to the server running Roon Core. My players are all over the house directly connected to Ethernet drops. In the main listening room my Raspberry Pi/DAC and power amp are on dedicated electrical circuits (unique runs to the panel), and the Raspberry Pi and DAC are on a separate electrical mains cleaning system. Whether I use my PoE switch, my non-PoE switch or WiFi makes no difference to my system. And I further suspect, that an audiophile switch would also make no difference, because the switch is physically and electrically separated from the audio circuits by some margin.
