Let me get a little pedantic here and qualify my position exactly (so that I’m not being misrepresented/misquoted).
At the physical layer, an Ethernet signal transmitted over a copper CAT6 cable, or over an SFP/fibre optic cable, is in fact an analog signal (representing the 1s and 0s of a digital bitstream), and as such, will experience the typical analog impairments and signal variations.
When looked at in isolation, one Ethernet cable, or an SFP, or a PHY circuit can be objectively better than another in terms of how well the end-point physical layer can retrieve the digital bitstream. So, the quality of the Ethernet Port Magnetics, or the SFP, or the PHY layer circuit, etc., is important (to some extent).
However , one cannot hear those differences because that physical layer (layer 1) is exclusively coupled to a full networking protocol stack; an Ethernet datalink layer (layer 2), a TCP/IP layer (layers 3-4), and the Session/Presentation layers above that (layers 5-6, e.g. UPnP/Roon-RAAT etc).
Those layers 2-6 were specifically designed to eliminate any physical layer differences impacting the source bitstream (through error-detection, error-correction, packet reassembly, and buffering etc).
So, even though there are variations in “quality” at the physical layer, the final bitstream that gets converted to music is an identical replica of the original source bitstream.
Uptone (and the other Audiophile Ethernet Switch makers) argue that those physical layer difference can and do make it all the way into the upper layers thus impacting the music that one hears, but there’s absolutely no science or facts supporting that position. And lots of facts, evidence, and Science on the contrary.
It’s ironic that these “Audiophile Ethernet Switch” makers totally depend on chips from the very manufacturers who explain exactly how and why the physical layer is independent from the upper layers, and yet Uptone and others are happy to tell their customer$ a different $tory. No need to wonder why.