It boils down to designs that are intended to minimize "noise" which is not something you would hear directly, like tape hiss, but the absence of which allows more musical detail to be heard. My first experience of this was with an expensive commercial switch from Cisco that was sold used for $40. One of many posters on the Naim Audio forum, where people report their experiences and speculate about the new "science", attributes this to:
"better clock circuitry, electro magnetic compliance and twisted pair transmission line termination compared to some other devices because of the environments they can be used in, and the high data transmission quality and reliability they offer. These Catalyst devices also implement IGMP snooping by default, which depending on home network and connected applications can improve streamer SQ as it removes irrelevant data and processing overhead from the streamer. Less processing and work tends to lead to better SQ."
Now I own an Uptone
etherREGEN which addresses additional sources of obscuring "hash" which are further explained in their infamous White Paper.