Hubs and switches are not an issue, nor are the various ethernet terminations standards. Same with speed. Users by an Ethernet cable, 99% of the time it'll be suitable for gigE (and thus suitable for 100Mb/sec down to 10Mb/sec).

Power over Ethernet: I haven't studied this one, so I'll assume it has some subtleties involved.

Hubs vs. switches: Hubs are awful (have used one), burn them with prejudice. Collisions bring everything down to 1 Mb/s. And hubs are disallowed in gigabit anyway, due to no CSMA/CD.

Category cables: USB has different cables too, and at least Ethernet didn't change the number of conductor wires. But the biggest difference is that Ethernet cables are often in walls or inaccessible places, whereas USB cables are user-replaceable.

100 Mb/s vs. 1 Gb/s: One problem I noticed is that a link might start at 1Gb/s, but after many weeks a momentary error (e.g. power blip) makes the link drop to 100Mb/s or even 10Mb/s and does not automatically recover to the higher speed without manually replugging. IIRC USB doesn't drop down to a lower speed in this manner.

Crossover cables: IIRC if you are using purely switches (no hubs), you will never use crossover cables. Other than that, pervasive MDI/MDI-X has silently solved the problem. Though I wish the standard didn't create this problem in the first place.

I do appreciate the fact that Ethernet treats nodes as peers. Whereas USB is based on a tree rooted at the master and the master must poll devices to ask them to transfer data. This tidy idea gets into trouble when, for example, a phone can be either a host (OTG) or a client. Also it is impossible to directly share USB devices between two hosts (e.g. a webcam accessible to two computers simultaneously).

Ethernet is definitely a standard with fewer issues than USB, but I'd say that is down to a narrower scope. That has allowed us to converge to essentially a working standard of 1Gbps over Cat5e.