Hmmm... Ricoh's web site claimed otherwise. But USB would be fine, if you can find the right cable.

If I do "nc wobbegong.sharks 9100" and type random text, the printer warms up and say "Waiting". After I press Ctrl-C it prints the random text I wrote

If you do cat sample_file.ps | nc wobbegong.sharks 9100, does it print? If so, what does tcpdump/wireshark show being sent over the wire when printing with cups, and how does that differ from what works by just sending it to the printer with nc? My guess would be that something other than PostScript is being sent over the wire. There might be some job submission wrapper around the PostScript that your printer is objecting to, or it might just be in some altogether different format. Is your cups setup sending the file via JetDirect, or is it just LPD/IPP that aren't working?

• Can you print to it from Linux if you plug it in directly using a parallel cable? Even without cups/lpd/whatever, you can just cat some PostScript to /dev/lp0 and see if it works.
• When on the network, can you ping it?
• Can you connect to the relevant port (515, 631 or whatever else you're using)?
• If you're using lpd, I've come across many printers that are very sensitive to having the correct queue name. Check what the correct one should be for that printer.

Cups is very capable, but it's also (necessarily) very complex. I went to a lecture about it by Kurt Pfeifle many years ago, and it was only after that, and after seeing some of his architecture diagrams, that I really understood it. Of course, I've forgotten most of that in the intervening years, but if I can be of help, I will.

I've seen several claims of this from North America. Although my gut instinct was to call bullshit, I thought I'd double check with someone that knows. In my case, that's my ex, who has a degree in infectious diseases and a doctorate in molecular biology. She says she can't 100% rule it out, but the chances are very, very slim.