Why wasn’t Kermit used by BBSes? Even at the time I remember having read about it in computer magazines, but I got my first modem in ‘82 or so and I don’t recall ever seeing it in the wild. My first actual contact with it was uploading firmware I wrote to an embedded controller. But the thing it was actually designed for? Not even once.

So… why? We didn’t collectively settle on YMODEM and ZMODEM until many years later. There were lots of early protocols to choose from back then, too. Punter, anyone? No Kermit, though.

Kermit had licensing issues in those days. And many bbs were focused on one specific system so many of the auto translate and nice compatibility features of Kermit weren’t that important.

And finally Kermit didn’t have great defaults especially compared to ymodem and zmodem. So getting good performance out of Kermit was much harder than the others.

All of that added up to much lower adoption outside of the mainframe world.

The very oldest (early 80s) Kermit implementations were limited to 96-byte packets. Kermit gained support for 4K and 9K jumbo frames, sliding windows, etc., and was generally faster than ZModem -- when there was a modern implementation. But a lot of BBSs (and also client-side communication software) did their own from-scratch Kermit implementation, which was tiny packets and no sliding windows, so it was slow. There's some discussion of it here: https://www.kermitproject.org/misconceptions.html#notslow

GOO. I remember Punter well.