I also briefly worked in Student IT support junior year of my university and "Winmodem" sent similar chills down my spine. An idea that never should have happened!
You can boil a lot of tech changes down to either A: Let's take this problem that has been solved in hardware and move it to software! and B: Let's take this problem that has been implemented in software and bake it into hardware.
Somehow, A is always a train wreck, and B usually pushes the abstraction stack upward and moves the industry forward. Yet, we as an industry keep trying A and expecting good results.
Yeah, in the case of winmodems/softmodems, it was because A is cheaper. Or, at least, you could externalize the costs.
In our case, we technically did not support your hardware - you had to show up with a working modem. But in practice, if you want to retain your customers, you need to support their hardware. At one point we used to have CDs full of known good drivers for all of the common softmodems that we'd send out if we couldn't figure out a configuration workaround. Even then, I had a handful of discussions with folks where I basically told them that their thing wasn't going to work - they either needed a different modem, of which we'd recommend a few that we knew some stores carried, or they needed to find a way to cut down their line noise. I'm one of those types that takes it a little bit personally when I spend a bunch of time on something and still can't solve it, so that always sucked. Maybe you could say that wasn't strictly the modem's fault, but even the cheapest hardware modems had better tolerance for line noise.
I think a lot of the problem was how difficult it was for the average computer user to tell the difference between a real modem and a Winmodem. Some manufacturers deliberately failed to distinguish them in marketing, pretending they were both "modems". Retailers were in on the scam, too. The whole puddle got muddied to the point where a savvy consumer needed to keep a whitelist of "real modem" make and model numbers with them going to the store. You could usually tell by the price, though, as you say they were cheap (garbage).
This is unlocking memories for me. I think we used to tell folks something like "If it's under $50 and Walmart sells it, that's a winmodem" or something like that.
In theory, one of the selling points was that as standards changed, you would just upgrade your drivers/software and not buy a new modem. That probably made a lot of sense if you bought a USR Winmodem, but those $20 unbranded models were lucky to ever see an update. If you were lucky, you had a reference model and could use the OEM drivers which did occasionally get updated. But by the time these things came about, V.90/V.92 existed, and dialup standards were kind of frozen in that 56k-if-you-were-lucky state. There wasn't anything to upgrade to - you got DSL if you wanted more bandwidth over POTS lines, or you went to cable.
Also I could be completely full of shit on the above. These are memories from 16-18 year old me.
You can boil a lot of tech changes down to either A: Let's take this problem that has been solved in hardware and move it to software! and B: Let's take this problem that has been implemented in software and bake it into hardware.
Somehow, A is always a train wreck, and B usually pushes the abstraction stack upward and moves the industry forward. Yet, we as an industry keep trying A and expecting good results.