> a worse version of Anki but for the web

This is actually a remarkably common failure pattern for a lot of language learning apps. Devs see Anki and think "I'm gonna do it better! I'm gonna build Anki, but for a specific language and make it a web app." ... I've lost count how many of these I've seen over the years!

> It seems like the ETA before enshittification begins is about 2~3 years

It's funny, I actually learned the term "enshittification" specifically from friends of mine who were Memrise users. It's honestly a textbook example of the phenomena.

Negativity aside though, I'm actually pretty optimistic about this space despite all that I've seen so far. I think that there's genuinely room to build great language learning software that people will really benefit from. I'm just really pessimistic about most of the people working in the space. Without trying to exagerrate, I'd estimate that probably less than 10% of people working in the language learning industry are actual language learners (at best, they might've learned English as a kid). When you're not actively, seriously learning a language, you become numb to the problems of people who actually care about becoming fluent and just end up building tinder-esque games to addict people with.