The endless gamification, nagging, and destruction of useful features would have driven me from the site. But because I'm a technically inclined, instead they drove me to sideload an "outdated" version of the app. Amusingly it still uses the same API, so you get all the core features and then some (turns out premium or not is a client side check lol) without any of the increasingly intolerable junk of the new app.
But what I have also noted is that the activity of users seems to have fallen off a cliff, which I'd tend to assume is because the number of users has fallen off a cliff. I went from a high activity user to a near zero activity user (Memrize is a much better primary tool for learning languages now IMO) yet my place in the leagues remains nearly unchanged. So either everybody also really started "engaging" with the site much less, or they simply left.
If anything Duolingo seems to be a warning about the dangers of pursuing the maximization of metrics at any cost. Work on making a better product and utilize metrics to see how you're doing. Work on mindlessly maximizing those metrics instead and you risk not only destroying your product, but also destroying your ability to measure the damage you're causing.