They followed in Facebook's footsteps and chased after meaningless engagement metrics, thus driving content-quality down. I don't know if they ever noticed - or cared. As long as the metrics look good these days you know, no one will ask questions.
There was a series of dubious "upgrades" to the feed algorithm which weighted popular content on the platform more than your own preferences or those of the people you followed. Overall, the content they pushed on you was to appeal to the lowest common denominator among your category of users (e.g "following Programming languages") which of course meant that it was absolute trash.
The great takeaway here is that one should be wary of playing god with feed algos. They create a web of complex feedback loops that no one actually understands. And building a product you can't understand will ineluctably crash down on you.
Sure, that might improve engagement metrics momentarily. I'm certain that at the same time, it devastate the lifetime user value, as people will inevitably get sick of being fed off detritus.
My thesis is that lots of the experts got driven away by the Linkedin-ish, clickbaity, and self-glorifying answers being aggressively promoted by the feed algo.
It's a real shame because circa 2013/2014 I found tons of excellent content there and was happy to engage a lot.