Indeed there's no clear answer as to what content is "harmful", which is why I'm not advocating for censorship but merely to let the user be in control of what they see.

Currently, the algorithm will prioritize content that produces the most "engagement", including even content that you otherwise have no link with and don't follow the user who posted it. Divisive, outrageous, hateful or blatantly fake content will typically generate tremendous amounts of engagement (as people start arguing over it) as opposed to mundane content such as updates from your friends.

There's also no way to have "civil discourse" when the algorithm prioritizes the hottest takes as opposed to more reasonable arguments.

There's another factor that I haven't see mentioned: I have no interest in most sections of the news. And even if I select only some kind of news, say "computers" I'm not very interested in consumer's hardware or commercial programs including apps or games, just programming, security and little more.

But social media live on ads. They tend to promote contents that cater to wide audiences.

If I'm only interested in a few niches, social media are a big waste of time. The contents and the discussions around them.