How about my third link?

I'm getting a 502 error trying to access the original content. It doesn't appear to be a scientific study, but rather a testimony from a Facebook executive talking about how they disregarded user safety in the development of algorithms that increased engagement. That's not quite what I'm looking for, though. I'd like to see something examining the effects of those behaviors on the population.

I will say that the lengths the executive goes to to compare social media with tobacco degrade the quality of the argument in my opinion; science tends to ask the question and then seek the answer. Arguments like this seem to start with the answer (it's like Big Tobacco) and then construct the argument accordingly.