Sadly, I don’t think the really, really, really annoying stuff will ever die though. It costs too little to throw out there digitally and some people apparently just cannot help themselves. Even if it is a terrible deal that only reinforces people marketing in the most intrusive way possible. The people who build platforms like Twitter, Facebook, Reddit, etc have it in their heads that their algorithm is the answer to all of that, and if it is still happening despite their algorithm then there the problem is too hard to actually solve and so they throw up their hands and blame someone else.
This goes greatly against what’s described in the article: a team/teams of data scientists/engineers datamining the platform in search of patterns of mass manipulation and malicious activity. Of course for that, they build statistical models, use the appropriate algorithms, etc. Like what else do you expect them to do? Again, if you’re implying their ML-based sorting is the only thing they’re basing their strategies around, you’re objectively wrong – at the very least in the case of Facebook (see the article).