Once again I am reminded that "knowing" which accounts are fake is a knowable thing and yet social media companies don't mitigate them "because money" or "because DAU" Etc. When I was running operations at Blekko (a search engine) we were busily identifying all the bots that were attempting ad fraud or scouring the web for vulnerabilities or PII to update "people" data bases. And we just mitigated them[1], even though it meant that from a 'traffic' perspective we were blocking probably 3 - 4 million searches / day.

[1] My favorite mitigation was a machine that accepted the TCP connection from a bot address and just never responded after that (except to keep alives) I think the longest client we had hung that way had been waiting for over 3 months for a web page that never arrived. :-)