Comment by goldenManatee

I imagine other factors have to do with the technicalities of executing on this, and the user’s visibility and “viral-ity”. On the technical, you can automate the job and have awkward success sometimes if you don’t get a human to intervene and verify what the algo’s flag as potential violation tweets. Now considering the user, the user’s number of followers, whether they are public figures (so because elected, celebrity, activist, etc. reasons) or whether the tweet has gone viral (regardless of the user’s pre-existing popularity). These kinds of things influence because someone with visibility and audience making calls to violence or some other questionable act is distinct in how actionable others around the world are to react to such a figure making questionable statements. So it’s not a matter of policing everyone, because there’s a lot of nuance and challenges. And the answer isn’t to give up because that’s just cowardice in the face of a big social challenge. We’ve got to carefully experiment and wisely assess these cases.