4 ms·

Did these 12 account for all or the bulk of 18 million pieces of information, and the bulk of anti-covid-vax postings? That is the lumping together I'm talking

• notahacker 5y ago
  One of the CCDH's criticisms of Facebook's internal analysis is that apparently they counted shares as discrete "pieces of information", never mind paraphrases. The "dirty dozen" highlighted producing 65% of the content CCDH class as antivax have a combined 59 million social followers and are extremely prolific. So if nothing else, we can be sure that [by Facebook's metrics] there's a lot more than 18m obvious "pieces of misinformation" by any reasonable standard posted to Facebook over that timeline. Of course it's not totally impossible Facebook's review process is so bad that they focused all their efforts on suppressing mainstream news stories about blood clots and people wondering whether they should still take it if they'd had adverse reactions to vaccines in the past. Certainly nobody would argue Facebook is perfect, but with the quantity of actual ludicrous antivax content out there clearly being greater than what they removed, it would have to be extraordinarily bad to have primarily targeted standard concerns. And that in itself would be worthy of discussion (and again CCDH specifically question whether Facebook's algorithms are adequate to distinguish between individual and organized vaccine hesitancy..)