Facebook has published official numbers on its enforcement efforts for the first time in its history. Providing users with a look into just how much work goes into keeping the social media platform free from those looking to abuse their freedom of speech.
The enforcement team breaks down infringing content into six categories: graphic violence, adult nudity and sexual activity, terrorist propaganda, hate speech, spam, and fake accounts. It’s detection algorithm is better at detecting some categories than others; and Facebook admits that hate speech still represents a detection problem.
Of these, the company says that it has shut down some 583 million fake accounts within the firsts quarter of 2018; most of which within minutes of being created. This number apparently doesn’t include millions of accounts that are blocked from being registered in the first place. Although the announcement doesn’t quite deal with a hard number of these attempts.
Similarly, the company has removed 837 million pieces of spam. This is trumpeted as a victory, as Facebook says that close to 100 percent of these incidents were detected by its own algorithms before users noticed anything happened.
On the other hand, Facebook only took down 2.5 million posts containing hate speech. The lower number is apparently due to the team needing to moderate this content manually, as the algorithm cannot tell the difference between actual declarations of hate and people relating a story that happened to them. In this case, only 38 percent of the posts deleted were detected automatically.
Publishing official numbers is part of the company’s efforts to become more transparent ahead of global criticism. There is an additional post about how Facebook determines the impact of posts and which need to be urgently moderated. It’s an interesting look into the thought process that goes on behind the scenes, but also highlights just how much guesswork actually goes into this kind of work.