- Facebook will ban Holocaust denial hate speech and paid anti-vaccination ads by the end of 2020.
- Facebook has faced huge public scrutiny in the past for allowing rampant misinformation and hate content.
- Identifying and removing this content is a combination of powerful A.I. and user contributions.
Facebook has announced two major internal changes, “banning” both Holocaust denial content and paid anti-vaccination ads. (Crucially, anti-vaccination posts are still fine.) What do the changes mean, how will Facebook try to make them stick, and how likely is the ban to work? Let’s take a look.
Dive deeper. ➡ Read best-in-class tech reporting and get unlimited access to Pop Mech , starting now.
The New York Times reports that Facebook CEO Mark Zuckerberg’s decision about Holocaust denial content is a quick, dramatic reversal from his infamous 2018 comment that deniers embodied free speech he felt should be protected. Indeed, Facebook sings a different tune in its announcement, citing polling data about the rise of anti-Semitism, especially among young people who are arguably most influenced by the selective information peddled by Facebook.
Vaccines are more of a lukewarm noncommittal, with Facebook “[r]ejecting ads globally that discourage people from getting a vaccine,” the company says, before quickly walking it back by allowing “political” ads about laws that require vaccination for public school students, and not limiting anti-vax content more broadly. Vaccine misinformation expert David A. Broniatowski told ABC News that “the largest source” of vaccine misinformation will still slide through as political.
Yes, removing hate speech and dangerous public health misinformation is good as an ideal. But if Facebook doesn’t really do it, it’s worse than lip service. As a case study, perhaps, we can look at Facebook’s October 7 announcement about the path to the 2020 U.S. presidential election. In 2016, many critics blamed Facebook for choking good information and enabling misinformation. In the years since, Facebook says, “We worked on more than 200 elections around the globe since then, learning from each, and now have more than 35,000 people across the company working on safety and security issues.”
The rapid rise of wild COVID-19 misinformation and ignorance online pushed Facebook a step further. “Between March and September of this year,” the company says, it’s rejected over 2 million ads and put warnings on 150 million pieces of factually inaccurate content. In 2019 alone, Facebook says it removed 6.5 billion fake accounts.
How does Facebook identify malicious content to flag and remove? Well, that part is more secretive, but it almost certainly involves a big mix of both powerful artificial intelligence that scans content as well as user input. Facebook explains in last week's announcement:
“Since 2016, we’ve built an advanced system combining people and technology to review the billions of pieces of content that are posted to our platform every day. State-of-the-art AI systems flag content that may violate our policies, users report content to us they believe is questionable and our own teams review content.”
Very few hard numbers float around about the ratio of computer to human, but Facebook says 99.7 percent of the fake accounts it removed in 2019 were “proactively identified,” which usually means by a computer. (You may remember when Facebook started telling people with unusual names, like activist Dese’Rae L. Stage, that they were not real people because their names contained punctuation, nouns, or other wild criteria.)
The same way Google Ads can prompt something based on the right combination of keywords you type in, computers likely scan fresh Facebook content by the millions of words and identify items from a programmed list. As more users enter more and more content, the list grows both larger and more sophisticated. And when human users flag content, that can be fed into the software as well, helping to identify things that “seem like” hate speech but don’t outright contain slurs, for example.
If Facebook really delivers on the promise to ban Holocaust denial hate speech, that will also involve a potential game of whack-a-mole with new accounts to replace old ones. But since Facebook pushed for those “real,” noun- and punctuation-free names, there are at least fewer new fake accounts—if only slightly.
You Might Also Like