Even Facebook admits it’s having a hard time tackling hate speech.
The world’s most popular social network said Tuesday that it removes about 66,000 posts a week — around 288,000 monthly — on what it considers hateful rhetoric. And, Facebook said, it’s working on doing a much better job.
Facebook defines hate speech as attacks on people based on their race, sexual orientation and other “protected characteristics,” saying it depends heavily on its nowto report any hate speech they encounter. Workers review the posts and decide whether to take them down.
The topic was the focus of the second part of Facebook’s Hard Questions, a series on how the company is handling tough issues, including . This latest installment of apparent transparency arrives as .
As far as hate speech goes, Facebook admits it isn’t perfect enforcing its policy, Richard Allan, the company’s vice president of public policy for Europe, the Middle East and Asia, wrote in a lengthy blog post.
Part of the problem, he said, is determining what exactly is hate speech. He said while there are some close calls, Facebook “too often” gets it wrong.
For example, Allan said a statement like “burn flags, not fags,” could be seen either as a symbol of protest, an anti-gay slur or a means by a group to reclaim the slur as their own.
“People who live in the same country — or next door — often have different levels of tolerance for speech about protected characteristics,” he said. “But sometimes, there isn’t a clear consensus — because the words themselves are ambiguous, the intent behind them is unknown or the context around them is unclear.”
Allan cited an instance involving Shaun King, the renowned New York-based activist and writer who posted hate mail that included slurs. “We took down Mr. King’s post in error — not recognizing at first that it was shared to condemn the attack,” he said.
Facebook is experimenting with technology to “filter the most obviously toxic language,” Allan said, noting the company is still “a long way” from being able to rely on AI and machine learning to tackle hate speech.
The company currently has about 4,500 workers reviewing posts and plans to hire 3,000 more in the next year, Allan added.