Earlier today, ProPublica published Facebook slides bluntly detailing the company’s hate speech policy in the form of a quiz. Facebook bans attacks on specific protected classes, which include people of a given race, gender, or sexual orientation. But it’s more lenient on statements about subsets of these categories, and about “quasi-protected” categories like refugees.
In practice, the results look more like a parody of anti-hate speech rules. One slide, for example, asked whether “black children” or “white men” were a protected subset of people on Facebook. The answer? White men, because men and white people were both protected classes, while “children” are not. Similarly, calling to hunt down and kill “radicalized” Muslims is seemingly all right, but saying “all white people are racist” is ban-worthy.
As law professor Danielle Citron tells ProPublica, these rules have a kind of “color-blindness” that sees no difference between attacking an oppressed group or an oppressive one. Groups are defined in the most legalistic way possible, and hate speech isn’t based on what’s actually likely to harm people or even seem offensive, but on what’s easy to define in a manual. It looks bad. But can hate speech rules on a platform like Facebook ever look good?
Facebook isn’t bound by the First Amendment, and it’s free to limit speech as much or little as it likes. (Right now, it seems to be following the bare-minimum rules that will keep it operational in countries with hate speech laws.) But what role do we want the company to play? If the company decides to promote a positive social environment without taking clear political sides, it will trend toward a faux neutrality where “hate” is any negative opinion, punishing people who criticize the status quo. But if it admits to an ideological bent, it will have to start formulating political stances on which groups worldwide deserve the most protection from hate. The more social responsibility it accepts, the more liable it is for failing to police its users, and the more power it has to control speech — not just comments to other users, but personal timeline posts or photographs.
Mark Zuckerberg describes Facebook in ways that sound more and more like a socially responsible government, and all governments set codes for their citizens to follow. But Facebook isn’t a democracy, where citizens are setting those codes (however indirectly) themselves. They can’t even see the codes. Almost everything useful we know about the platform’s hate speech rules, including today’s ProPublica report, comes from leaked documents.
I’m not trying to defend, Facebook, exactly. I’m saying that even leaving aside the logistical nightmare of moderation, Facebook is too big and centralized to address hate speech in any way that won’t seem either laughably perfunctory or dangerously overreaching. It’s not just aiming to be our new digital public square, but our digital church, digital school, and digital living room. What do we want people to be able to say and do in all these places? I can’t think of an answer to this question that seems right, because I’d rather not have it be in this position in the first place.
The most important problem for Facebook might be mitigating harm: stopping users from directly sending abuse and threats to each other, making sure that people can control what goes on in their own spaces, and working with law enforcement to watch for criminal action. It’s far from the utopian ideal that Mark Zuckerberg has put forward as a new global community, but it will prevent Facebook from accidentally doing harm — or setting itself up to fail again.