Last month, in Wired, Adrian Chen published a vivid story about an obscure group of dot-com workers. It was called “The Laborers Who Keep Dick Pics and Beheadings Out of Your Facebook Feed,” and its subject was the content-moderation industry: a constellation of independent companies, many based in the Philippines, that employ an army of censors who try to delete objectionable posts and videos as soon as they appear. One pseudonymous moderator said that having to watch “bestiality with children” bothered her; another, an American, was affected by videos of animal torture. An expert estimated that there were “well over” a hundred thousand content moderators worldwide, working for companies like Facebook and YouTube.
Most social-media companies prefer not to talk about their content-moderation systems, but last week Twitter divulged a few details. The company confirmed that it is now working with Women, Action, and the Media, an advocacy group that promotes “gender justice.” To that end, WAM developed an online tool designed to help Twitter users report harassment—“gendered harassment,” in particular, but also other kinds, including “racist” and “transphobic” harassment.
Upon closer inspection, Twitter’s collaboration with WAM turned out to be less impressive than it first seemed: WAM can’t do more than forward complaints to Twitter, and Twitter says only that “WAM is one of many organizations we work with.” Even so, Amanda Marcotte, at Slate’s XX Factor blog, voiced her hope that this collaboration would improve Twitter’s harassment-reporting system, although she also argued that “pretty much anything is better than the current system.” Meanwhile, the independent blogger Andrew Sullivan sounded the alarm, asserting, “Twitter has empowered leftist feminists to have a censorship field day.” In a follow-up post, Sullivan cited the example of a Breitbart editor named Milo Yiannopoulos, whose Twitter account was apparently suspended and then, about twelve hours later, unsuspended, possibly because of some tweets that were intemperate but not explicitly threatening.
What counts as “censorship” on a platform like Twitter? Magazines and blogs are typically free to reject articles—and, for that matter, to delete offensive reader comments—without being accused of censorship. Users of social-media sites typically expect more latitude, though the amount and type they get varies. Facebook asks users to use their “authentic” names; Tumblr doesn’t allow “promotion or glorification of self-harm”; Instagram bans some kinds of nudity. Like all of these platforms, Twitter thrives by creating for its users a world with a lot of freedom and a little bit of regulation, which means that its executives must constantly try to figure out exactly how anarchic its users want this world to be.
Twitter’s current anti-harassment initiative seems like a response to the growing awareness about online harassment, and especially the particularly tenacious and disturbing harassment that many women face. One writer, Imani Gandy, suggested that many of the unwanted tweets she received came from one user with multiple accounts, whose linguistic arsenal included racial slurs. (She eventually got a sympathetic response from Dick Costolo, Twitter’s C.E.O., but she said that she wanted more “transparency” about the company’s policy on harassment.) In August, after the death of Robin Williams, his daughter quit Twitter after receiving upsetting messages from two accounts. Early this year, in Britain, two Twitter users were sentenced to jail after sending a series of menacing tweets to a feminist writer, one of which said, “Rape?! I’d do a lot worse things than rape you!!’” Although some observers argued that Britain shouldn’t jail users for obnoxious or abusive tweets, virtually everyone agrees that Twitter has the right to set and enforce its own rules.
But “harassment” is a slippery word, especially on Twitter, whose architecture turns public posts into personal messages. You don’t just tweet about someone, much of the time; you tweet at someone, even if you’re mostly talking to your own followers. And that someone may experience your tweet as a kind of invasion: on a cell-phone screen, a new tweet can seem as intimate, or as unsettling, as an incoming text message. This is the genius of Twitter, which has intentionally (and lucratively) blurred the line between different modes of communication. On a platform designed to turn idle thoughts into urgent messages, it can be hard to distinguish between a crude or splenetic observation and a menacing threat.
Around the same time Twitter was confirming its new protocol, yet another high-profile user was discovering just how much trouble a few tweets can cause. The troublemaker was the darkly Dionysian comedian Artie Lange, who regaled his quarter of a million followers with a sexual fantasy about an African-American television personality. “I’m T. Jefferson & she’s my slave,” he wrote last Tuesday morning. “She beats the shit out of me & runs free.” In follow-up tweets, which were increasingly graphic, he included her Twitter name.
In the hours that followed, Twitter users rallied to her side, denouncing Lange, whose upcoming appearance on the Comedy Central show “@Midnight” was cancelled. On Wednesday, he issued a semi-apology to his target: “if this hurt u in any way I’m sorry,” he wrote, using her Twitter name again. (She didn’t tweet back, except to send an all-purpose “THANK YOU” to supporters; since she evidently isn’t eager to be associated with this incident, it seems unnecessary to name her.) A few hours later, he added, “I’m sorry to whoever was hurt by the tweets.” But later that night, at the Comedy Cellar at the Village Underground, he was his old self again. “My liver is bloated, I got type two diabetes, I got about two and a half years, tops—I’m just going to embrace the racism,” he said. “I’m going to become a racist country singer. I’m starting to write some songs. If you want to help—anybody know any words that end in ‘igger’?”
Lange’s problem, of course, wasn’t Twitter “censorship” but the opposite: the platform efficiently amplified his thoughts and then, just as efficiently, amplified the response. (If he had said the same thing at the Comedy Cellar, there wouldn’t have been as much of a reaction.) Twitter executives surely know about this episode, and they haven’t removed the inflammatory tweets or blocked Lange’s account—perhaps they feel that his fantasy didn’t quite cross the line into harassment. Were they wrong?
Users who want Twitter to adopt much stricter policies against harassment might consider the case of Bill Cosby, whose extraordinary comedic career has come to be overshadowed, especially online, by allegations of sexual assault. On Monday, a message (since deleted) on Cosby’s Twitter page asked fans to add their own text to images of Cosby. A number of users responded by creating grotesque juxtapositions, adding caustic sexual-assault references to the smiling photographs of Cosby. This was probably harassment of a sort, too, although—depending on your feelings about Cosby, and about the allegations, and about gender violence more broadly—you might consider it noble harassment. It’s easy to scoff at Sullivan’s fears about the coming reign of “leftist feminists,” but he’s right to suggest that Twitter’s new harassment protocol won’t be ideologically neutral. Censorship—even the kind of small-“c” censorship, without which no magazine or social-media platform or comedy club can endure—never is.
