If You Can’t Stop It, Hide It

Twitter has a new plan to make its community less abusive—by siloing off abuse.

Illustration by Alex Eben Meyer

Twitter is getting slightly quieter. That’s the gist of Twitter’s new changes designed to combat abuse, which it unveiled Tuesday. By its own admission, the social network has a problem dealing with abusive users: As Twitter CEO Dick Costolo wrote in a leaked internal memo, “We suck at dealing with abuse and trolls on the platform and we’ve sucked at it for years.” Every week seems to bring a new story of harassment: This weekend, U.K. food writer Jack Monroe quit Twitter after receiving a torrent of homophobic abuse. Lena Dunham quit in January, tired of threats and name-calling. Louis C.K. quit last week, not citing abuse but simply saying, “It didn’t make me feel good. It made me feel bad instead… It’s too instant.” Beyond the harrowing abuse, an outsized share of which is directed at women, Twitter increasingly resembles a school lunchroom where tribes of kids sling food at each other. A fix is long overdue. Still, what Twitter has announced won’t satisfy the company’s most vocal critics.

The network did itself no favors with the other new feature it revealed this week, one that lets users receive direct private messages from anyone, not just people they follow. While users must manually turn on this feature (no thank you) before the trolling floodgates open, journalists like Ezra Klein have criticized it and incorrectly suggested that users must opt out to avoid getting DMs from anyone. While Twitter should add an “Are you sure you want to do this?” confirmation to that feature’s opt-in, it’s clear from Tuesday’s announcement that Twitter’s evolving treatment of abusive content requires a careful reading.

Twitter is clearly loath to restrict its users’ speech, which may not be enough for those clamoring for a wide-scale crackdown on bullies and trolls, but will reassure those fearful that Twitter could overly police its users. Of the trolls, Costolo wrote, “We’re going to start kicking these people off right and left and making sure that when they issue their ridiculous attacks, nobody hears them.” But the changes announced Tuesday by the director of product development, Shreyas Doshi, aren’t nearly so aggressive. There is only one targeted change to prohibited content, which is a wider reading of what constitutes a “violent threat.” Previously, Twitter would only ban users for “direct, specific threats of violence against others,” but will now also sanction users for “threats of violence against others or promot[ing] violence against others.” In other words, where previously you would get in trouble for saying, “I’m going to drive to David Auerbach’s house and punch him in the face,” you can now also get in trouble for saying, “Someone ought to punch David Auerbach in the face” or “Punching David Auerbach in the face is a great idea.” Fair enough. (“David Auerbach’s face makes me puke,” however, is still very much OK.)

Reported violations will not always result in total account suspension, however. Twitter may force users to delete certain tweets before being allowed to log on again, or it may selectively lock functionality: An offender may be suspended from tweeting for some period of time, even though she can still log in and direct message her friends. This sort of “time-out” can be deployed much more readily than total account suspension. (There are other restrictions enforced against accounts Twitter determines to be spammers. A Twitter spokesperson told me, “Individual users and coordinated campaigns sometimes report abusive content as spam and accounts may be flagged mistakenly in those situations.” Now, Twitter’s spam restrictions won’t intentionally be used against abusive accounts.)

So far, so small. None of these changes will be enough to get Iggy Azalea back on Twitter. Up until now, Twitter’s abuse policies have been purely reactive: If a user complains, Twitter investigates. This is less than ideal, since it’s very easy to create a Twitter account and harass someone, putting the burden on the target to file a complaint and block them. If you are a popular target for abuse, this can make Twitter downright unusable unless you protect your account and go private, which defeats the purpose of Twitter as a public medium. This is where the most technically interesting change comes in. Twitter will attempt to detect abusive tweets proactively, using a combination of signals including the age of the account and a tweet’s similarity to other tweets deemed abusive by the social network’s internal team. This automated flagging will not block the tweet from being sent or cause any restrictions on the sender’s account. Instead, if my nemesis creates a new account and declares his desire to punch me in the face, I won’t see it in my notifications, even if he tags my Twitter account in his tweet.

Depending on how aggressive Twitter is, this change could actually make Twitter a nicer place. Twitter is attempting to thread the needle here by taking a view of abuse as something that one person directs at another person, rather than a statement simply made in public. By preventing a harasser from making contact with his target, Twitter hopes to leave the harasser off in his corner to rant and threaten away while no one is paying attention. My detractors can detail their ugly plans for me publicly on Twitter, but Twitter will try to keep people (including me) from seeing them. And this should be taken as a sign that Twitter is trying to avoid proactively restricting speech as much as possible. In the event of a false positive—a tweet that Twitter flags as abusive when it is in fact not—the worst that happens is a user does not receive a notification of a tweet mentioning him.

Combined with Twitter’s promise of speedier and more aggressive enforcement of complaints—Twitter general counsel Vijaya Gadde writes that “We have also tripled the size of the team whose job is to protect users, which has allowed us to respond to five times the user complaints while reducing the turnaround time to a fraction of what it was not long ago”—these changes allow for a few more shades of discipline in between full privileges and total account suspension. If Twitter follows through, it will hopefully become much easier to restrict problematic accounts far more quickly. The changes do not, however, do much to address the confusion that arises from Twitter’s strange conflation of the public and the private, where strangers can interrupt a conversation you’re having with a friend in order to argue with you, or you can be dogpiled by a Twitter mob politely (or impolitely) telling you how wrong you are, or a group of users can loudly debate whether you’re more ugly than stupid.

Furthermore, despite Costolo’s commitment to weed out real abuse more aggressively, Twitter shows no sign of expanding its definition of abuse to encompass merely unpopular or controversial speech. Ridicule, mocking, insults that aren’t profane, and simple puerility aren’t going anywhere. People can still call me “dumb as hell” or compare me to Hitler and Twitter won’t keep their tweets out of my notifications. People can be rude, sexist, and bigoted and Twitter will not stop them unless they go way over the line into abuse. Recently, Costolo has even said that some people have too broad a definition of harassment: “I’ll get emails from people that say, ‘I agree, and here’s a great example of someone being harassed on the platform’—and it’s not at all harassment, it’s political discourse.” This, again, is a consequence of Twitter’s character limit and generally thoughtless pace, thanks to which a bluntly expressed political opinion can seem rude to the point of offense—especially when delivered by 20 people in tandem. But there is little Twitter can do without radically changing its format. I have written that Twitter’s social problems arise from its structure rather than from its content. Fixing it would require revising the underlying communication paradigm in drastic ways.

To the extent that Twitter is broken, it is broken by nature, not through a lack of policing. Users will simply have to ignore content that doesn’t rise to Twitter’s definition of abuse; Twitter’s biggest move Tuesday is not a change in enforcement but a change in functionality that makes content easier to ignore. Overly restricting the content would only put Twitter in the uncomfortable position of arbitrating discourse, which is not a business it should be in, since standards of discourse differ from country to country and Twitter responds to government requests for takedowns as well as individuals. As the EFF’s Jillian York has written, the lack of transparency and the potential for arbitrary decisions can easily cross the line into censorship, even accidentally, so Twitter’s caution is explicable, if not necessarily reassuring to those who wish that everyone would just be nice all the time on Twitter.

Besides, while “Out of sight, out of mind” may not sound like a reassuring maxim, it is in reality what we do every day. It is only because of the radical public transparency of Twitter that we are made excessively aware of the unpleasant words of a lot of people who we’d be better off ignoring. If Twitter survives, it will not be through content policing, but because its users’ tolerance will outweigh their toxicity. It’s a long bet. Twitter can reduce the abuse, but it may not be able to make Louis C.K.—or many of us—feel good about ourselves.

This article is part of Future Tense, a collaboration among Arizona State UniversityNew America, and Slate. Future Tense explores the ways emerging technologies affect society, policy, and culture. To read more, visit the Future Tense blog and the Future Tense home page. You can also follow us on Twitter.