Twitter's Winning the War on Harassment—So Says Twitter

Twitter has vowed to make harassment a priority, and today issued its 6-month progress report. But the company could still use more transparency in how it responds to abuse.
Image may contain Logo Trademark Symbol Home Decor and Wall
Jeff Chiu/AP

When people get harassed on Twitter—which happens all the time—it's terrible for them, and also for the company, which has earned a reputation as a place that doesn't take abuse seriously. For years it hasn't focused on fixing the problem of harassment in any meaningful way. Then six months ago, Ed Ho, Twitter’s general manager of consumer product and engineering, vowed to address the problem “in days and hours, not weeks and months.” What followed were a slew of updates—seemingly every month or so. Today, in a blog post, Ho announced what affect those updates have had on curbing Twitter harassment.

Based on its internal data, Twitter appears to have made pretty good progress. According to Ho, Twitter is taking action on “ten times the number of abusive accounts every day compared to the same time last year.” It suspends and limits the reach of thousands more abusive accounts each day, and over the past four months, Ho says, new tech deployed by Twitter has been able to weed out and remove twice the number of repeat offenders—those abusive Twitter users who find themselves suspended then promptly create a new account to continue harassing someone under a new name. But Twitter did not disclose the year-over-year drop in total abuse reports, or release any kind of baseline—making it hard to know how much harassment was actually eliminated from the platform as a result of the company's efforts.

Ho also highlights in the post a team of humans at Twitter who work reviewing reports. Twitter says it is now doing a better job of communicating to people who are accused of harassment on the site. For one thing, when it restricts someone's account over abusive content, Twitter informs the account owner that they have been put into a time-out phase, and why. During this Twitter "timeout," only the followers of the account can see what these users have said—and Twitter can enforce this penalty even when no one has reported these accounts as abusive. Twitter says accounts in this kind of timeout, meanwhile, have subsequently received 25-percent fewer abuse reports. More than half of the accounts accused of abusive behavior receive this warning state just once, Twitter says, which could indicate that people learn their lesson. But by this calculation, about one in three abusive accounts repeats abusive behavior, a proportion that highlights how much work there is left to do.

James Grimmelmann, a law professor who studies social networks at Cornell University, cautions that none of these statistics are meaningful on their own. "Is ten times as many account flags good or bad?" he asks. "It depends on how many Twitter was flagging before, and how many they were missing." Of course, ten harassing tweets aimed at a user is obviously better than 100. But ten harassing tweets is still harassment.

"You want to look at crime rates, and not arrest rates," says Grimmelmann.

To really know if Twitter's efforts had a positive effect on people's experience, the company should ask people directly, Grimmelmann says. "You need some kind of user-facing instrument," he says. "You need a survey of people about their experiences: 'How do you feel Twitter is doing on harassment?'"

Twitter could also do a better job of explaining to people why their reports are not followed up by Twitter's human reviews. A recent BuzzFeed report from earlier this week, recounted that a call for anecdotes about harassment on Twitter yielded 89 messages from people who said they received improper dismissals of harassment reports, including one still active account with 899 abusive tweets all targeted at a female sportswriter. Why didn't these reports yield action? No one knows. Twitter is doing a better job of being transparent with those it blocks about why it has blocked them, but still isn't communicating to people who have reported harassment that doesn't result in action exactly why no action was taken.

Today's report shows Twitter is starting to focus on a problem that's plagued people on its platform since its earliest days. And it's aiming to be transparent—on its own terms. Along with progress, Twitter today reveals it can and must do more.