It Takes a Troll to Catch a Troll

Trolls gonna troll? Not for long, maybe. Reuters/Morris Mac Matzen

Comments sections, or "the comments," as they're usually called (sometimes with a little venom), have been a dilemma for online media practically since media went online. Media outlets want to encourage open and free discussion, and there's the possibility that an incisive comment might add value to a story. Most comments sections, however, are full of dreck, a filthy Cheetos-encrusted beard on the handsome face of a well-written story.

"It's impossible to brace yourself emotionally for how it feels the first time it happens to you, the first time there's a real horrible shit-storm [in the comments]," says Erin Gloria Ryan, managing editor of Jezebel, a Gawker Media blog about women's issues. If she thinks a topic is likely to rile up the trolls, she refuses to look at the comments section entirely. "It sucks, because Gawker can be at its best when the writers and commenters are interacting." But sometimes even seemingly uncontroversial subjects inflame the commentariat, according to Ryan. "You write a piece about something innocuous, and you scroll down and they're posting 15 GIFs of violent sex," she says.

The trolls seem to be winning. Open and free comments sections are being replaced on many sites by moderated comments sections. In response to this summer's frenzy of rape GIFs, for example, Gawker Media revamped its comments. Now, by default, only moderator-approved comments are visible. Not even the so-called traditional, legacy media is immune to the insidious influence of trolls. In November, Dan Colarusso, executive editor of Reuters Digital, announced the newswire was doing away with comments on its news stories, though it would retain comments on opinion pieces.

Given the choice, most media organizations would prefer having comments, but the time and energy required to moderate them to make sure they are interesting, or even on-topic, is increasingly outweighing the perceived benefits. And, even when moderators ban trollish commenters, there is nothing to stop those trolls from registering new accounts with fresh email addresses and starting over. It's a maddening situation for those tasked with enforcing civility in comments sections.

One troll thinks he's found the answer. In his free time, Rurik Bradbury runs the popular @ProfJeffJarvis Twitter account that lampoons CUNY journalism professor and occasional slinger of technobabble Jeff Jarvis. At work, Bradbury is the chief marketing officer of Trustev, an online retail fraud-detection software startup. Trustev sniffs out fraudsters with stolen credit cards on online retail sites using a combination of behavioral clues, location data and thousands of other data points. Trustev passes the identity of the fraudster along to the retail company, and he or she is banned—permanently, Bradbury says. You can't just re-register with a brand-new email address—Trustev already has your scent. It's built a profile of you that goes beyond your email and your IP address.

Bradbury is a very active Internet user. He reads about 50 or 60 blogs a day, he tells me, and loves reading the comments. "I feel like it always leads to a better story," he explains. But, increasingly, they've become "completely out of control," he says. So, when self-identified members of Gamergate—the amorphous online movement that claims to be fighting for increased transparency and ethics in video games journalism—issued death threats to several prominent female members of the video games industry for daring to voice their opinions about gaming, Bradbury had had enough. "I said, 'Someone needs to fix this.'" If his company's software could beat credit card fraudsters, he figured it could beat trolls.

Trustev for Publishers is "a slight variant on the existing software," Trustev's CEO and co-founder Pat Phelan tells me. Where Trustev is designed to ferret out credit card fraudsters, Trustev for Publishers uses similar technology to target trolls, harassers and abusers. "What's happened up to now is, when you see a guy being absolutely vicious, you ban him," he says. "But what you're really blocking is an email address." And with free email services like Gmail, Hotmail and Yahoo! Mail at a troll's fingertips, there's nothing stopping one from creating a new email address and a new account from which to spread vitriol, which then has to be ferreted out and blocked. "It's like whack-a-mole. It's just not economically viable," Bradbury says.

"But everybody has just accepted it and said, 'There's nothing we can really do,'" according to Phelan. But he thinks his company's software will solve this problem.

It works by creating a "digital fingerprint" for every user who signs up for an account to comment. The software logs identifying information—your IP address, your device ID (each device, be it a computer, mobile phone, or tablet has a unique device ID), your browser ID, your cookies, your device's stated location, your actual, physical location, and more. It also logs "behavioral biometrics"—things like how quickly you type and how you move your mouse around your screen as you surf the web. Researchers at the College of William and Mary say "mouse dynamics"—that is, the pattern you create as you move your mouse around your screen—"are relatively unique from person to person and independent of the computing platform." In other words, Trustev for Publishers can tell you're you, even if you switch devices, IP addresses, or even your physical location. "What we've built here is a complete key that's absolutely able to ID trolls at a very, very high likelihood," Phelan tells me.

Once the software has your fingerprint, Phelan says a publisher can use it to ban you from commenting—forever. "We have an intelligent blacklist that allows us to constantly scan for these behaviors," Phelan says.

For now, Trustev is marketing their software to publishers, but Bradbury thinks content platforms like Twitter and Facebook could also benefit. "All it takes is a small number of people being abusive on Twitter to spoil it for everyone," he says. He hopes his company's new software will "change the dynamic" between abusers and publishers. "Hopefully, if [abusers] realize they can be blocked, it will modify behaviors and change the places they congregate."

Maybe, in other words, we can finally drive the trolls back under their bridges.