Study predicting troll behavior finds reader mood and discussion environment key factors
As publishers become more and more frustrated with the quality of their comment threads, some (like TNM) have decided to eliminate them altogether, some moving them to social media – others continue to look for ways to keep the conversation civil
Trolls remain the bane of the Internet. But why do so many web readers engage in behavior that can be considered to be trolling?
“It’s a spiral of negativity,” said Jure Leskovec, associate professor of computer science at Stanford, and one of the authors of the study Anyone Can Become a Troll: Causes of Trolling Behavior in Online Discussions (PDF).
“Just one person waking up cranky can create a spark and, because of discussion context and voting, these sparks can spiral out into cascades of bad behavior. Bad conversations lead to bad conversations. People who get down-voted come back more, comment more and comment even worse.”
Study participants were given a quiz to determine their mood before participating in online discussions. They were given 15 open-ended questions, and included logic, math, and word problems (the math problems would have immediately put me in a bad mood). The participants then began to take part in an online discussions.
In short, the authors found that “anger begets more anger” and that “trolling in a past discussion, as well as participating in a discussion where trolling occurred, can affect whether a user trolls in the future discussion. These results suggest that negative mood can persist and transmit trolling norms and behavior across multiple discussions, where there is no similar context to draw on.”
Justin Cheng, Michael Bernstein, and Jure Leskove of Stanford University, and Cristian Danescu-Niculescu-Mizil of Cornell University, authored the study which both tries to predict trolling behavior, and recommends some ideas of controlling it.
The study finds that “not only are some banned users likely to be ordinary users just having a bad day, but” banning these users “does little to curb such situational trolling, which many ordinary users may be susceptible to.”
The authors recommend that rather than banning all trolls, it might be better to consider “measures that mitigate the situational factors that lead to trolling may better reflect the reality of how trolling occurs.” This might include instituting a “cooling off” period after a reader has had a comment flagged.
The study, however, offers few other recommendations, which publishing will likely find frustrating. It finds that there is a spiraling effect at work that leads to more and more trolling, as a discussion thread declines into negativity.
“At the end of the day, what this research is really suggesting is that it’s us who are causing these breakdowns in discussion,” Michael Bernstein, assistant professor of computer science at Stanford , told Stanford News. “A lot of news sites have removed their comments systems because they think it’s counter to actual debate and discussion. Understanding our own best and worst selves here is key to bringing those back.”