Internet Trolls Are Just Like Us, Researchers Say

Why do people become internet trolls, leaving comments on articles and posts meant to disrupt conversation, inflame emotions and provoke an impassioned response? A team of researchers at Stanford University wanted to find out.

In a new paper coming out as part of the 2017 Conference on Computer-Supported Cooperative Work and Social Computing, the team led by Justin Cheng, a computer science researcher, found that context is deeply important to predicting whether or not people will leave trolling comments.

"A lot of news sites have removed their comments systems because they think it's counter to actual debate and discussion," said Michael Bernstein, assistant professor of computer science at Stanford and co-author of the paper.

"We wanted to understand why trolling is so prevalent today," Cheng said. "While the common knowledge is that trolls are particularly sociopathic individuals that occasionally appear in conversations, is it really just these people who are trolling others?"

But they found that, rather than simply being a result of a bad personality, trolling stems from a set of factors over and above individual dispositions. In a press release, the authors, perhaps too strongly, conclude, "Under the right circumstances, anyone can become a troll."

More cautiously, the evidence suggests that many people have the propensity to become trolls in situations when they otherwise would not.

One study the team carried out tested 667 subjects and their responses to a sample article comment section. Some of the subjects were given an easy test to complete beforehand, while others were given a hard test. Some of the articles already had trolling comments on them, while others had neutral posts.

Here's what they found:

• 35 percent of people were inclined to post trolling comments (they were asked to post a comment as part of the experiment)
• 50 percent of people left trolling comments when they took a difficult test before the experiment, and who thus already felt frustrated, or when there were already tolling comments posted
• 68 percent of people left trolling comments when they both took the hard test and saw other other trolling comments

This suggests that the likelihood people will become an internet troll is strongly tied to their present mood and the behavior of others.

"It's a spiral of negativity," said Jure Leskovec, associate professor of computer science at Stanford and senior author of the study. "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."

Another study analyzed millions of comments on CNN in 2012. It found that trolling comments were associated with times when people are known to have worse moods, such as late at night and at the beginning of the week, as well as with conversations that included comments that had been flagged.

"People who get down-voted come back more, comment more and comment even worse," said Leskovec.

The final study involved creating an an algorithm that predicted trolling behavior. This time, mood-related factors seem less important.

But unlike what many people assume, user IDs were not the most predictive factor with which to anticipate a trolling comment — instead, the level of civility in the existing conversation was much more predictive.

In other words, while some people have a tendency to be trolls, it's the tone of the discussion itself that is most decisive.

So how can we reduce trolling and make the internet a more friendly place to be?

"Understanding what actually determines somebody to behave antisocially is essential if we want to improve the quality of online discussions," explained Cristian Danescu-Niculescu-Mizil, assistant professor of information science at Cornell University and co-author of the paper. "Insight into the underlying causal mechanisms could inform the design of systems that encourage a more civil online discussion and could help moderators mitigate trolling more effectively."

The authors suggest features that could pause commenting when the conversation gets heated could give users time to cool off. They also suggest implementing "systems that automatically alert moderators to a post that's likely to be a troll post or 'shadow banning,' which is the practice of hiding troll posts from non-troll users without notifying the troll."

"At the end of the day, what this research is really suggesting is that it's us who are causing these breakdowns in discussion," said Bernstein.

Photo credit: Twitter Screenshot