Internet trolls are made, not born, researchers say

May 5, 2017 by Bill Steele, Cornell University
Credit: CC0 Public Domain

You, too, could become a troll. Not a mythological creature that hides under bridges, but one of those annoying people who post disruptive messages in internet discussion groups – "trolling" for attention—and off-topic posters who throw out racist, sexist or politically controversial rants. The term has come to be applied to posters who use offensive language, harass other posters and generally conjure up the image of an ugly, deformed beast.

It has been assumed that trolls are just naturally nasty people being themselves online, but according to Cornell research, what makes a troll is a combination of a bad mood and the bad example of other trolls.

"While prior work suggests that trolling behavior is confined to a vocal and anti-social minority, ordinary people can, under the right circumstances, behave like trolls," said Cristian Danescu-Niculescu-Mizil, assistant professor of information science. He and his colleagues actually caused that to happen in an online experiment.

They described their research at the 20th ACM Conference on Computer-Supported Cooperative Work and Social Computing, Feb. 25–March 1 in Portland, Oregon, where they received the Best Paper Award. The team included Stanford University computer science professors Michael Bernstein and Jure Leskovec, and their doctoral student Justin Cheng '12.

To tease out possible causes of trolling, the researchers set up an online experiment. Through the Amazon Mechanical Turk service, where people can be hired to perform online tasks for a small hourly payment, they recruited people to participate in a discussion group about current events.

Participants were first given a quiz consisting of logic, math and word problems, then shown a news item and invited to comment. To compare the effects of positive and negative mood, some participants were given harder questions or were told afterward that they had performed poorly on the quiz. To compare the effects of exposure to other trolls, some were led into discussions already seeded with real troll posts copied from comments on CNN.com. The experiment showed that and bad example could lead to offensive posting.

Following up, the researchers reviewed 16 million posts on CNN.com, noting which posts were flagged by moderators, and applying computer text analysis and human review of samples to confirm that these qualified as trolling. They found that as the number of flagged posts among the first four posts in a discussion increases, the probability that the fifth post is also flagged increases. Even if only one of the first four posts was flagged, the fifth post was more likely to be flagged. This study was described in a separate paper, "Antisocial Behavior in Online Discussion Communities," presented at the Ninth International AAAI Conference on Web and Social Media, May 2015 at Oxford University.

Using day and time as a stand-in for mood, they found that ordinary posters were more likely to troll late at night, and more likely on Monday than Friday.

It might be possible to build some troll reduction into the design of discussion groups, the researchers propose. A person likely to start trolling could be identified based on recent participation in discussions where they might have been involved in heated debate. Mood can be inferred from keystroke movements. In these and other cases a time limit on new postings might allow for cooling off. Moderators could remove troll posts to limit contagion. Allowing users to retract posts may help, they added, as would reducing other sources of user frustration, such as poor interface design or slow loading times.

The point of their research, the researchers conclude, is to show that not all trolling is done by inherently anti-social people, so looking at the whole situation may better reflect the reality of how trolling occurs, and perhaps help us see less of it.

Explore further: Research shows that anyone can become an internet troll

Related Stories

Algorithm able to identify online trolls

April 14, 2015

A trio of researchers, two from Cornell the other from Stanford has developed a computer algorithm that is capable of identifying antisocial behavior as demonstrated in website comment sections. In their paper uploaded to ...

Exploring the world of the Madeleine McCann trolls

March 2, 2017

A decade has passed since the disappearance of toddler Madeleine McCann on holiday in Portugal, but activity online regarding the case is constant, with some of this commentary being directed in the form of abuse, a behaviour ...

Online trolls and cyberstalkers aren't the same

February 26, 2014

The recent death of television personality Charlotte Dawson and the possible role that online abuse played in her struggles with depression shows how damaging this behaviour can be.

Recommended for you

What do you get when you cross an airplane with a submarine?

February 15, 2018

Researchers from North Carolina State University have developed the first unmanned, fixed-wing aircraft that is capable of traveling both through the air and under the water – transitioning repeatedly between sky and sea. ...

3 comments

Adjust slider to filter visible comments by rank

Display comments: newest first

julianpenrod
1 / 5 (3) May 05, 2017
Among other things, a chicken-and-the-egg situation. If other trolls can make some turn into a troll, what makes the other trolls trolls? And what made the trolls who made the other trolls trolls trolls? At some point, there have to be individuals who have a genuinely innate malignant nature.
And, don't forget, genuine trolls are always trolls. It's not the result of something bad happening to them every single last time they place a comment! Also, genuine trolls display a model of the world, such as that whites hate everybody, anyone who isn't a Democrat is necessarily stupid, sex deviances are all normal, Do they necessarily make that model up when something bad happens to them to give them a bad mood?
gkam
2.6 / 5 (5) May 05, 2017
Oh, Julian, . . are we on your bridge?
NIPSZX
not rated yet May 08, 2017
I am a huge TROLL! The reason I became a troll is I started reading Financial Economic Business blogs as a hobby because I always wanted to become a smart investor. After reading business blogs over the past 7 years, I noticed that the blogs were always exactly 180 degrees perfectly wrong with their predictions. I paid dearly for trusting these financial liars such as BusinessInsider and ZeroHedge. They mislead investors with their wrong financial outlooks. One could say that financial investing outlook predictions published before the outcome are not considered misleading if they turn out to be wrong. But posting "Market Crash" doom and gloom consistently everyday over the course of 7 years on every page, in every article, over and over again, is definitely misleading! Especially when the market has gone straight up, day after day, wrong post after wrong post! Losing all that money taught me a lesson not to believe anything written on the internet.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.