Psychologists find that Internet trolls seem impervious to any efforts to change their behaviour.
Some scientists peer into active volcanoes and try to read rocks. Others sift signals from space or analyse how animals behave. And then there are the cyber-ethnographers, who dedicate their careers to studying the way that people behave online. Some of these digital researchers must surely envy the ‘peaceful’ life of a volcanologist, for, as geologists like to say, one cannot argue with a rock.
Arguments rule the online world — witness the attention given this week to a Twitter row between Harry Potter author J. K. Rowling and journalist Piers Morgan. And although sometimes amusing, it doesn’t take much for online banter to slip towards insults, harassment and worse. That is the grim domain of the Internet troll, and it’s this murky online environment that brave cyber-ethnographers are now trying to study.
This May, it will be a full ten years since the abduction of three-year-old Madeleine McCann from her family’s holiday villa in Portugal and the worldwide coverage that followed. Yet, a decade later, people on the Internet still swap 100 messages or so an hour about the case. Many of these accuse and insult her traumatized parents, celebrating their daughter’s disappearance and gloating over their misery.
Such people are among the basest and most antisocial Internet trolls, and in a paper in Computers in Human Behaviour, psychologists describe how they tried to engage with this troll community, to study their attitudes and behaviour, and to work out what makes them tick (J. Synnott et al. Comput. Hum. Behav. 71, 70–78; 2017). Their research put them in the cross-hairs for several weeks, and the trolls did not disappoint. Once the goal of their study was exposed by others in the anti-McCann community, “you need better English to do a PHD luv!” was among the more polite messages sent in response to questions from “the psychology student studying trolls”.
Things got heated when the scientists tried to introduce some science into the debate. Much of the suspicion towards the McCann family was generated by a claim from the Portuguese police that sniffer dogs had found evidence of a cadaver in their holiday apartment (no charges were brought). When one of the psychologists posted a reference to an academic paper showing that such dogs made frequent mistakes in hot weather, and invited discussion, the trolls were more interested in insults and attacks on the researcher’s motive, labelling them a “shill” and blocking them when they tried to steer conversations back to the findings.
Previous research on trolls has identified key phrases that act as calling cards and draw activity. In this study, the word ‘shill’ — meaning that the researcher was paid by the McCann family to protect its reputation — was a red rag, and led to more and more trolls circling the discussion and piling in.
What can we learn from the study? One powerful theme of the anti-McCann messages is motherhood — and how the trolls argue that they would have behaved differently, both before and after the abduction. Psychologists call this disassociation, and it could arise from an irrational belief that parents who explicitly distance themselves from the plight of the McCann family somehow keep their own children safer. But there were much nastier motives on show, too: although most of the trolls argued that they were fighting for justice, the researchers conclude that this was thin cover for being able to hurl insults anonymously.
There are two other notable points. First, most of the abusive and offensive messages sent and received were against the rules of the social-media provider, yet no action was taken. And second, to ‘not feed the trolls’ has little impact. They are cultural scavengers who feast on alternative facts and false news already in the system, and thrive on condemnation. Rocks are so much easier to deal with.
Related links in Nature Research
Related external links
About this article
Cite this article
The dark side of social media. Nature 542, 272 (2017). https://doi.org/10.1038/542272a