Hateful tweets multiply dramatically as temperatures become more extreme, an analysis of 4bn geo-located tweets in the US has found.
Scientists logged rises of up to 22% in racist, misogynist and homophobic tweets when temperatures rose above 42C, and increases of up to 12% when the mercury fell below -3C, according to a study by The Lancet Planetary Health.
Annika Stechemesser, its lead author and a scientist at the Potsdam Institute for Climate Impact Research (PIK), said: “We found that both the absolute number and the share of hate tweets rise outside a climate comfort zone. People tend to show a more aggressive online behaviour when it’s either too cold or too hot outside.”
The research used machine-learning algorithms to identify around 75 million English-phrased hate tweets – around 2% of the sample – in 773 US cities between 2014 and 2020.
These volumes of hate speech were then logged and statistically evaluated against variations in local temperatures by the PIK team.
They found that the lowest number of abusive messages occurred when temperatures were between 15-18C but when thermometers fell below 12C or rose above 21C, hate tweets began to rise – most dramatically at climatic extremes.
Incendiary tweets flared across all climate zones irrespective of factors such as income and political or religious outlook, the paper says.
Diego Naranjo, head of policy at the European Digital Rights network, said that the impact of the climate crisis on hate speech could be reduced by banning current corporate models aimed at maximising attention, some of which have been accused of promoting hate content.
“The Big Tech business model drives platforms to promote polarising content and as long as very large online platforms are incentivised to create viral content by spreading polarising posts, we risk more hate speech being propagated faster and putting our mental health at risk,” he said.
In June, Facebook was criticised for its failures in picking up hate speech in certain other languages, when a researcher tested its systems by submitting hate-filled fake advertisements, which Facebook approved for publication even though they called for the murder of people who belonged to different ethnic groups.
Surveys indicate that four in 10 Americans have experienced some form of online harassment, which can trigger maladies ranging from anxiety and depression to self-harm.
“Being the target of online hate speech is a serious threat to people’s mental health,” Stechemesser said. “The psychological literature tells us that online hate can aggravate mental health conditions especially for young people and marginalised groups.”
Almost three-quarters of women globally said they had been exposed to online violence in one UN study while one in four black Americans have reported facing racial harassment online.
Twitter was chosen for the PIK research as it is used by one in five Americans and many of its tweets are geo-located. Maximum temperatures are also typically recorded between noon and late afternoon, a popular time for tweeting.
The study used a UN definition of hate speech as being “any kind of communication in speech, writing or behaviour, that attacks or uses pejorative or discriminatory language with reference to a person or a group on the basis of … their religion, ethnicity, nationality, race, colour, descent, gender or other identity factor”.
The Guardian approached both Twitter and Meta for comment.
