How can we better understand the role of social media in spreading climate misinformation?
Max Falkenberg and Andrea Baronchelli explain how a more nuanced view of online climate scepticism might be achieved and why it is essential that social media companies share better quality data with researchers.
Is social media making the spread of climate misinformation worse? Since Elon Musk’s takeover of Twitter, we have been asked this question a lot. Our analysis, originally developed for The Times, shows that the answer is probably yes – the last six months have witnessed a large increase in climate-sceptic terminology on Twitter. However, the focus on classifying content as ‘misinformation’ may not be not that useful, as it is hard to define: understanding the structures that enable the spread of information, accurate or not, is just as important and potentially less problematic. Unfortunately, the studies that strive to build this understanding continue to suffer from the lack of data made available for research by social media companies.
Moving away from the focus on misinformation acknowledges how hard it is to fairly define the boundaries of climate misinformation. Yes, there is climate denial on social media. Our recent study published in the journal Nature Climate Change highlights common claims like ‘climate change is not real’ and ‘climate change is not anthropogenic’. However, such statements of outright denial are in a small minority in the broader context of the climate-contrarian views found on social media.
Indirect forms of scepticism – discourses of delay
Where there has been major growth is in the indirect forms of climate scepticism. These include ideas that question the pace of the green transition or attempt to discredit the actors involved in debating and implementing climate action. Both fall under the umbrella of claims called the ‘discourses of delay’, which can be used to oppose action, but not everyone discussing such ideas is a climate denier or is acting in bad faith.
The best example of such a claim is perhaps the accusation that ‘politicians are hypocrites’ in calling for climate action, for instance by arguing that they maintain their climate-unfriendly habits while placing the burden of action on the poorest. There has been a huge rise in statements of this kind in the past two years, most notably in relation to politicians’ use of private jets. However, a critical distinction between these claims and claims of outright climate denial is that the former are made by contrarians and pro-action activists alike.
In the example of ‘action will hit the poorest’, a climate activist may highlight the burden faced by low- and middle-income countries, emphasising the need for loss and damage funds, whereas a contrarian may argue that phasing out fossil fuels will hurt society’s poorest, including in wealthy countries. Such rhetorical differences emerge at the second level, but often blur together at the first.
Nuance matters, but social media and nuance rarely mix. Attempts to use technology to automate the classification of climate misinformation rely on the language of strong ‘climate denial’, and as a consequence miss the much larger community with softer views expressing scepticism about climate action. From a policy perspective, understanding this larger community whose views are harder to classify is likely more important than focusing on individual accounts that heavily engage in climate denial.
How can the full breadth of views be better considered?
In our recent work, we opted to study structure – looking at the network of interactions between users where interactions may include likes, retweets, comments, etc. – instead of content. The benefit of such an approach is that it is robust to the emergence of new trends, and captures the broader climate-contrarian echo chamber – an environment in which users amplify and reinforce a shared narrative, isolated from opposing views – without focusing on specific contrarian terminology. This is critical since many of the most prominent actors hoping to delay climate action use deliberately moderate language. Few claim that climate change is not real. However, they often share the same structural echo chamber as those with more extreme views.
Our recent study uses this approach to investigate discussions about the UN COP summits and finds a large increase in climate polarisation over the past few years. Notably, the study shows that engagement, in the form of retweets, with members of the contrarian echo chamber increased 16-fold between COP21 (in 2015) and COP26 (in 2021) (a factor of 4 larger than the increase among the pro-climate echo chamber). Critically, this growth appears driven by climate contrarians broadening their audience across the political right.
More data is needed for a fuller picture
Our findings have been made possible thanks to the data Twitter makes available to researchers. But there is no guarantee that this data will remain accessible in the future, given Twitter’s change in ownership. If anything, Twitter is the outlier: most platforms are far more restrictive with their data access policies.
A review by our co-author Warren Pearce and colleagues highlights this well: of the 35 papers studying climate change communication on social media published up to January 2018, 28 focused on Twitter, five on Facebook, and only a handful considered other platforms. Not a single study looked at Instagram, a platform that already in 2018 had twice the number of active users as Twitter. Why? Because we do not have access to the data. Studies on Instagram have been published since, but they remain in a small minority.
Some platforms appear to be accommodating, but only offer public posts to researchers – a tiny fraction of all Facebook content, for example – in an attempt to preserve user privacy. Of course privacy concerns are important, but there is a risk that these considerations become an excuse social media companies use for not being more transparent with their data and algorithmic choices.
Teams like ours, part of the IRIS Academic coalition studying ‘infodemics’, will continue to try to understand the systemic elements of the social media ecosystem that enhance or inhibit certain types of information. But we cannot force social media companies to provide the necessary data to make this possible. It is incumbent on politicians and regulators to ensure that social media companies cannot hide behind the veil of data privacy, otherwise future studies on infodemics, related to climate change or other issues, will continue to be limited to anecdotal evidence and simple term-based analyses.
The views in this commentary are those of the authors and do not necessarily represent those of the Grantham Research Institute. Max Falkenberg and Andrea Baronchelli are based at City, University of London and The Alan Turing Institute.