Despite near-unanimity among scientists regarding climate change, misinformation persists, muddying established climate science and destabilising public consensus. Ultimately, this threatens climate policy support. And as social media use has proliferated, the dynamics of climate misinformation have shifted; fake messages spread rapidly by malign actors who evade accountability. 

In the UK, the Ultra-Low-Emission Zone (ULEZ) policy has been a flashpoint for climate misinformation. The policy was met with harsh resistance by residents, particularly in Greater London, with many protesting. For instance, the vigilante group, The Blade Runners, destroyed 1,760 cameras, and ULEZ was blamed for Labour losing its seat in the Uxbridge and South Ruislip by-election, as well as significantly influencing the mayoral election, where six of eleven candidates pledged to repeal it.

Alongside genuine economic concerns, online climate misinformation has undermined support for the policy and galvanised action against it. Investigations into anti-ULEZ groups on Facebook revealed endorsements of violence, climate denial and conspiracy theories, unfounded links between freedom restriction and ULEZ and overt racism and islamophobia. Keen to investigate why and how ULEZ became a hotbed for misinformation, I used Natural Language Processing, data science and visual semiotics to analyse climate misinformation within ULEZ discourse across Reddit and Telegram. 

I employed a mixed-methods approach to analyse how climate misinformation spread online in discourse about the ULEZ policy, beginning with quantitative analysis, which served as the foundation for subsequent qualitative research. I first collected around 50,000 messages from Reddit and Telegram. These platforms provided an interesting contrast: Reddit, the self-proclaimed ‘front page of the internet,’ is open and community-moderated. Alternatively, Telegram is known for its association with far-right organising, criminal activity, and has a more closed-off structure with minimal content moderation. Following data collection, I used Natural Language Processing (NLP) tools to identify themes in the text and flag posts similar to typical examples of climate misinformation. This allowed me to use data science to analyse who and how often spreads climate misinformation.

Using a qualitative lens, I then analysed an under-researched area of misinformation: the role of images. Using the flagged posts from the NLP, I sampled the images shared alongside misleading messages. These images were analysed through visual semiotics, examining what they showed literally, what they implied symbolically, and how design choices, such as colour or composition, shaped meaning (denotation, connotation and visual grammar).

The integration of quantitative and qualitative analysis allowed me to zoom out, see the broad picture and then take a tailored, more thorough approach to analysing specific examples in detail. This provided a more comprehensive picture of both the scale of the problem and a detailed analysis of specific misinformation narratives.

My research culminated in a report designed for Climate Action Against Disinformation (CAAD), a global coalition of over 50 international organisations working to stop climate disinformation. CAAD both raises awareness of climate disinformation and lobbies social media organisations, advertising companies, and policymakers to curb its spread online. I amalgamated the most pertinent findings and used them to advise on crafting more resilient climate policy narratives. Among those findings were:

1. ULEZ discourse on Reddit tended to focus on the policy and adjacent topics, whereas on Telegram, it was situated within a more conspiratorial ecosystem. Interestingly, across both platforms, a shared resentment towards increased surveillance due to ULEZ cameras emerged as the third top theme on Reddit and the second top theme on Telegram,

2. Climate misinformation was more prevalent on Telegram than on Reddit, accounting for 1.2% and 0.8% of the discourse, respectively. Despite this, general climate discourse was more prevalent on Reddit, as reflected in the topic modelling results, where ‘air quality, emissions, and climate change’ was the second most prominent theme.

3. Five broad climate misinformation claims were prevalent throughout ULEZ discourse, although they were more pronounced on Telegram. They include a link between climate policy and restrictions on freedom, claiming that climate solutions won’t work, responding to climate discourse with an unrelated rebuttal (‘whataboutism’), presenting multiple issues alongside climate misinformation (issue-stacking), institutional distrust, and discrediting climate science and the movement.

4. Images played a very significant role in climate misinformation on Telegram, but a negligible role on Reddit. Among the images on Telegram, several were AI-generated, indicating that AI tools are being misused to generate climate misinformation.

5. Regarding who spreads climate misinformation, 2.1% of those using Reddit to post about the ULEZ contributed to climate misinformation, while 10.2% of those using Telegram to post about the ULEZ did, a nearly fivefold difference.

Outcomes aside, during the capstone process, I learned that when it comes to researching complex issues like climate misinformation, having a plan B (and probably a plan C!) along with a lot of determination and grit is key.