This month we have witnessed the end of arguably the most consequential Presidential Election in US history. Many of us have followed the contest between Vice-President Kamala Harris and former President Donald Trump via social media. Even those with no prior interest in US politics will have found it hard to avoid the social media commentary, video footage, and memes about the two candidates. Light relief has been found in memes about Kamala Harris’s ‘Brat Summer’ and the viral video remix of Trump’s claims that Haitians were eating dogs and cats in Springfield, Ohio. However, these have been overshadowed by the sheer volume of online hate and misinformation being spread online. There have been concerns that the racist and misogynistic attacks upon Harris and other women of colour standing for Congress may deter others from running for political office in the future. Meanwhile, Twitter/X owner Elon Musk has amplified conspiracy theories about the election, using his microblogging site to promote false claims that electronic voting machines can be easily manipulated to allow Harris to win. The Centre for Countering Digital Hate has beenone of many organisations to provide evidence that Twitter/X is failing to counter false claims about the election. They found that 74% of misleading posts did not display a Community Note flagging that they were promoting inaccurate information. Reading through online comments can be an ordeal at the best of times, but interactions between Trump’s ‘MAGA army’ and supporters of the Vice-President have been particularly toxic. One mightalso have been forgiven for thinking that Trump was bound to win by a landslide, given the number of TikTokers declaring their support for the former President over the past few months. A network of political propaganda accounts powered by AI were found to have posed as real people to promote Republican candidates on various platforms. This has contributed to a febrile atmosphere in which it is highly likely that many Trump supporters are likely to dispute the result and legitimacy of the election.

This is not the first time that social media have been accused of distorting public opinion.There was evidence during the recent UK General Election that automated bot accounts were being used to exaggerate the level of support for the Reform Party. Chatbots, produced using Google, Microsoft and OpenAI, were found to amplified false information during EU elections across Europe in the first few months of this year. Twitter/X was temporarily banned by the Brazilian Supreme Court in September 2024 for its failure to suspend accounts found to have spread disinformation. It is also not that long ago that Facebook was forced toapologise for its role in amplifying disinformation that contributed to rioting in Myanmar and Sri Lanka. As Siva Vaidhynathan notes, the amplification of inflammatory content is a design feature of these platforms, not a defect, that generates revenue through every click, like, or share.

Social media are not a proxy for public opinion, especially on contentious political issues.Algorithms can be manipulated through the use of bots to artificially inflate certain narratives or drown out others through misinformation and information pollution. Not every social media user will leave a digital trace showing how they feel about such issues. This reflects the fact that most social media users do not follow journalists or politicians, nor participate in political debates online. News avoidance on social media continues to increase, with a recent study suggesting that 43% of US citizens consume news less than once a month. In the UK, the percentage of people avoiding the news sits at an all-time high of 39%. There has alsobeen some evidence to support the ‘spiral of silence’ theory, namely that some users will refrain from voicing opinions online due to a fear of being ostracised or subject to abuse. A comparative study of online adults in France, Germany, the UK and US found support for this theory, expressing concern about the implications of self-censoring behaviour for democratic values.

Moreover, a cursory look at who uses social media shows why they are unreliable barometers of public opinion. Twitter/X, the platform of choice for many politicians and journalists, was used by fewer than one in four US citizens prior to Musk’s takeover in 2023. The most recent data shows that the number of active daily users in the UK has declined from 8.5 million to 5.5 million in the past year. Research also shows that white men tend to have more followers and disproportionate influence on information flows on the microblogging site. At the other end of the spectrum, sites like TikTok are synonymous with young people and attract less attention from older social media users. A recent Ofcom report found that Instagram, Facebook and TikTok were the most important news sources for 16-24 year olds in the UK, with those aged 55 and more likely to rely on television news such as the BBC News Channel and Sky News. There is no social media site that can credibly claim that its user base is representative of the entire population of countries like the UK or US.

Like it or not, our online lives are increasingly curated by algorithms. What we see on social media reflects our reality, not necessarily that of others. Therefore, we should treat all claims about online public opinion on politics with a degree of caution.

Read more on Dr Paul Reilly's work and his book, Digital Contention in a Divided Society.


First published: 7 November 2024