Social and journalistic media play a big role this election. In convincing the shrinking group of undecided voters, but especially in motivating voters to go to the polls. Almost half of Americans see social media as important news sources (48%). That percentage is higher than for the medium of newspapers (16%), but is in line with television (51%), according to previous research by the Digital News Report. Professor of Political Communication Wouter van Atteveldt (Vrije Universiteit Amsterdam) is researching the effect of social media on elections.
The algorithm and the social network
Algorithms ensure that users are shown news items that match their interests. This happens because the algorithm looks at X (the user's behaviour, such as previously read articles), Y (demographics, such as location or age) and Z (the preferences of other users with similar interests). If you remove this preselection, the news becomes less personal and therefore often less relevant to the individual user.
Van Atteveldt: ‘There are good reasons to look critically at the role of social media in politics. Even if most news ultimately comes from journalistic sources, the social network's algorithm takes over part of the journalist's gatekeeper role - without the accompanying editorial statutes and regulation that ensure the independence, quality and reliability of news. In addition, news on social media is less of a shared good: it is not clear who sees what on social media, which makes it harder to hold politicians accountable for statements made - from broken promises to falsehoods and incitement.'
This criticism is not new and already after the 2016 US elections, there were well-founded fears that the elections had been influenced. Among other things, by spreading fake news via Facebook, whether from foreign sources or not. Thereupon, Meta, Facebook's parent company, had decided to enter into a very extensive collaboration with a group of (US) scientists to research the role of social media in the 2020 elections. Although researchers doubted the independence of this research, it did allow scientists to study the platform and conduct experiments on it. This showed, for example, that ‘turning off’ the algorithm led to a much less interesting platform for users. Proportionally, more political news then appears, but less interaction with political content. Also, the absence of algorithms leads to more news from dubious sources.
Content different for users
Other research within the same project found that left-wing and right-wing Facebook users were shown completely different content on their timelines. Although this was more due to who they were following than the algorithm, the results were telling: right-wing users in particular were in a ‘bubble’, in which almost all fake news identified by Facebook could be found. Especially after the election, a lot of fake news circulated in right-wing circles, culminating in the storming of the Capitol on 6 January 2021. Whistleblower Frances Haugen previously indicated that Facebook put in extra security during the election to prevent fake news, but that they undid this right after the election. Van Atteveldt: ‘This raises questions about responsibility for the Capitol Storming, but also makes us wonder whether Facebook would do the same for European or Dutch elections.’
'Without collaboration with platforms like Facebook, scientists do not get the data to answer basic questions about the functioning of these same platforms within our society and democracy. Although collaborative research with Facebook appears to be sound, it remains a commercial company that itself determines the initiative and framework conditions of the collaboration, while that collaboration is partly aimed at controlling the same company,' van Atteveldt said.
New research with data donation
A research team at Vrije Universiteit Amsterdam led by Wouter van Atteveldt will set up a completely independent study on how social media works over the next few years. Van Atteveldt is leading a new Horizon Europe project in which they will make use of data donation: they will ask respondents who participate in the study to voluntarily donate their social media data, which they can retrieve from social media platforms using the GDPR (General Data Protection Regulation). The research team is building a digital twin of the media system with open source language models to better predict the impact of regulation.
Wouter van Atteveldt: ‘Until platforms are forced to share data with researchers on a meaningful scale, this is currently one of the few ways to independently research the functioning of social media during elections. Events during and after the last US election show that this is much needed.'