Mapping social media news in Europe: A comparative actor-network investigation of authors, articles, publishers and platforms in France, Germany, Italy, Spain, and the UK

Guided by actor-network theory, this study analyzes the news ecosystem as a complex network of users, news stories, journalists, publishers, and platforms in five major European countries. The study consists of an empirical and theoretical part. It develops an empirically-based model of the journalistic network of digital artifacts (e.g. news articles), humans (e.g. users, journalists) and non-human actants (e.g. news feeds). The aim is to better understand the changing role of European legacy news outlets that operate in an environment increasingly dominated by platform technologies (Bell & Owen, 2017; Kleis Nielsen & Ganter, 2017).
In the empirical part, a quantitative analysis of news shared on social media platforms was performed. The objective was to obtain a representative sample of online news articles published by major European media organizations and their corresponding social media interactions. The sample consists of online news articles (n = 198.921) published between March 2017 – February 2018. It includes data from 20 legacy publishers and broadcasters from France, Germany, Italy, Spain, and the UK. Examples are BBC News, El Paìs, La Repubblica, Bild and France24. For each organization, up to 10.000 most shared online articles were extracted using commercial social media analytics software (BuzzSumo) that queries public data APIs (application programming interfaces). Article variables include the URL, title and the number of shares on Facebook, Twitter, LinkedIn, and Pinterest. Further variables are text length (word count) and author name.
These data were used to perform a comparative analysis across countries, social media platforms, and organization types to draw a representative image of the European social news network. The empirical results suggest significant differences in social media interaction patterns across nations, especially Germany and the UK, as well as platforms. Moreover, the downward trend of social media interactions with news content – primarily induced by tweaks to the Facebook News Feed that favor other content (Newman, 2018) – can be reproduced. Methodologically, the empirical part uses descriptive statistics, correlation analysis as well as keyword-based content analysis.
In the theoretical part of the paper, insights from ANT-studies of journalism (i.e. Lewis & Westlund, 2015; Schudson, 2015; Turner, 2005) are used to critically assess the empirical findings. For example, the number of social media interactions some (viral) articles have generated while others remain unsuccessful are explained as a result of differences in material agency. Furthermore, the downward trend in social interactions tweaks to the Facebook News Feed have created, can similarly be understood as the result of a behavioral change of a powerful algorithmic actant (Lewis & Westlund, 2015).
Such a non-anthropocentric view on the digital news ecosystem, we argue, may also help overcome problems of inscrutability of the black-box-nature of algorithmically-driven tech platforms that continue to disrupt journalism.