In the aftermath of May’s European Parliamentary Elections, the European Commission stated that their “efforts… contributed to limit the impact of disinformation operations, including from foreign actors, through closer coordination between the EU and Member States. However, much remains to be done. The European elections were not after all free from disinformation; we should not accept this as the new normal. Malign actors constantly change their strategies. We must strive to be ahead of them. Fighting disinformation is a common, long-term challenge for EU institutions and Member States.”
Investing in independent research and sophisticated data analytics is a crucial part of the effort to fight disinformation and help inform responses. In the months leading up to the European elections, Alto Data Analytics researched signals and networks of disinformation in Europe’s digital public sphere.
Alto’s analysts collected data from a wide range of public digital sources, including social media, public forums, blogs, digital communities, discussion boards, news, video, wiki sites, and other sites, from mid-December 2018 to the end of May 2019 in five key countries: France, Germany, Italy, Poland, and Spain. The data lake included more than 898 million results indexed from over 95 million users interacting and generating content in the digital public sphere.
The Commission is right that the May elections were not free from disinformation. Alto’s independent research discovered several patterns by which disinformation took shape and functioned in pre-election campaigning.
Europe’s populists from right and left flooding social media
The research identified a very small number of highly active social media profiles sympathetic to right- and left-leaning nationalist and populist parties (such as Germany’s Alternative for Germany (AfD) and Spain’s Vox or Podemos). These profiles accounted for a substantial portion of total posts on Twitter, Facebook, YouTube, Instagram, and other digital domains.
In several cases, a range of between 0.05 percent and 0.16 percent of high frequency profiles contributed between 9.5 percent and 11 percent of total activity across each of the five countries. That amount of activity and noise from a comparatively tiny portion of participants in the public sphere digital debate is noteworthy. Abnormally high activity does not necessarily mean the presence or circulation of disinformation, but it is often an important signal of the intent to propagate narratives and information in a systematic way.
In France, for instance, Alto’s researchers identified 280 users with abnormal activity, comprising 0.13 percent of all users. The most active French users produced at least eighty-one posts per day. Over half of these high frequency users were in digital communities associated with the right-wing populist party Rassemblement National, and these users produced 57 percent of the comments produced by the group of 280.
Abnormal digital activity in France
New and emerging media domains
Among the top media domains shared in each country, Alto’s researchers identified a number of sites as potential sources of disinformation and radicalization. In many cases, these new and emerging domains bear several common features, such as a similar aesthetic look and feel, a lack of transparency in ownership and authorship of articles, recent creation within the past one to three years, funding via programmatic advertising, and strikingly similar content themes.
Other characteristics of sites created for disinformation purposes include high penetration in key communities of interest and extremely segmented content focused on specific polarizing themes.
In Italy, for instance, domains such as imolaoggi.it, voxnews.info , ilprimatonazionale.it, and riscattonazionale.org are examples of relevant sites that were shared with high frequency and achieved substantial impact over the period analyzed. Other sites such as cronacapiu.it and cronacaora.it were discovered during the research and then became inaccessible in early March.
Media content aimed at disinformation
A key focus of the research aimed to identify specific themes, issues, and content actively propagated in ways which could signal disinformation efforts. A predominance of anti-immigration and anti-establishment content was consistently pushed into mainstream debates. There were also strong signals of coordination in some networks. These included the ultimately successful efforts to influence the public agenda on the UN Migration Pact, a bizarre story which started on the American conspiracy theory site InfoWars, and a controversy over an open letter apparently signed by senior French army officials accusing President Macron of “treason” which had a significant impact in the French media space.
The influence of foreign media
The presence of foreign media, notably Russia Today (RT) and Sputnik, varied significantly depending on the country. In both France and Germany, content from RT and Sputnik was distributed across several communities and reflected levels of penetration higher than several well-established national and international media outlets. In France, for example, RT’s and Sputnik’s production of anti-Macron and anti-establishment content around the Yellow Vests movement helped propel them into the ranks of key media, with heavy interaction by communities with an affinity for the Rassemblement National and La France Insoumise parties.
Polarization has become a characteristic of politics in the United States and parts of Europe over the past few years. Digital discourse in online communities magnifies this new reality. Alto’s analysis demonstrated a shift from traditional left-right communities and political discourse to a well-defined axis revolving around anti-establishment versus establishment positions.
The legitimacy of established groups, including political parties, national governments, the EU, and the UN, proved divisive in online conversations. Contentious political and social topics such as the Yellow Vests movement in France and the reaction to the UN Migration Pact across European online communities exposed wide gulfs between online discussion participants.
Themes with an anti-establishment bias tended to blend national and local narratives around identity, sovereignty, culture – such as guns and bullfighting in Spain – and gender. For each country, Alto analyzed the most frequently used keywords present in the public digital conversation to identify salient themes.
The example below is from Italy and shows the top 273 keywords & hashtags which represent 47.4 percent of all results analyzed based on the highest frequency terms interconnected from the public discussion narratives:
Key issues discussed in Italy
Networks of multi-language users
Alto’s team identified a network of 533 users posting in several languages with strong anti-immigration messaging. Many of these posts had a specific theme related to the UN Migration Pact. The users’ patterns of behavior and activity, in addition to their themes of discussion, were tracked up to the elections. The tone of their discourse was polarizing – notably anti-Islam and anti-immigration – and the domains with which these users most frequently interacted included several previously identified sites producing disinformation and misleading content. Following a trend of recently created accounts demonstrating anomalous activity, Alto’s analysts were able to identify that a third of these accounts were created in either 2018 or 2019. It is interesting to note that Twitter appears to have removed some of the individual accounts between March and May 2019.
Independent research and insights
The key findings of Alto’s EU elections research programme demonstrate the complexity of disinformation in the digital ecosystem. It’s particularly concerning that a small group of highly active users can generate enough content to significantly shape the digital discussions around polarizing issues such as immigration.
In a world where individuals are increasingly reliant on digital media to inform and educate them about topics within larger policy debates, this level of influence by a comparatively small section of online discussion participants can shape how wider political issues and events unfold. While it is notoriously difficult to quantify disinformation’s effect on large votes such as the EU elections, its very presence merits continued investigations into its sources and concerted efforts to halt its spread.
Alto’s findings demonstrate the importance of tracking the constantly evolving vulnerabilities of the digital ecosystem and disinformation tactics. Continuing to invest in independent research and intelligence is crucial to help inform the responses from regulators, politicians, technology companies and other stakeholders in democratic societies.
Alex Romero is the CEO and Founder of Alto Data Analytics.