We built an algorithm to track bots during the European elections – what we found should scare you

Emmi Bevensee, Alexander Reid Ross, and Sabrina Nardin, May 22nd 2019, The Independent,

Computer programmes aren’t going to drag the pencil out of your hand and scrawl an X on the ballot paper for you, but they are doing everything they can to make sure you head to the polling station with a warped idea of popular politics. And by a large majority, it is the right wing of politics that is doing the warping.

Using a programme we developed to study bot interference in elections, we found that some 12 per cent of tweets using hashtags promoted by far-right EU parties came from users which showed tell-tale signs of full automation. That is double the average across all parties.

We studied Twitter hashtags supporting respective EU political groupings across five different countries – Spain, Italy, Germany, the UK and France – and looked at the most active users from about 12,500 of the most recent tweets. While bots exist in each study area, we found that they target southern Europe the most, with Italy and Spain taking the lead.

Overall, we found that nearly 6 per cent of tweets which promoted hashtags for all parties and nations studied originated from such accounts. However, the bots were markedly overrepresented in hashtag campaigns supportive of the far-right Movement for a Europe of Nations and Freedom group at a full 12 per cent.

Second to that grouping came the right-wing Eurosceptics with the Alliance for Conservatives and Reformists in Europe with around 9 per cent.

In some ways, the far right’s relatively large proportion of bots is unsurprising, because it tends to have more ambitious and aggressive social media strategies. Matteo Salvini, the leader of the far-right party League has energetically adopted social media, and is the top influencer by far for the hashtag #26maggiovotoLega.

It is also true, though, that the social media influence of Salvini’s account is at least partially inflated by bots logging more than 400 tweets per day on average. The majority of all bots we found in the #26maggiovotoLega hashtag were created in May 2019 and almost exclusively retweet everything that Salvini’s account tweeted. In two days, from 13 May to 15 May, the total number of tweets of these accounts increased drastically (from seven to 428, on average).

On 10 May, League launched a social media game with a promotional video in which Salvini declares: “They are all against us, big newspapers, big professors, big intellectuals, analysts and sociologists, but we use the internet… And we win online.”

By contrast, left and liberal-centrist leaders in some countries that we studied did not use hashtags at all, or simply placed them haphazardly without an apparent understanding of their utility.

Our model weeded out all but the most obvious bots in just those five countries, and still found more than 700 bot-like accounts that we can say with the highest confidence are indeed bots. In a separate study conducted in March, SafeGuard Cyber found 6,700 “bad actors” linked to Russian disinformation operations spreading propaganda to up to 241 million users.

While we have no way of knowing the origins of the bots we detected, it is important to note the context of Russian support for disinformation campaigns and that the parties that comprise the Movement for a Europe of Nations and Freedom do have close ties with Russia. The League and the Austrian Freedom Party have signed cooperation agreements with Moscow, and the National Rally (formerly Front National) has received funding from Kremlin-linked backers.

While the Freedom Party was recently brought down by a scandal involving trading public contracts for campaign contributions from someone posing as the niece of a Russian oligarch, League has denied allegations of trading oil contracts for support to Kremlin representatives.

The far-right Alternative for Germany, which votes with the far-right group, is embroiled in a scandal over Russian strategy papers that apparently said that German MP Markus Frohnmaier could be “absolutely controlled” by the Kremlin. Manuel Ochsenreiter, Frohnmaier’s employee, has also been implicated in an alleged fascist false flag attack in Ukraine, allegedly orchestrated by Russian secret services, which he has denied.

These are only links of course, which prove nothing about the origins of the bots that we found, but it is useful to remind ourselves that Russia has built a powerful track record in this area.

As Europeans go to the polls on Thursday, they will be making important decisions about the direction of the future of supranational institutions like the European Union. Far-right parties focusing on anti-immigration politics as an antidote to the crises of neoliberalism hope to make big gains against those seeking a more inclusive, left alternative or a safe return to the status quo.

In the midst of these important issues, information warfare is being waged in support of an increasingly online and internationalised far right fighting to tip the scales against a united Europe and in favour of nationalist alternatives.

Our findings suggest that social media providers must do more to combat these attacks, and to build users’ capacity to identify and marginalise malicious online content. The bots want to swing your vote, and you may not even realise that they have succeeded.

Emmi Bevensee is a PhD student at the University of Arizona studying machine learning and disinformation; Alexander Reid Ross is the author of Against the Fascist Creep (AK Press)​; Sabrina Nardin is a graduate student at the School of Sociology at the University of Arizona