Oxford study raises alarm on 'junk' news in France
Fake news is playing a massive role on Twitter in the run-up to French and German elections, a British study has said amid concerns on Russian meddling.
The study, by the Oxford Internet Institute, part of Oxford University in the UK, out on Friday (21 April), said that one in four political news stories being circulated by Twitter users in France in the run-up to Sunday’s presidential election were deliberately false or “junk” articles that voiced “ideologically extreme, hyper-partisan or conspiratorial" points of view.
It said that the run-up to German presidential elections in February “found Germans sharing four professionally produced news stories for every one piece of junk”.
It noted that the problem was less acute than in the US elections, where junk news, for instance in the state of Michigan, at times accounted for one out of every two links shared.
It also said that the bulk of France’s junk traffic targeted Emmanuel Macron, a centrist, pro-EU, and Russia-critical candidate, but that “highly automated accounts” also circulated large amounts of traffic about Francois Fillon, a centre-right and Russia-friendly contender.
The Oxford study comes hot on the heels of a similar one by Bakamo, a private-sector internet research firm based in the UK.
Bakamo said on Wednesday that 19.2 percent of links shared by users of social media in France in the past six months pertained to articles that did not “adhere to journalistic standards” and that expressed “radical opinions … to craft a disruptive narrative”.
It said a further 5 percent related to “narratives [that were] often mythical, almost theological in nature” or discussed “conspiracy theories”.
The bogus news mostly favoured anti-EU or pro-Russia candidates, including also far-right leader Marine Le Pen, radical left candidate Jean-Luc Melenchon, the self-described "Frexit" candidate Francois Asselineau, and Trotskyist candidate Philippe Poutou.
It said one in five of the “disruptive” stories had been influenced by Russian media and that one in two of the “conspiracy” articles bore Russian fingerprints.
Two of the Russian state’s biggest disinformation outlets, RT and Sputnik, both have dedicated French-language services.
Research by the Atlantic Council, a US-based think tank, out this week said that “much” of RT France’s coverage was “adequately balanced”.
“[RT France] has tended to report criticism of Macron, but at least mentioned the candidate’s own stance, albeit often briefly”, it said.
But it said Sputnik France showed a “distinct bias” against Macron and in support of Le Pen and Fillon.
It claimed, for instance, in five articles in February and March that French broadcaster BFMTV had given more airtime to Macron, whom it dubbed the mainstream media’s “darling”, than to rivals, even though the French media regulator, the CSA, said this was not true.
It also claimed in a series of articles in March that polls were predicting a Fillon victory even though most polls were not.
It based the reports on figures published by Brand Analytics, a Moscow-based firm, whose methodology was said to be unrepresentative by another French regulator, the Commission des Sondages.
The Atlantic Council said in an earlier report that RT France and Sputnik France had just 102,000 Twitter followers compared to mainstream media, such as the AFP news agency, which had 2.6 million.
It said their users were much more active, at almost 5 tweets per follower in a sample period compared to 1.5 tweets per follower of other media, and had a “strong political bias” that was broadly pro-Russia and anti-Macron.
It also said Sputnik France’s output was being circulated by automated “bots”, or Twitter accounts that gave no personal information on the user and that tweeted more than 100 times per day.
It said almost one in three (28%) of Sputnik France followers were of this type.
It also said the most prolific bots RT France or Sputnik France bots, such as @RTenfrancais or @heelleclech, were tweeting or retweeting content an average of 392 to 777 times each day.