Fake news failed to rock French election

President Macron has unveiled plans to tighten rules on what the media can report during election cycles. Lindsey Johnstone reveals the history of fake news that has led to that decision

Published Modified

On Friday, May 5, 2017, two days before the final round of France’s presidential election, and four minutes until the legally mandated media blackout came into effect, a statement from Emmanuel Macron’s En Marche! party landed in newsroom inboxes.

It announced that, earlier that evening, 9GB of emails and documents from his election campaign had been posted to a document-sharing site.

Mr Macron’s En Marche! said it had been a “victim of a massive, coordinated act of hacking”. Emails, documents, accounting files and contracts were included in the information that was published.

At the time, campaign officials said real documents had been mixed with fake ones to sow “doubt and misinformation” and that it was a clear attempt to undermine Mr Macron.

Fake news had already affected both the EU referendum in the UK and the US presidential campaign, so it was no surprise that the French presidential election was likewise a target.

What is perhaps surprising is how comparatively little of it there was in France.

The Computational Propaganda Project at the Oxford Internet Institute investigates the spread of online misinformation by automated scripts – known as ‘bots’.

Researchers studied tweets posted during five recent elections – in the US, Germany, France and UK – and found that the ratio of real stories to fake was 1:1 in the US, 4:1 in Germany and the UK, and 7:1 in France.

Figures from social network tracking service NewsWhip paint a similar picture – their analysis found that 72 of the 200 most-shared stories in the run-up to the US election came from proven false news sources. The equivalent figure in France was 20.

The Computational Propaganda Project’s Lisa-Maria Neudert said France seems to have been “less susceptible” to false stories.

She said: “This was demonstrated by the Macron email leaks – although they were only released [two days] before the second round of the elections, we could still see that people were really not all that interested.

“They didn’t really care, it didn’t sway people.”

According to the project’s findings, manipulative hacking and fake news are on the rise – although Ms Neudert is quick to disown the label.

“The use of the term ‘fake news’ has increasingly become contentious with people like Donald Trump using it to discredit whatever kind of content they dislike.”

Her study shows that during the US elections, 20% of Twitter traffic was generated by social bots, with Trump bots outnumbering Clinton bots five to one. During Britain’s EU referendum 1% of accounts drove nearly a third of traffic.

So why the discrepancy in France? High consumption of traditional news media is not a factor in France. Per head, the French read fewer newspapers than the other nations mentioned in the Reuters Institute for the Study of Journalism’s Digital News Report 2017.

Its figures show that only 25% of French respondents read a newspaper. This puts them behind the US (26%), Germany (34%) and the UK (41%). France’s neighbours Switzerland, Belgium and Spain also returned significantly higher figures.

But online news readership is much higher, with 71% of people regularly consuming digital news. Facebook is a common news source in France with 43% of people using it regularly. The figure is lower than the US (48%) but ahead of Germany (25%) and the UK (29%).

Ms Neudert believes the French online news habit could well be making people more savvy.

“People who are used to spending time reading news online will be far better equipped to distinguish between what is a real Twitter account or a fake one, an accurate news story and a false one.

“So the smaller ratio of computational propaganda online in France could well be partly because they were more able to spot a bogus story, and therefore less likely to give it any currency. Digital literacy is a key component.”

In addition, she said that the news France is reading online is dominated by established news organisations. “Tweets from the big traditional outlets were dominant throughout the campaign – there was a much higher proportion of tweets from established media, as opposed to junk sites than, for example, during the US election. The fact that there was such a high volume of content from established news outlets being spread suggests people were relatively well-informed.”

The 2017 Reuters report agrees, showing that mainstream media generated more than seven times as much activity as digital-born media during the election. It also reports Twitter tends to play a broader political role in France.

Ms Neudert said: “In addition to serving as a platform for news distribution, political campaigns use it to mobilise support and shape political discussions.”

The first electoral debate saw the highest amount of news-related activity on one day in Twitter’s history, with more than 126,000 of those tweets posted by or mentioning traditional media outlets.

Despite this dominance, trust in traditional media in France is among the lowest in Europe, with figures from the Reuters report putting it at just 30%. This low figure is often attributed to the combined factors of France’s similarly low level of trust in political institutions, and the fact that the government subsidises newspapers.

But this culture of suspicion does not appear to have produced a preference for ‘alternative’ sources. On the contrary, French scepticism seems to be applied across the board. Ms Neudert said: “France’s trust in mainstream media may be low, but that doesn’t mean their trust in bogus news sources wouldn’t be even lower.”

One factor that differentiates the French election is the sheer number of fact-checking services that have sprung up to address this mistrust.

Fact-checking is long-established in France. Libération newspaper launched its Désintox service in 2008, and was joined by Le Monde’s Les Décodeurs a year later. Similar offerings from the state-owned news network FranceInfo and radio Europe1, plus private service Facta Media, and Les Observateurs, from France24 followed – leading former Prime Minister Alain Juppé to last year complain about the “mania for fact-checking that has overtaken France’s media”.

Concerns prompted the addition of two more to the roster ahead of the 2017 Presidential election – Le Monde’s Les Decodex team focused on fact-checking French news websites, and the collaborative project CrossCheck worked with 37 media partners over the election campaign including AFP, L’Express, Nice Matin, Ouest France, Buzzfeed, Google and Facebook.

Ms Neudert warned, however, that fact-checking to dispel digital propaganda could be counter-productive – “Some studies show that fact-checking rewards content with additional attention in the media space.

“What is more, misinformation travels in different social networks and with different dynamics as compared to a debunking message.”

But CrossCheck’s Aimee Rinehart says independent research commissioned by the organisation shows that it was successful in building trust.

“One of our hypotheses going into the project of the French election was that if we have five, six, seven organisations who have verified something, does that build trust? And our research shows that it did.

“But then France got an election result that the majority of people could live with. Had [Marine] Le Pen won, people might have declared CrossCheck a total failure.”

But for her, what stands out as a possible explanation for the lesser impact of the bots in France is the election process itself. “Their election period is much shorter than in the US – if France had an 18-month campaign they might have seen a lot more misinformation.

“The other thing is the media blackout. When the Macron email leaks were released, the news organisations just weren’t able to cover it. These things help, in terms of [limiting] spreading misinformation. These might be good measures for other countries to look at.”

At the time of the presidential election, France seemed to be fertile ground for fake news – with a series of terror attacks fuelling Islamophobia and anti-immigration feeling, lack of trust in the political system, high unemployment, and long-standing tensions in economically deprived, multi-ethnic suburbs. And the campaign came on the heels of two misinformation-heavy campaigns in the US and the UK.

But it could be argued that these circumstances allowed France to mitigate misinformation more effectively, by arming them with one essential weapon in the fight against fake news: preparation.

This was a situation they could see coming. For the bots, France was ripe for the picking – but France was ready.

A study of tweets posted during recent voting showed:

When Donald Trump won the 2016 US election the ratio of real stories to fake was 1:1

UK Prime Minister David Cameron, pictured above, quit after the 2016 Brexit referendum when the ratio of real to fake was 4:1

Emmanuel Macron won the 2017 presidential election in France when the ratio of real news to fake was 7:1