The alarming spread of fake news on Twitter in France

While the context of COVID-19 in France and around the world highlights the volume of false information and the importance of accessing reliable content on networks, the audiovisual gendarme published a report describing their mechanisms. It must be said that at this time of great uncertainty in the face of an unprecedented virus, social networks have paradoxically become one of the first sources of information about health. So the creator of Facebook declared in March last year: “We know thanks to previous emergencies […] that in times of crisis means of communication are used more often than usual ”.

The World Health Organization (WHO) is talking to us about their part of an “infodemic” to denote an entirely different pandemic that is informative and has a strong impact on the population. As the Freedom House report on online freedoms since the beginning of the global crisis has shown, false information, often shared by those in authority or legitimacy, is as dangerous as a virus. In early November 2020, an infodemia working group published a report with recommendations to social networks to improve their fight against disinformation.

In the same category

Snapchat launches Spotlight to compete with TikTok

In addition to these reports, the CSA provides information about Twitter. The Council has no power to restrict the network and only proposes to highlight the phenomenon of fake news and its powerful mechanisms.

A report made possible by the transparency of Twitter

An interesting aspect is that of the modalities of the existence of these revelations by the CSA. The data collected was created thanks to an API, an interface through which users can determine access to the site data. Through Twitter, the CSA was able to access public content free of charge, although the platform was relatively transparent on this point and the only one that offered such access. Facebook, on the other hand, has restricted all access to its data since the Cambridge Analytica scandal. The report is therefore based on an analysis of the content of Twitter, whose users have an obviously different profile than other networks that may have been the subject of the study. In addition to accepting people who are on average socially and culturally higher than Facebook, Twitter ranks 4th in France (9% of the population uses it) behind Facebook (43%) and YouTube (23%). , Facebook Messenger (12%) and connected to Instagram (9%), according to Reuters Institute’s 2020 Digital News Report for the Study of Journalism.

An analysis of the reliability of the information

The advice is based on the categorization of the Decodex of the World, a directory of potential sources designed to provide benchmarks for internet users lost in this deluge of content. Each news page is classified according to its reliability based on various criteria such as author, sources or misspellings. 50 disinformation or dubious locations were examined, including Valeurs Actuelles, Dieudosphere, TV Libertés, Boulevard Voltaire, l’Antipresse and also Russia Today.

“The Different Categories of Twitter Accounts and Their Specifics”. Despite a much higher number of accounts rated as “reliable”, incorrect information is retweeted much more often. Graphic: Century Digital / CSA

Categories 2 and 3, classified by Le Monde as “spreading false information” and “questionable”, have fewer accounts on Twitter than reliable websites: a 37 times lower number. However, the report shows that these accounts are very active with a higher number of retweets: an average of 19 per tweet. This explains how tweets calling for the destruction of 5G pylons can go viral in a short amount of time.

What are the most frequently shared infoxes?

Of the accounts that reported false information, the study focused on each of the 10 most viral and recent tweets. It indicates that certain issues arise. Below this is politics (contained in 46 tweets), followed by immigration and health (in 29 and 27 tweets, respectively). The study, which began before the emergence of the COVID-19 crisis, is easy to imagine that health has gained the upper hand today as the infoxes have been numerous since the epidemic began in France. Before the health crisis, anti-vaccine publications dominated and made up substantial content.

“Number of tweets from disinformation accounts by subject”. Divisive and current affairs are the most popular accounts that reveal false information. Graphic: Century Digital / CSA

The report not only reveals the most shared topics on Twitter, but also adds that this false content is not necessarily obvious, as it sometimes uses the codes of professional and reliable journalism. The language level, the quality of the photos and videos can be anything but amateur.

Infoxes come in various forms to manipulate the user who reads them: The CSA shows that divisive topics such as politics, immigration, religion or terrorism are breeding grounds for false news on Twitter. Disinformation accounts use external sources of legitimacy to influence the interpretation of data and manipulate users. These videos, images or links, mostly reliable or official, possibly from the Élysée or the world, are most of the sources of information that are clipped and modified on Infox accounts on Twitter in France.

So it’s important to check facts, says the CSA. Fact-check accounts or checking information are pillars of the Twitter network. Accounts like Factual AFP, Check News, Fact Check EU, or Décodeurs du Monde remarkably correct viral tweets and help to moderate the information to some extent. For example, the de facto AFP had corrected the allegations of the arson of Notre Dame de Paris, these allegations in particular using photos in which we saw a silhouette on the roof. As can be seen from the factual account, this silhouette was nothing more than the statue of the Virgin of the Trumeau of the monastery.

No echo chamber phenomena

In a social network, an echo chamber is a phenomenon of the repetition of information due to the accounts one follows and the network itself, the. It uses algorithms to show us what we prefer. This is also known as the filter bubble: This personalization of our “newsfeeds” is increasingly moving towards our usual decisions and beliefs. If we see the same news on a hundred accounts, we’ll end up believing it.

On Twitter, the report concludes that these phenomena were indeed very limited. It is also the specialty of the network, which describes itself as a “showcase of what is happening in the world and current topics of conversation”. Since 2016, the user can effectively add an algorithmic component to the display of tweets, which is then personalized for the content based on various criteria such as the number of user interactions. However, it is a choice: the user can also select the ante-chronological display mode in which the tweets are displayed according to their publication.

The report also showed that subscribers to Infox accounts are also 80% of a “reliable” account. The problem, therefore, is not in the existence of an echo chamber, but in the attractiveness and efficiency of creating fake messages, given that the vast majority of users have access to the correct information.

Infodemia, infobesity: the evils of social networks

The CSA study is in line with a 2018 study by an MIT data scientist, the largest ever conducted on Twitter. Having considered more than 126,000 topics (the “Trending Topics”) for more than 10 years, it contains the data of more than 3 million users, and its conclusions are final. The truth just cannot compete with jokes and false rumors. In every statistical analysis, researchers have systematically shown that fake news reached more people, penetrated deeper into the network and spread faster.

An infodemy that adds to the infobesity, this information overload that is absorbed by users. Indeed, the world today creates as much data in 48 hours as it did in the entire 20th century. As early as 2013, the Economist magazine report “Daten, Daten, everywhere” showed the growth in the volume of data available on the Internet and introduced the yottabyte for quantification, a measure that cannot be assessed due to its astronomical evaluation.

One fundamental problem concerns our social networks, as all of these studies show, and particularly that of the CSA, which focuses on Twitter. “In the case of the topics with the highest volume of tweets, incorrect information predominates. And studying the timeline shows that, contrary to hope, the real information does not drive away the wrong. Some issues die out before the truth is told, ”stresses the Council.

In an essay published in Science in 2018, a group of political scientists and lawyers called for “rethinking our information ecosystem in the 21st century.” Accordingly, many pathologies reach social networks, of which infodemia or infobesity are just examples, and a new impetus of interdisciplinary research is necessary to reduce the spread of false news. A dynamic that the CSA could have intensified despite the limits of its powers vis-à-vis the giants of social networks.

Report Rating