Décodex: on Classifying News Sources and Fact-Checking Fact-Checkers

On February 1st, the biggest daily newspaper in France, Le Monde, launched an initiative to combat fake news and biased information sources called « Décodex ». Or at least try to.

Décodex is made of several components. It consists in a search engine, Mozilla and Chrome add-ons, a Facebook Bot and pedagogical documentation on how to be a more careful online news reader. It catalogues 600 websites and classifies them according to their « reliability ». I put the term in quotation mark, as this reliability is assessed by the team building the tool, team called Les Décodeurs. Les Décodeurs is a branch of Le Monde specialized in data journalism and fact checking headed by Samuel Laurent.

Firefox add-on: -What does it mean? A barely reliable website propagating conspiracy theories. -Is this website reliable? This website regularly features fake news or misleading articles. Be careful and look other more reliable sources. If possible, look for the origin of the information.

The tool has a straightforward – yet arguably unattainable – ambition: verify if a website is, or not, reliable. It can be used by journalists, but also by anyone reading an article online. Its functioning is very simple, perhaps even too simplistic. You, curious yet naive reader (until now) surf the web, end up on a website, click on the Décodeurs extension that will give you one of the 5 following answers concerning the visited website’s credibility:

From Top to Bottom: [GREY] – Warning, this website is not a source, or its reliability is too variable to fit our criterias. To know more, look for other sources and the origin of the information. [BLUE] – Warning, this is a satirical website that is not made to propagate real information. It is a second degree read. [RED] – This website regularly propagates fake news or misleading articles. Beware and look for other more reliable sources of information. If possible, look for the origin of the information. [YELLOW] – This website can regularly be imprecise, not giving information about its sources and not conducting regular fact checking. If possible, search for the origin of the information. [GREEN] – This website is considered as reliable. Do not hesitate to confirm the information by looking for other reliable sources or the origin of the information.

The Facebook bot version leads to the following user experience:

In many instances, you will end up clicking on the link to the documentation, and learn how to find the original source of an information, and how to cross check what you are reading. If the website is not classified yet, you can also report it to the Décodex Team.

This documentation is, to me, the most innovative part of the tooI. I had never heard of any team in a newspaper sitting down to write proper guidelines to information search and evaluation. Perhaps because there are no definite guidelines, but I find the effort legitimate and fair in the current state of the industry, as social medias are blurring the line between fake and truth. At the heart of the Décodex initiative is the will to give the power back to the people and avoid the propagation of erroneous information. As they put it themselves, the tool represents « a first step towards mass checking information ». Hence, a democratization of the ability to identify what is reliable, and what is not. However, a first critique can already be made when looking at the Décodex classification. Websites categorized as « reliable » are mainly mainstream medias ( Le monde, NYT, CNN, etc…). Is it because you are an independent blogger that you are unreliable? Not necessarily, would I argue, but Décodex’s answer to this question is yes.

Additionally, the very existence of this tool triggers concerns. First of all, many websites, especially some that are considered as being on the right side of the political spectrum, have fired back at the initiative as some were classified by Décodex as « biased » newsources. Le Figaro rightfully asked the question « who will fact check the fact checkers? », and this is, I think, important: how can one claim, as Décodex seems to be doing, that she is completely unbiased, hence has the aptitude to judge others’ biases? How can one claim her classification of reliable sources is the « right one », that everyone should abide by? What prevents other sources to build the same tool, and classify the first classifier as non reliable in retaliation?
In some instances, however, the classification can hardly be argued. The Onion, for example, is classified as being parodic: everyone knows and acknowledges that The Onion is a satirist « news source ». The editors of the publication acknowledge it themselves and that is the very essence of their editorial line. The entire « not fully reliable category », is vague on purpose and encompasses platforms from breitbart.com to Russia Today. It seems hard to find generalized and cross cutting definitions of what makes a website enter this category.

Another important problem with this tool is the following: are the people who tend to share and read fake news going to use it? As a tool built by the mainstream media that is Le Monde, and the media undergoing an unprecedented credibility and trust crisis, is the initiative going to be considered as having enough credibility? Or is it going to be judged as biased and be a subject of an even more intense criticism? Is it going to be truste or not? Can it be relied on? A large amount of articles have been released criticizing the Décodex classification. Which brings in the big question: does Décodex even make a difference?