what follows after the quack of the platform launched by the RTBF?

It is certain, the carriers of the project Faky had not imagined such a disappointment. A few hours after its launch, the platform to help fight misinformation has been suspended.

"As we tackled a sensitive issue, we expected that there would be reactions," responds Gregoire Ryckmans, one of the carriers of the project. "But we did not expect some media to find that the results were not relevant to them and that there is such a reaction, which is understandable". "We tasted a bitter pill, we will do with it and we will try to turn it into a positive result", says Georges Lauwerijs, also involved in the creation of this tool.

What happened ?

This Friday morning, the Faky tool is made available to the public. As we explained in this article, this tool proposes among other things to estimate the reliability of an information by copying / pasting the URL of an article in a search window. After a few seconds, a result falls: the reliability of the article is estimated by a score on a scale of 1 to 5. This result is the aggregation of the indications provided by tools that work on the analysis of sources and content of the article.

But Faky will quickly show his limits.

Olivier Bailly is a journalist for the magazine Medor. He decides to test several of his articles on Faky. And what is his surprise to find that they are endowed with an "unreliable" mention while they are the fruit of a journalistic work pushed.

Thequotation is particularly deadly for us

"It is particularly pernicious to end up with algorithms that emit a peremptory judgment of reliability", he reacts. "I find myself with an article that is considered unreliable without argumentation.We are a young media.The people who know us will not use Faky.But those who discover us could end up with the identification 'Medor = unreliable "We have no explanation for this judgment, I do not know how the algorithms organize the listing but it is particularly deadly for us and any other young media".

Recourse to Faky must not prevent us from being critical of the results. In this case, this critical spirit has been exercised and Net surfers as journalists did not fail to point out surprising results. A horoscope deemed reliable, articles deemed unreliable out of the ordinary, different results for similar information depending on the media that publishes them, etc.

Internally, writing is almost nothing more than that. Nobody wants the serious work of colleagues in other media to be unjustly pinned. "It's catastrophic," we hear here and there.

Why were a priori reliable articles considered unreliable?

Why have Medor's articles been so poorly quoted, when we know the usual seriousness of the work of this magazine? Elements of explanations.

"We have five tools normally in the platform but they do not work systematically for all the contents", explains Gregory Ryckmans. These are Decodex, Decoders, Disinformation Detector (developed for the project), Textgain and Neutral News. However, several tools did not deliver any results. "The Decodex of the World has not referenced Medor as a reliable source because it works mainly on French sources and some mainstream sources in the French-speaking world of Belgium." Medor has also gone under the radar of the Neutral News tool. "It takes into account the traffic generated by the site and SEO, but Medor is not strongly referenced." It must be said that the site was created very recently, in September. His recent creation impacts the perception of his credibility.

In fact, only one tool has turned, the one that "detects elements of subjectivity in articles."

Surprise of Olivier Bailly: "I have in mind my paper on hunting, at no time do I speak in 'I' … I do not do subjective journalism, I refer users to my work, that they judge the relevance of the work done".

For each amazing result obtained for articles from other titles, different explanations can be made. These are, for the creative team, as many ways to improve the tool.

A test version, to improve

The project team directly drew conclusions from the critics that reached them. The word "beta" has been added to the home page of the site, for clarity on the unsuccessful side, under improvement, of the tool. Channels of change were mentioned in the wake: better referencing of Belgian news sites, adaptation to bring when are submitted articles of opinion or investigation, display a warning in relation to the number of tools actually put implemented according to the articles, better explanations on the objectives of Faky, etc.

Si had waited for a perfect tool, it was never launched

The released version was a test version, which -visibly- did not appear clearly enough. But why not wait to present a more powerful tool? "If we had waited to have a perfect tool in relation to this theme, we never launched it," responds Gregory. "The idea is to launch it, even imperfect, and to have feedback, feedback from users and the press to improve the tool". For Gregoire, there were also other relevant things in the platform (fact checking, image search) that were more developed. "Searching via the URL was the trickiest thing."

We learn from his mistakes

How was Faky tested before being made public? Has it been sufficiently tested? "We sent the link to all RTBF editors to ask everyone to test this tool", explains Georges. "It is true that we had few returns, we should have said that we did not have enough feedback to validate its effectiveness, what we did not do … We learn from his mistakes. We will relaunch a battery of tests to ensure a more reliable result. "

The decision to suspend public access

Initially, it was a question of working on the improvements to make while leaving the tool accessible to the public (with more explicit mentions on its evolutionary character and its objectives). Finally, decision is taken to temporarily suspend access to the public.

"When there is a bit of confusion, you have to take a step back, especially as a public service media, when you see that the public might not understand our purpose," explains RTBF spokesperson Axelle Pollet. "Some decode that the RTBF is a prescriber, puts itself in the position to say what is good information and what is not, which is not the case.It is really a tool This is an RTBF initiative but it is driven by a set of partners and so again, we were a bit weak or too optimistic about how the audience would understand it. " Hence a desire for greater clarity in the future in the intentions. Even if one wonders whether this will be enough to answer all the reservations already expressed.

Is this the right way to fight misinformation?

Beyond the defects of the test version, some question more generally about the relevance of this type of tools to fight against misinformation. Not to mention the errors that would escape anyway to this type of tool, like the widely shared about Xavier Dupont Ligonnes (including by RTBF, we talked about it here).

Is it not utopian to think that a tool of this type will one day become sufficiently powerful to avoid being open to criticism, in a field as complex as that of information? Georges continues to believe in the relevance of the tool: "I think there is not a lot of work to improve it very seriously, there are some very concrete things to do to avoid aberrant results," believes there. While stating that we must take Faky for what it is: a help. "Faky does not say if an info is false or true: Faky is a tool in the field of media education, which facilitates a little work but does not prevent critical thinking, on the contrary: it leads to to awaken the critical spirit ", He explains. "The idea is to give keys, to be transparent about the approach and the tools used and to leave the critical mind to the user", complete Gregory.

This approach of critical thinking is also that of INSIDE, in writing. These two projects are also a little cousins ​​because they are the result of a common reflection about the quality of information today and the link with the public, while offering very different angles of approach .

---

►►► This article is not an info article like the others … About the INSIDE page of the editorial, daily news reporters take the pen – and a bit of hindsight – to reveal the backstage of the profession, answer your questions and reflect with you on their practices. More information : the. And for your questions about our info processing: it's here.

---