Journalistes Solidaires

We're a fake news debunking factory. Now we want to tackle them before they get viral.

Comment 8

Inspiration

A huge amount of COVID-19 centric news are published each day. Some are true, some are suspicious, some others totally fake. In the era of social media, it's proven very hard to be a well informed citizen. We created A collective of 60+ journalists, developers, rewriters, correctors, project managers and more willing to fight fake news.

We built a 100% virtual and open newsroom where everyone is working together towards the same goal, under the same values : Expertise, Credibility, Transparency and Agility. Anyone can ask us to verify a suspicious publication. Everyone can see how we do it.

What it does

We build an infrastructure that:

  • organises efficiently people in task-forces to verify specific publications/claims
  • worked on 119 submissions from citizens
  • ran 64 investigations
  • published 26 articles to state if an information was true, false or if it was too soon to tell
  • produced 7 videos from those articles
  • is now composed of more than 50 journalists and students**, experimented fact-checkers, health specialists, developers and more
  • publishes on a daily base our articles and videos on the main social networks with a strong community to amplify our work
  • has a website audience close to 1,000 visitors per day.
  • covers fake news in France, Belgium, Germany, Greece and more.
  • have 26 ongoing and investigations open for anyone to see

During the hackathon, we launched checkmy.news an English language version of JS.

How we built it

Journalistes Solidaires was built using citizen expertise and online tools like Airtable, Integromat, Protonmail,Armada.digital, Wipster, Pytorch, Discord and more

We also could rely on skilled professionals from the media field from Belgium, France, Finland, Germany, Italy, Switzerland and more.

We designed and set up an Airtable environment where our journalists can store all the data generated during their investigations. This collaborative tool can be entirely tailored to our specific needs and serves as a place to store and classify dubious publication (for active work and later research), as well suspicious sources of information as building video script.

We linked Airtable to a custom-made EtherPad solution (open-source collaborative edition tool) in order to provide our journalists a Content Management System where they can write publications using Markdown syntax.

Our Airtable is also linked to our custom-made website where our readers can follow in real time live the investigation of our teams by joining the Open Newsroom. They can also participate in the investigation if they have information using the form included in each investigation page.

After having built a strong dataset of fake news spreaders we are now working on a tool who could help our investigation teams to work on a topic before it spreads.

Challenges we ran into

We faced a lot of challenges on our journey from moving to a simple Facebook group idea to a 7/7 working newsroom producing daily fake news debunk.

  1. Finding the right organisational model Our first and main challenge has been to be able to bring people from all over Europe and the Francophonie working together in a single environment. Give them all the tools they needed to work independently from the situation. We moved all the lines and the processes, creating our own guidelines keeping in ming our mission and our values as equity, transparency and agility. But we could disrupted the usual newsroom, investigation and diffusion processes in order to make them fits into our needs. We are a 360° newsroom thinking social media, article, video, audio productions at the same time.

  2. Accept and have everyone accept to open their work to others We were willing to not only give information but also help people to understand how we are able to say if a news is true or not. We needed to open ourself, open all our weakness and create from them our strength.

  3. We still have a lot to achieve but have the motivation of over 70 journalists, experts and our each day growing community of citizens pushing us to create better tools, better process to tackle the fake news problem.

  4. Our Economic model requires some work. Given the emergency of the situation, we secured partnerships with our technical partners which agreed to give us access to their services for free for a limited time. We also applied for grants to anticipate the need to pay our contributors. One other option is for us to become a service provider for other medias. In a B2B relationship, we would offer debunking services for a fee.

Accomplishments that we are proud of

We are proud of having build an entirely virtual newsroom in less than a month starting from nothing except of our knowledge. Our newsroom is currently producing a debunk per day and handling more than 25 investigations in the same time.

We worked several times with established media companies in France such as France Info and building partnerships with them to spread in more channels our content or produce new content with them.

We build an entire video production department in a month able to write and produce social media video adapted format in a day or less.

We created an Open Newsroom where we are totally transparent on our processes. Every citizen can follow in real time what our investigators are doing while keeping for them and our investigations all the necessary privacy.

We build a democratic organisation where everyone is involved applying the best of the class models.

We are not fearing to fail, we are learning everyday and are able to adapt our organisation in order to improve ourselves.

We are united as journalists, experts, developers, citizen to have an impact.

What we learned

We confirmed something we already strongly believed in: all united across borders, there is no limit to greatness. Who couldn't have imagined 6 weeks ago that we would have reached such milestones?

We learned that tackling fake news is not only a journalist's work. It requires data scientists, developers, psychologists, product designer, project managers and even more.

There is no such thing as "one absolute rule" to follow or robot to build, we need to work as a united organisation to tackle this issue properly.

Whatever the algorithm we use, we'll always need journalists to work on what comes out of it. Whatever the amount of journalists are dedicated to fake news verification, they will not be able to handle the amount of publications without automation.
Whatever the amount of contacts we have, it will always be smaller than the power of a community.

We learned to work united.

What's next for Journalistes Solidaires

  • Getting up to speed with our machine learning model and make it readily sharable to other teams so that fake news are contradicted and blocked before they reach the masses (more below).
  • Transposing our model in other countries, and other languages: we can quickly duplicate our processes and teach our methodology
  • Creating an English language corpus of debunks across the continent, directly created with by our national teams, also for researchers to use

We came up with the idea to try to tackle fake news at the source, before they spread instead of trying to contradict them one it has reached a large audience. The idea of the tool we are willing to develop would be to help our newsroom to maximise the chances to debunk fake news as early as possible so that the truth and well sourced content are spread before fake news.

So we would like to form a team with Machine Learning / Natural Language Processing specialists, behavioural science specialists in addition to our existing team to build a model that can:

  • crawl our own set of known sources and discover new one to aggregate their content and analyse it. Most of the content will be in French and English (for a start). They can be as varied as 4chan, Facebook groups or profiles, Twitter or websites. The bonus would be the analyse of videos from YouTube or others sources
  • recognise potential fake news through the dataset and identify patterns of semantic elements already used in other fake news
  • sort them by sources (which would have untrustworthiness scores
  • identify distribution patterns "Super-distributor" profiles on social media

The tool should be able to notify our newsroom by itself when a piece of misinformation is susceptible to get viral so that we can work on the topic before it spreads. This notification is a simple URL to be called by the tool with params (Discord hook).

We bring in this project a large set of suspicious sources, in French and English, and the knowledge of our journalists in the fake news creation process and their spreading. We can also help to get the ressources needed to have our tool running quickly. Lastly we can bring a large set of fake news debunking in French.

Try us !

French version: Journalistes Solidaires ¦ English version: Check my .News

Built With

+ 224 more
Share this project:

Updates