Project Comprova, a collaborative fact-checking initiative in Brazil, has abandoned the labels "false," "misleading," "satire," and "proven" in its verifications and adopted a broader approach to combating misinformation.
José Antonio Lima, the project’s assistant editor, said the change was motivated by the understanding that the previous approach focused too much on the content itself and neglected other essential components of misinformation, such as where it came from and what makes it credible to so many people.
"We realized that simply refuting allegations is not enough," the journalist told LatAm Journalism Review (LJR). "Fact-checks now investigate not only the evidence and claims of viral posts, but also the creators of the content and their interests, as well as the tactics used to persuade people and lead them to believe what was published. The overall idea is to make the content more palatable in order to contribute to public debate, allowing people to make their own decisions, regardless of their partisan or ideological affiliation, based on true facts."
Additionally, Lima said that in ending the use of these labels, they expect for at least part of the public most affected by misinformation to begin engaging with content that provides truthful information.
"The labels ended up acting as an obstacle or a barrier to the connection between verification and the public," he told LJR. "We stopped using them precisely after we concluded that people have an aversion to content that contradicts their worldview."
The changes, definitively implemented after four months of testing, represent a methodological overhaul of the initiative, which launched in 2018 and now brings together 42 media outlets from different parts of Brazil who work collaboratively.
Comprova has its own team of editors who gather suspicious content and send it to the group of fact-checkers participating in the project. Reporters from the media partners in the coalition volunteer to fact-check the content, which is typically verified by three journalists from different Comprova member outlets. After the investigation is complete, the journalists write a report that is submitted for peer review and is only published when at least three other participating newsrooms review and validate the investigation and the results obtained. The final article is published on the Comprova website with the brands of the outlets that participated in the entire process (the investigation and subsequent verification of the work done) and can be incorporated or cited by partner newsrooms in their own coverage.
In addition to the end of classification labels, a new way of thinking about fact-check headlines was also instituted. The proposal is that all headlines should be "truth-affirming" and avoid reproducing false information, even if it's to deny it.
"In the current scenario, it's difficult to imagine a public debate without misinformation. As some experts say, misinformation is no longer a flaw in public debate, but rather a characteristic that will be difficult to eliminate," Lima said. "Therefore, in practice, we are committed to reducing the harm caused by misinformation. The issue of headlines fits into this."
Reporter Gabriella Braz, part of the Comprova team from newspaper Correio Braziliense, said the biggest challenge of the new guidelines has been adapting the thinking and the way of constructing headlines and texts to explain to the reader that it is a piece of misinformation without saying that it is false.
"Before, articles would say: 'It's false that such and such happened.' Not anymore. We have to think of a much more indirect headline, in a way, because we understand that this [false] label can alienate the very audience that fell for these pieces [of misinformation]," Braz told LJR. "This whole idea of telling the truth first, of not constantly using 'false,' or 'misleading,' or 'not,' really changes our mindset when it comes to writing."
Beyond the text itself, the new format brings more reflection to the entire fact-checking process. For example, to account for all the elements that make up a piece of misinformation—who creates the content and the tactics used to make it credible—an in-depth analysis is conducted of the pages that produce the fact-checked content, the discourse used by these profiles, the way headlines and captions are created, and the content of other shared posts. All of this is dissected and can be used in the fact-checking.
"Because the misinformation chain itself is more complex today, this has required a much more reflective process from us," Braz said. "During this process, we've seen many common characteristics, such as the use of warnings, 'urgent,' content that seems alarming. These characteristics, and also new techniques we see these profiles using, can serve as a warning so that people can learn to identify [misinformation] over time."
Braz said she's also become more empathetic toward people who believe this content and has sought to understand their reasoning. She said the content she produces now is much more educational.
"Just because we're fact-checkers doesn't mean we don't sometimes glance at something and think, 'Wow, I could have missed this if I hadn't had the habit of researching,'" she added. "This is something Sérgio [Lüdtke, editor] and Zé [Lima, assistant editor] talk about: that sometimes people themselves are embarrassed about having missed it. So we can't adopt this attitude that it's false, that it's misleading. You have to be more explanatory and tell that person that you understand why they missed it, but that now they can be alert."
Researcher Taís Seibt, a PhD in Communication and Information from the Federal University of Rio Grande do Sul, said abolishing labels can be a path to dialogue and mitigate public resistance. The journalist has been researching fact-checking in Brazil for 10 years. She said that in addition to being a form of informative journalism, fact-checking is also a form of media literacy, and this channel of dialogue can be a tool against misinformation.
"Abandoning labels is a possible path. I don't know if it's precisely the future of fact-checking, but it's a possibility to mitigate resistance to correction, which is most blatantly reflected in labels," Seibt told LJR. "If you share information and someone responds with a fact-check indicating it was false, you feel exposed and may react to this exposure. The response is often to attack the correction and the correctors. There's an emotional element to this reaction, just as there's an emotional appeal in adhering to certain beliefs."
Seibt also notes that in the current misinformation landscape, there's a clearer understanding that isolated fact-checks have limited effect. This is because the misinformation narrative is already consolidated in the collective memory of people with a certain worldview. According to the researcher, it's more difficult to circumvent this mentality using fact-checking as it was born a decade ago.
"There's little point in labeling one piece of information as X; many others will do the same damage without checking. This is also why the research and practice of prebunking, which is a kind of 'vaccine,' is growing. You somehow prepare the audience for when misinformation appears, without attributing the correction to a specific piece of content," she said. "This also has to do with media literacy and the educational role of fact-checking—I would say, of journalism in general."
In this sense, Lima said that Comprova also aims to expand its ability to anticipate the spread of misinformation, among other things, by helping people identify AI-generated content. The fact-checking project also began verifying online scams in April.
"We also intend to reinforce to the public the importance of being aware of the persuasion tactics used to spread misinformation and how algorithms work," Lima added. "The goal is to contribute to creating a healthier information environment, something that depends not only on initiatives like Comprova, but also on public authorities, the private sector, society in general, and every citizen."
