How people, technology help Facebook fact-check posts for accuracy
Facebook, which has also been facing flak over allegations of political bias, works with independent third-party fact-checkers certified through a non-partisan international fact-checking network. In India, the social media company works with eight partners including India Today Group, Vishvas News (Jagran), FACTLY, NewsMobile, FactCrescendo, BOOM Live, AFP, and the Quint, which fact-check in 11 Indian languages and English.
“The minister has made a blanket statement and has not targeted anyone in particular,” Rahul Namboodri, the co-founder of FactCrescendo, said.
“IFCN has very stringent methodology and a completely non-partisan outlook which we follow,” Namboodri added. “The fact-check is done solely on evidence and not on opinion. We tackle misinformation, irrespective of ideology.”
HT reached out to BOOM Live and India Today Group; they declined to comment.
Facebook told HT over email that the fact-checking system is a hybrid between people and technology, adding Facebook’s machine-learning models identify posts that may be false by using signals to check what’s going viral, user reports, and comments expressing disbelief. Facebook also allows several rating options that third-party fact-checkers can apply to content.
FACTLY founder Rakesh Dubbudu said the role of a fact-checker is limited to assessing misinformation and Facebook has to take further action. “Our process is simple. We ensure that when we do a fact-check, everything is completely transparent and the public can repeat the same actions to confirm the information.” He said they focus on virality and potential for real-world harm when it comes to flagging the posts.
According to a fact-checker, who did not wish to be named, if the right-wing ecosystem is throwing up more misinformation, then how is their fault. “Ahead of the Karnataka elections in 2018, we had also pointed to various pages that were spreading Congress propaganda.”
People familiar with the matter said the fact-checking system is a hybrid between people and technology. They added Facebook’s machine learning models identify posts that may be false by using signals like how something may be going viral, user reports, and comments expressing disbelief. Facebook also allows several rating options that third-party fact-checkers can apply to content.
Content that has no basis is marked as false. Image, audio, or video content edited or synthesised beyond adjustments for clarity or quality in ways that could mislead people is marked as altered. Content that has some factual inaccuracies falls under the partly false category, etc.
When fact-checkers rate a story as false or altered, Facebook significantly reduces its distribution. Pages and domains that repeatedly share false news also have their distribution reduced and their ability to monetise and advertise removed. Facebook users are also allowed to mark content as false or misleading.
Facebook’s strategy to tackle misinformation is three-fold, the people said. It also involves the removal of accounts and content that violate community standards.
The Parliamentary standing committee on information and technology on Wednesday summoned Facebook officials following media reports that an executive from the company’s India operations team intervened to protect a ruling Bharatiya Janata Party politician from action over hate speech.
The Congress has sought a joint parliamentary committee investigation into the allegations against Facebook, a criminal inquiry into the company’s conduct, and termination of its contracts with the government.