Social media giant Meta has announced it will implement an Elon Musk-inspired solution on its platforms Facebook, Instagram and Threads to combat misinformation.
The end of its fact-checking partnership will begin in the United States first, with the possibility of it being extended to other countries after the social media company examines regulatory requirements.
Rather than pay independent fact checkers to flag and debunk misleading posts, Meta founder Mark Zuckerberg said the company would introduce a different model already being used by rival platform X.
Community Notes, formerly BirdWatch before it was rebranded, is a feature on X that relies on crowdsourcing to flag potentially misleading content that became widespread in 2023.
Mr Musk has praised the program as "awesome" and he applauded Mr Zuckerberg after the announcement.
So what is Community Notes and is it effective?
How do Community Notes work?
Community Notes are breakout boxes that appear underneath a post on X that has been flagged as misleading or false. These notes are written by platform users who are eligible to sign up for the program.
The note appears after users reach a majority a consensus on its accuracy and it includes a correction, often accompanied by a link to an online source supporting the fact check.
Any X user can sign up to write a Community Note, so long as their account is at least six months old, they have a verified phone number, and no "recent violations" of X's rules.
The notes are only shown on a post if they are rated helpful "by enough people from different perspectives", which the company says "prevents one-sided ratings".
The company says it defines these different perspectives based on a user's history of rating notes, though it is unclear how new users without much of a digital footprint are assessed.
X says it does not write or moderate any notes unless they break platform rules.
The company also says a post being flagged with a note does not disrupt its reach, in contrast to Meta's fact-checking program, which de-prioritises content rated false.
Posts with a Community Notes correction are not eligible for the platform's ad revenue sharing, which Mr Musk has said is to "maximise the incentive for accuracy over sensationalism".
In its announcement, Meta directly referenced X's approach, including the requirement for the agreement of people from "diverse perspectives" and the end of throttling the reach of fact-checked posts.
The company also criticised its current program as one that "intended to inform [that] too often became a tool to censor". It attacked the integrity of fact-checking groups, claiming they were "too politically biased" and had "destroyed more trust than they created".
It also indicated Meta would be "simplifying" its policies around content in the coming months, including removing restrictions on topics "like immigration and gender".
Mr Zuckerberg said in his announcement video the election of former US president Donald Trump to a second term presented the opportunity to "restore free expression".
Do notes stop misinformation?
Whether Community Notes are effective on X depends on who you ask and what you're measuring.
Academic research has found that the program has been effective at countering vaccine misinformation, encouraging users to retract false or misleading posts, and improving users' ability to identify misleading content.
But there are significant limitations too.
Tim Graham, an associate professor in digital media at the Queensland University of Technology, said the Community Notes process was slow and misinformation was allowed to spread while the notes were being written and debated.
"The damage is already done in an hour or two, once you get into five hours, a day, two days, everyone moves on," he told ABC NEWS Verify.
Dr Graham also noted the program's commitment to only show notes rated helpful from a diverse range of perspectives made it less useful in the context of political claims.
"The issue with crowd-source fact checking for political content is not only does the system fail to get consensus most of the time, but a lot of the claims that are made in the posts by nature can't be verified," he said.
He said time was often wasted debating whether content was an opinion or whether a claim was even verifiable, citing the claim "big governments don't work" as an example of when users got lost trying to prove or disprove a proposition.
He has observed "massive wars" occurring in the notes.
"There are so many claims where people don't even know how to fact check them," he said.
This can lead to many posts that might contain misinformation going without a note because of a failure to reach consensus. Other studies and reporting have found that this makes the program much less effective at stopping misinformation.
While research suggests notes can be effective in countering misinformation when they are present, there is still the issue of visibility.
Analysis from The Washington Post in October using X's Community Notes data found only 7.7 per cent of proposed Community Notes ended up being shown on the platform that month, down from a peak of just over 11 per cent in January 2023.
The masthead highlighted several misleading X posts across several posts that had gone without a note because Community Notes users could not agree.
The Post's findings are supported by research from the Centre for Countering Digital Hate, which found that 74 per cent of accurate community notes on a sample of 283 misleading political posts between January and October last year did not get shown, amassing 2.2 billion views.
The analysis also found that among a sample of 20 misleading posts that did receive community notes, the posts received 13 times as many views as the notes attached to them.
And joint research published by the University of Luxembourg and the University of Melbourne found that even though there was an observed increase in the volume of fact checks conducted via Community Notes, particularly against verified users with large followings, there was no evidence it reduced engagement with misleading tweets.
In simple terms, telling people something was wrong did not stop people from spreading it.
Dr Graham said there were surprising instances where the notes worked, but the design of the system — consensus from users with diverse perspectives, often polarising — was flawed for certain types of content.
This includes political content, parody, borderline speech and hate speech.
"Community Notes is billed as a panacea, as a cure-all, a kind of Wikipedia solution to fact-checking … but when you get into the nitty-gritty the system fails to get a consensus most of the time.
"[Consensus] is a fundamental misreading of truth and how fact checking works."
