Bird watching: What’s going on with X/Twitter’s Community Notes?
It’s always embarrassing to be called out on a lie.
That’s the cruel genius behind Twitter’s Community Notes feature, which brutally exposes mistruths with a cool, simple fact box. As Labour’s shadow foreign secretary David Lammy recently found out, the hard way. He tweeted “As I said on Radio 4 Today this morning, it is morally wrong to bomb refugee camps”.
But a ‘community note’ soon appeared under his tweet, informing users of Lammy’s full radio comment: “It’s wrong to bomb a refugee camp but clearly if there is a military objective it can be legally justifiable. It’s for Israel to explain its actions.”
Countless other examples abound of people being called out on untruths – many of which have collated on the account aptly named ‘Community Notes Violating People’ (some are crude: a man claiming women can’t orgasm before science corrected him; Taylor Swift’s pubic bone being pointed out – some are serious like Lammy’s lie).
First introduced in 2021 by Twitter to fight misinformation, Community Notes was hatched as Birdwatch (a “creepy” name, according to Elon Musk). It became available to all users in the US just before Musk’s acquisition of the social media platform in October 2022.
The billionaire, who bought the platform known as X last year, has called the widget a “gamechanger”. Under the (boring) new name Community Notes, the feature was expanded out of the US to 15 other countries.
Presumably this was part of Musk’s plan to make X “by far the most accurate source of information about the world”.
Since the Tesla billionaire has taken over, the social media site has been subject to increasing criticism over problems with its algorithm, the prevalence of fake news, and the re-reinstatement of previously banned accounts. All of this has accumulated to a valuation of $19bn, in comparison to the price Musk paid for it – $44bn.
X has all but abolished its teams that oversaw content moderation and ‘trust and safety’. In this context, community notes has actually done some good, and even won a few fans among the tech analysts.
How does it work?
Rather than having an oversight board – “X doesn’t write, rate or moderate notes” it proclaims online – the notes are approved by an insanely complicated algorithm assessing volunteer users’ ratings of suggested notes.
This has been cleverly designed to prioritise notes that receive positive ratings from users across a diverse range of perspectives. Any users registered as a “community notes member” are then given a rating, based on their public political leanings, so that only notes sanctioned by people from across the spectrum are eventually published. It works this out by analysing users’ tweets.
This is a way to save money by “outsourcing moderation costs to users [allowing Musk to] avoid paying professional fact checkers by getting the punters to add context and vote” according to Joseph Teasdale, head of tech at Enders Analysis.
However, overall Teasdale “likes” community notes. He told City A.M. notes are “a nice middle ground between doing nothing about false or misleading posts and simply removing them.”
He pointed out the “problems with a central platform applying content policies designed on the West coast of the USA to products that people all over the world rely on for information”.
However, there are definitely imperfections.
The issues
- Manipulation
Wired recently published serious allegations of manipulation amongst Community Notes users. A contributor told the magazine they, along with around 25 others, had been able to manipulate visibility of notes by upvoting and downvoting.
“We have a group so we can coordinate who writes what Community Notes, and if you write one, you can coordinate so you get one voted as helpful, and there’s no validation on any of that,” the source told Wired.
In this case, the members were part of a group of online volunteers trying to fight Russian disinformation in the aftermath of the invasion in Ukraine.
“I’m sure community notes are open to abuse, if enough users organise against a given tweet or opinion,” Teasdale agreed.
Musk recently announced that sources would now be required – to put a stop to these shenanigans. However, the guidelines around what constitutes a source appear to be murky.
Musk has said: “We will have to watch this one. Links to actual source data, not some bs press article, are what matter.”
This system is similar to how Wikipedia works. Meta by contrast has a global team of over 15,000 reviewers “keeping Facebook safe”, plus stakeholders informing community standards.
- Lack of transparency
Despite X saying “anyone has free access to analyse the data, identify problems, and spot opportunities to make Community Notes better”, the actual algorithm is opaque.
Some sources have flagged in-fighting and disinformation within the company and there appears to be no real oversight from X.
- Visibility
The largest problem appears to exist around visibility. Sixty per cent of the most rated notes have not been made public, according to Mediawise, an organisation run by the Poynter Institute.
Nearly 150,000 notes have been filed but only 10,000 or so have been seen by X users.
There are nearly 133,000 Community Notes users.
Another aspect of this is speed. In the wake of the Israel-Hamas war, Musk has been under fire for “slow” fact checking prompting X’s owner to pledge greater speed.
The European Union recently launched an investigation into whether X has failed to comply with the new Digital Services Act by allowing graphic illegal content and disinformation linked to Hamas’s attack on Israel to swarm the site unchecked. It is the first time the EU has implemented this mechanism.
Teasdale was more charitable to the platform, saying: “I can’t judge how community notes have fared recently, but I will say that when an event like the Israel/Gaza war occurs, there is no informational product that is going to go uncriticised.”
“People don’t just have differing opinions about it, but live in totally separate worlds of facts,” he said.
So who can apply?
Notes are written by Twitter users. That means anyone can apply to become a Community Note member – as long as your account has existed for six months, you’re willing to hand over your mobile phone number and agree to some very Brownie-esque guidelines.
All this was such a tempting offer City A.M. had to put some names down right away.
The process wasn’t smooth: the mandatory text message confirmation code didn’t deliver to one colleague’s phone for hours. Another was told she had an error with her account and that she must wait 48 hours before she could apply to join the club.
As it stands, one City A.M. employee got through the majority of the process.
After that, all you need to do is wait for X to approve you – which could take hours or weeks with the social media firm’s site saying it’s reviewing users “On a rolling basis”.
Stay tuned for our next update once City A.M. has fully immersed itself in the universe of notes users (hopefully).
Are you a member of Community Notes? We’d love to speak to you opinion@cityam.com