Community Notes is BY FAR, like really really by far a better way to do fact checking than what we were subjected to previously.
Eh, not really. If enough partisan people don't like a nonpartisan fact, they can make it go away. Apparently the metric is users voting on whether facts are considered "helpful" or "not helpful." That's not how facts work. Crowdsourcing facts is not better than what we were subjected to previously. And where will these brave new influencers be getting their facts? The same media with boots on the ground experts that half the people chose to ignore?
If you're pitting Fox News, Newsmax, Joe Rogan, and the dedicated apologists for every inanity that comes out of Donald Trump's mouth universe against the mainstream media gatekeepers and fact-checking services, it's not even close. You can get some mileage out of Hunter Biden and the Steele Dossier, but pure uncut disinformation is gonna get lots of play.
The algorithm decides
According to X's website, the purpose of its so-called
bridging-based algorithm is to "identify notes that are helpful to a broad audience across perspectives."
In other words, if the algorithm finds that contributors who voted on a given note represent an ideologically diverse group, then the note becomes visible on the platform. But if the algorithm finds that the voting contributors are too uniform in their political views — a possible sign of bias — "the public never sees it," Mahtani said.
Problems with X's system of crowd-based fact-checking arise when a valid note calling out misinformation isn't rated as helpful by a diverse enough group of contributors to satisfy the algorithm, and is therefore never seen by readers. The speed at which a note is made public is also important, so that false or misleading information isn't given the opportunity to spread unchallenged.
[SIZE=1.32rem]A[/SIZE][SIZE=1.32rem] [/SIZE]
report in October[SIZE=1.32rem] by the nonprofit Center for Countering Digital Hate analyzed the Community Notes feature and found that accurate notes correcting false and misleading claims about the U.S. elections were not displayed on 209 out of a sample of 283 posts deemed misleading — or 74%.[/SIZE]
