X's Community Notes Is Spreading False Information | Hacker News Reader

X's Community Notes Is Spreading False Information

bellingcat.com

18 pointsjruohonen18 comments

Wow, so many words written about nothing. Feels like a Daily Mail article itself.

I think the goal of this post is not to understand the inner workings of Taylor Swift's entourage. It is too try and pick apart how Twitter handles misinformation since it has been a giant source of misinformation about the war in Ukraine and bellingcat is one of the few Russian language investigative journalism groups that are trying to understand and report on what goes on in Russia. By demonstrating this with an innocuous example it removes any bias in interpreting twitter's technology as it reports on the way in Ukraine, for example.

> By demonstrating this with an innocuous example it removes any bias in interpreting twitter's technology as it reports on the way in Ukraine, for example.

How does it remove bias?

It's possible for folks biased against X/Twitter/the concept of 'Community Notes' itself to still write articles using cherry-picked weak examples.

If anything when it comes to controversial and highly polarizing topics, the accuracy rate will be higher because more people would spend more effort on proving the other side wrong. So it's unlikely even small mistakes would be ignored. Whereas for random topics, much less effort will be spent.

I do not believe the current Twitter ownership is motivated to present an unbiased view of the war in Ukraine, Putin, etc.

Did you mistake my comment with someone else's?

Fair enough, maybe i was too harsh with their choice of example.

Community notes (hello wikipedia) should be read with caution, is my takeaway. As with anything, that is collected by an anonymous community.

Full title: "X’s Community Notes is Spreading False Information About Taylor Swift’s Bodyguard"

The shortened title of the submission makes it seem like it's a systemic issue, but it's about a single case, where the Note isn't even that important in the context of the post.

I do find it concerning that apparently the contributors upvoting the Note didn't check whether the source supported the Note's claim. But apart from that it's really not a very important story.

Well, it shows the pointlessness/potential danger of “community notes”, but then I think that was already reasonably well-known.

I signed up to moderate community notes, and once I was accepted I became paralyzed by the realisation that I was now partially responsible for deciding what the truth was. I'm yet to moderate a single post.

It's also quite a revelation to see behind the curtain. The discussions that take place in the moderation area make it feel like a shadow Twitter, rather than a place that in theory should be restricted to mature people focused on objectivity.

That itself is an extremely valuable experience. People often decry systems and outcomes without realizing the pressures on the people 'responsible'. You've felt that pressure, and now have a better understanding of why such systems produce negative outcomes.

I guess we need a community notes for the community notes.

Something might be wrong when your tool for dealing with disinformation is spreading disinformation.

"Community Notes" is a tool for distributed fact-checking, basically. It depends on the diligence of individual fact-checkers who participate in this game - if the participants would be like 99% bad faith actors, the algorithm wouldn't work.

Yes, something is wrong. People are wrong, some of the time, about reality. So any attempt to inform people better will misinform sometimes.

The question is whether tweets with Community Notes are more often correct than the Community Notes attached to them. If yes, then this feature promotes disinformation on net and should probably be disabled. Otherwise, it's an improvement over the status quo ante.

Since tweets are promoted by viral engagement whereas Community Notes aren't, I think it's likely that the former are wrong more often. Though probably further improvements can be made, e.g. to prevent drive-by "this is helpful" votes from outweighing higher-effort "I checked the source and it doesn't say what you said it says" votes.

It doesn't seem obviously true that that's the question? Random tweets are just random tweets, a not insignificant number of people know that they should be regarded sceptically. Community notes are seemingly authoritative, definitive, purport to present an accepted consensus. If 50% of random tweets were wrong and a mere 49% of community notes were wrong, then your heuristic would decide they're helpful but to me it would seem strongly arguable that the more convincing nature of the wrong ones would far outweigh such a minor improvement in accuracy.

the only answer to this assertion is 'of course'

We are in the era of contesting narratives, with 'disinformation debunkers' being amongst the worst culprits. All is propaganda or meta-propaganda, our anguish is coming to terms with this bitter reality.

A TL;DR would be appreciated for this one.

That sure looks convoluted.

Israeli newspaper reported that Taylor Swift's bodyguard returned to Israel to join the IDF in their fight against Hamas. X users added Community Note saying he wasn't actually her bodyguard but hired by the venue, citing a Daily Mail article as a source. The Daily Mail article contains no mention of the bodyguard's supposed employer and as such does not support the Community Note. Bellingcat then go on to show the bodyguard has been seen working with Swift for a while and in different locations, further disproving the Community Note.

Reply on news.ycombinator.com