Hacker News Reader

Comment by bjourne | Hacker News Reader

Haven't they noticed what a complete cluster fuck "Community Notes" are on Twitter? Just way to easy to game and just way, waay to easy for hordes of trolls to "Community note" real accounts into oblivion. For example, a note on a story may read: "This image is likely AI-generated. Source: Some Random Post". Except Some Random Post is written by an idiot who is completely wrong. Doesn't matter one bit because the troll hordes can generate more clicks on the up button than the honest users can generate on the down button.

View on HN

Community Notes aren't simply voted on by counting clicks on an up button, it's more sophisticated than that. Go sign up and do some rating yourself to see how it works.

CNs are vastly superior to the NGO swamp Facebook have been funding for the past decade:

1. The CN community focus on the ordinary boring stuff that NGOs ignore. A huge number of community notes are highlighting financial scams, obfuscated adverts, stuff faked for viral engagement and so on. They aren't political but this is the bulk of what needs fact checking on social networks. Because they're crowd sourced, CN posters are incredibly good at this and easily identify obscure videos or pictures that have been taken out of context (or indeed AI generated). In contrast the ecosystem that FB funded seemed to focus primarily on contentious political claims.

2. To go live a CN needs agreement from people across the political spectrum. The algorithm behind this decision is quite smart. It's reminiscent of the old Slashdot meta-moderation system, for instance you are asked to rate notes across multiple different criteria that come with objective definitions. The result is that CNs are almost always written in a neutral style that just briefly presents facts with sources, and the extremely biased "fact checks" of the sort that have been embarrassing Facebook for years are basically absent. By the nature of the system sometimes bad CNs appear briefly, but they get fixed quickly.

3. Because it's algorithmic and anyone can write notes, it doesn't discriminate in favour of the rich and powerful. Elon Musk gets community noted, he even sometimes just posts stuff and then asks CN specifically to check it. Facebook's "fact checks" were written by the sort of people who simply assume academics, civil servants and activists are always right, with the result that they never fact checked such people even when they were saying things that were completely wrong.

4. They are much more concise.

5. Even the name is better. They don't claim to be final judgements of truth, just notes created and voted on by an amorphous community. This is a much more defensible claim: much of the scorn towards fact checkers acknowledged by Zuck is because they claim to be "checking facts" but often just pick fights or try to block true facts they don't like.

Agree with all that, but would counter for Musk's CN's specifically there is definitely monkey business going on. Most of the CNs never show up for his disinfo, so either he's manually removing them or they are getting brigaded to hell.

Sometimes Musk does post a factual claim that's wrong in some way, specifically asks for a fact check, or posts something misleading, and then a CN does appear. Examples:

https://x.com/elonmusk/status/1876435054622388257 (from yesterday)

https://x.com/elonmusk/status/1876554867785007175

But why not sign up for a CN account like I did and go look. Last time I did this almost every tweet he posted had a gazillion proposed CNs but one of the rules that the community is good at enforcing is that a CN is not meant for just arguing with someone. Many of the proposed CNs on Musk's posts didn't meet the criteria to be a note and so they get meta-moderated out on the grounds of "no note needed" (NNN), which is the reason people pick when a tweet isn't making a factual claim to begin with.

Most of what Musk posts are either opinions or just retweets, and arguing with an opinion is always supposed to be marked as NNN. And CNs go on the original tweet not retweets.

There is no note on the first one.

I’m agreeing with you.

Twitter used to have a problem with retaliation if you dared talk back or correct the “wrong person” (mass reporting, suspending, taking down anyone who bruises an influencers massive ego - which still works unfortunately and is still a problem under Musk. But CN made sure this wouldn’t happen to you for telling the truth.)

Or a minor nitpicking comment or correction becomes a flame war and now someone’s digging up your high school year book photo or your DUI mugshot from 20 years ago. (Root cause being, again, twitters retaliatory culture) CN just kinda stops people from shooting the messenger.

And sometimes the messenger sucks but still has a point.

CN is a twitter-specific solution to Twitter-specific problems.

It would have also been good if they’d stop suspending people with no human review just because an influencers army of flying monkeys reported them. It “only” takes about 50 and then you’re done forever and all appeals denied. This is one of several reasons I don’t use Twitter, none of which are political or about it’s supposed “owner.”

You can’t block these guys, god forbid you reply to one- no thanks

> Community Notes aren't simply voted on by counting clicks on an up button, it's more sophisticated than that. Go sign up and do some rating yourself to see how it works.

I did. I was almost immediately locked out since too many people disagreed with me. And no, it is not more sophisticated than that. Users have become very good at exploiting the system resulting in only clicks counting.

> To go live a CN needs agreement from people across the political spectrum.

This is false and is not the "security hatch" you think it is. Rather than arguing these points, just look at a few suggested CNs:

Tweet: "A white grooming gang caught abusing 30 girls. The youngest was 5. Why did this not receive as much attention?" CN: "It did get attention which is demonstrated from the screen grab attached to the post which is from the BBC website."

The CN is not correcting misinformation. It attempts to argue with the twitterer because, apparently, replying is not good enough. Another one:

Tweet: "Doctor admits Israeli pathologists harvested organs without consent" CN: "The full article says Israel admitted to harvesting organs from deceased IDF soldiers. Israel denies the claims of harvesting organs from Palestinians, which originated from a Swedish news outlet. No evidence exists that Israel harvested organs from Palestinians at this time."

Except the CN is wrong and Israel indeed harvested organs from slain Palestinians: https://www.jstor.org/stable/24590832 Another one:

Tweet: "Rennes are completing Brice Samba deal from RC Lennes as planned, it’s all set to be sealed." CN: "The team is called RC Lens, not Lennes."

The CN author is collecting points here. Correcting a misspelled name of a football team doesn't add much, but by doing so the author can farm karma for use on more impactful notes. Another:

Tweet: "Shut up, Elon Musk! We don‘t want your fascism in Europe!" CN: "Elons actions don‘t seem to support the claim of fascism according to the defining 14 Points of fascism of Umberto Eco:"

Again someone who prefers the CN feature over replying. Yet another, from the former president:

Tweet: "Four years ago, violent insurrectionists attacked the Capitol, threatened the lives of elected officials, and assaulted brave law enforcement officers." CN: "Neither Trump, or any of the January 6th defendants, were indicted, charged or convicted of "insurrection" as a result of the Jan. 6, 2021 riot. Calling the rioters "violent insurrectionists" does not accurately reflect any of the charges filed. "

Yeah, that just, uh, your opinion, man.

These examples are all from English Twitter, where the huge number of users balance the situation out somewhat. On Swedish Twitter, where a famous left-wing politician might collect 20 likes for a tweet if they are lucky (drawback of being only 10M people), the right-wing trolls utterly dominate. If one politicians complains about high electricity prices the trolls will be there with a CN about how "they destroyed nuclear power" (utterly false BS) or if it's about crime the CN will be about how "you let the immigrants in".

> Because it's algorithmic and anyone can write notes, it doesn't discriminate in favour of the rich and powerful.

Uhu. They said the same thing about Wikipedia, didn't they? In reality, fanatics with a lot of free time on their hand and those who can pay for meat puppets will decide what the truth is.

You're assuming Meta's goal is to prevent misinformation and disinformation.

Reply on news.ycombinator.com