from Hacker News

What do I think about Community Notes?

by fbrusch on 8/28/23, 10:27 AM with 221 comments

  • by ryukoposting on 8/28/23, 1:46 PM

    A very interesting article. It seems like the Community Notes system is built on (in my opinion) a very mathematically sound model. It seems far less susceptible to biased "mob actions" than a conventional upvote/downvote system, even if it's not impervious to such actions. "Bias" exists in the eye of the beholder, so it's unreasonable to expect any content rating system to be perfectly unbiased.

    Community Notes and crypto share a common thread of "complicated math that makes it both extremely powerful and totally opaque to 95% of people." Honestly, that's a pretty weak relationship for how much page-space it got in the writeup. Looks like Vitalik is writing a crypto blog, so I guess you gotta appeal to your readers. Great read nonetheless.

  • by major505 on 8/28/23, 12:28 PM

    I really like community notes. It's been a while in Brazil they are being used to check politicians who always lie in their post and try to rewrite history.

    Its like a small note of shame in their posts, and it seens so far they been using it in politicians in all sides of the political spectrum.

  • by pjc50 on 8/28/23, 12:23 PM

    > deep learning vs crypto is a clear divide of rotators vs wordcels. the former offends theorycel aesthetic sensibilities but empirically works to produce absurd miracles. the latter is an insane series of nerd traps and sky high abstraction ladders yet mostly scams

    Can anyone translate this from meme to English please?

    Also, it would seem to be a very anti-crypto statement for Vitalik to be posting?

  • by mellosouls on 8/28/23, 1:34 PM

    Community Notes will likely have some of the same pros and cons of wikipedia.

    Community Notes are not written or curated by some centrally selected set of experts; rather, they can be written and voted on by anyone ... It's not perfect, but it's surprisingly close to satisfying the ideal of credible neutrality...

    Well, that's the optimistic pro. But the con is that if a particular demographic is more drawn to contributing to those notes (or comes to overwhelm Twitter itself), we will see the same problems of bias we see on Wikipedia in (say) social-cultural subjects which - whatever it results in - is certainly not "credible neutrality".

  • by skilled on 8/28/23, 11:39 AM

    The few that I have seen have been pretty good. I also like that they can be added to Ads, and many Ads running on Twitter right now have community notes because they're clearly saying one thing while the product does something else.

  • by bandrami on 8/28/23, 11:24 AM

    Community Notes is literally the one good thing that has happened to X in this whole thing.

  • by Yhippa on 8/28/23, 2:00 PM

    Someone alluded to this in the comments here, but this does strongly remind me of the Slashdot moderation method but much more sophisticated. That was my favorite site moderation method. Giving meta judgement/commentary on things felt like a privilege and not a given.

  • by thomastjeffery on 8/28/23, 10:11 PM

    I think the biggest problem is treating "polarity" as a single axis. Many subjects align that way in practice, but the ones that don't are important.

    It's nice to see some effort behind distributed moderation, but this is still too centralized to realize that dream.

  • by chownie on 8/28/23, 11:25 AM

    The article repeatedly makes effort to assure the reader that the left/right dichotomy wasn't hardcoded but instead "emergently" appeared, then never goes into any detail about how that happened exactly...

    It's weird to me that the left right divide happened to appear, rather than a faith based/non faith based, or an authoritarian/liberal, etc dichotomy. Is there no public data on this we can use to prove it's true, just trusting the org at it's word?

  • by Analemma_ on 8/28/23, 11:17 AM

    An uncharitable but not exactly incorrect summary of this would be: most people love Community Notes; Vitalik really wants some of that goodwill for crypto (which is desperately lacking it), so he insists the two have a common philosophical and/or functional heritage. I think this is nonsense and he's trying to hitch a dead ox to a live one, but it's his blog and he can do as he likes.

  • by exo762 on 8/28/23, 11:24 AM

  • by dherikb on 8/28/23, 1:15 PM

    It's better than any other alternative that I saw until now.

  • by snowwrestler on 8/28/23, 11:49 AM

    It seems like an entire generation has grown up believing that vote features on the Internet exist to improve quality, because that's what they were told by the websites who marketed those features.

    Members of a prior generation may remember that voting features were introduced explicitly to improve engagement, not quality. Those little triangles next to this comment are there to give you something to do, and thereby hopefully increase the chance that you'll feel invested in this site to come back. Not to discover the truth, improve awareness of facts, or build community.

    Voting is a mechanism by which people express their values. If people know facts and value them, then a voting mechanism will deliver facts. But if they don't, then it won't.

    I will say it seems to take an impressive lack of introspection to spend thousands of words expertly fact-checking how Community Notes works, and yet still conclude that fact-checking by experts "seems risky."

    Vitaly's personal values are expressed most clearly here:

    > ultimately I come down on the side that it is better to let ten misinformative tweets go free than it is to have one tweet covered by a note that judges it unfairly

    It should be noted that he is less sanguine about letting misinformation run free when he is complaining about inaccurate press coverage of cryptocurrency. (Pretty much like anyone who complains about inaccurate press coverage, honestly.) It's always easier to be sanguine about someone else's misinformation that affects other people.

  • by pastacacioepepe on 8/28/23, 2:04 PM

    I was positively surpised by Community Notes. That's until I witnessed a correct, factual note, removed from a @NATO_ACT tweet[0].

    The removal was probably due to brigading of the "Not Helpful" feedback button, but for how good the Notes concept is, it doesn't take into account such a common opinion manipulation strategy.

    Wikipedia is facing similar issues, and I see the same happen with the flagging feature on HN.

    .0: https://twitter.com/NATO_ACT/status/1692565516924112916

  • by CamelCaseName on 8/28/23, 1:14 PM

    I've seen quite a few bad notes, either flat out incorrect or on jokes or shitposts.

    It was much better when it was applied on just a few tweets but much higher quality.

    Just another thing Musk has ruined.

  • by a_person_2017 on 8/28/23, 12:31 PM

    Does anyone know what blogging / wiki platform this is on?

  • by amai on 9/3/23, 7:09 PM

    https://pol.is mentioned at the end of the article seems to be very interesting.

  • by lbhdc on 8/28/23, 1:16 PM

    https://tsjournal.org/index.php/jots/article/view/100

    There has been some interesting research done around the effectiveness of false information labels.

    I wonder how much of their findings extend to community labels?

  • by Pxtl on 8/28/23, 1:02 PM

    Community Notes are a neat idea but

    a) Twitter has too many overlapping actions. Community Notes is just "credibility-oriented reply" in the same way that retweet is "quote tweet without comment".

    When you see a tweet you disagree with and wish to correct, there are now three different ways to register that disagreement -- QT, Reply, and Notes. That's kind of clumsy.

    I feel like from an ergonomic perspective, better integration and surfacing of credibility scores as related to replies (and better integration of the QT and Reply features) would've been cleaner.

    b) I'm done with X/Twitter now that it's a monument to Musk's internet poisoning.

  • by troupo on 8/28/23, 11:10 AM

    This really begs the meme with a guy pointing at a butterfly and saying "is this crypto".

    The rest of the analysis seems to be even-handed, but the intro...

  • by samyar on 8/28/23, 11:46 AM

    they are great

  • by drones on 8/29/23, 9:44 AM

    > Currently, participants are slowly and randomly being accepted, but eventually the plan is to let in anyone who fits the criteria. Once you are accepted, you can at first participate in rating existing notes, and once you've made enough good ratings *(measured by seeing which ratings match with the final outcome for that note)*, you can also write notes of your own.

    I think twitter is especially fortunate to have the audience diversity it has to make this work - there seems to be a good mix of both left and right wing leaning users. A consensus model with such diversity encourages compromise.

    I feel like the strength of using this model is that it's low stakes - at the end of the day it's a tweet. People are willing to make concessions, because it's a tweet. As soon as, say, money is involved, I think that willingness to find a middle ground goes away.

  • by soerxpso on 8/28/23, 5:49 PM

    I wonder how reducing factions to a single axis impacts notes on tweets about matters that the two "teams" haven't really picked their sides on yet. During the first weeks of covid, it seemed like "this is a nothingburger" and "this is a huge deal" weren't statements that had a left/right connotation yet, but were statements that people passionately disagreed about. How does the algorithm behave when a note is highly controversial, but that controversy doesn't seem to be correlated with users' single-axis leanings?

    Anarcho-capitalists and radical leftists strongly disagree on the vast majority of things, but both groups passionately support defunding the police (for very different reasons). Would the algorithm unfairly see a false note as having "bipartisan support" when these two opposite-radical groups manage to have their biases line up?

  • by hndamien on 8/29/23, 1:05 AM

    Community notes needs to carry a record on the user who get noted.

  • by drngdds on 8/28/23, 12:43 PM

    Amusingly, the community note he cites complaining some LGBT books were obscenity and therefore illegal to distribute to children is misinformation in itself. Obscenity is a very high bar to pass and there's no way the linked materials would be considered illegal obscenity in the US.

  • by tomcam on 8/28/23, 11:41 AM

    > what interests me most about Community Notes is how, despite not being a "crypto project", it might be the closest thing to an instantiation of "crypto values"

    Wut

  • by epolanski on 8/28/23, 11:23 AM

    > Community Notes are not written or curated by some centrally selected set of experts; rather, they can be written and voted on by anyone

    This reminds of a Michael Scott's quote from the Office which I find brilliant in it's irony:

    "Wikipedia is the best thing ever. Anyone in the world can write anything they want about any subject. So you know you are getting the best possible information."

  • by lucideer on 8/28/23, 11:25 AM

    It's great to see a deep analysis like this.

    Recently Community Notes has been opened up to a significant enough % of the Twitter populous that you're starting to see a very large quantity of "spurious" notes (argumentative / sarcastic responses that aren't addressing facts - more appropriate as actual replies), with even one or two examples of such notes being approved, in many cases I'd guess because approvers found them entertaining rather than useful. That's lead to criticisms of CN as something "in decline".

    I hope that's not the case - spurious notes don't actually do any harm (and are often very entertaining), but dismissal of CN as a system would: loss of reputation would mean a dilution of the impact of fact checks & context provision on rebutting misinformation.

    Another possible concern is, while it seems obvious to most that CN isn't a feature that would have been conceived under Musk, and it's interesting to see it continue to thrive now, one wonders how long it will be tolerated (the example from the first screenshot in this article is telling). In possibly related news, Musk has recently shown an interest in influencing elections[0] in Ireland (Twitter EU HQ) in direct response to talks of EU "misinformation regulation".

    [0] https://twitter.com/elonmusk/status/1695545217133490453 (Gript Media is a far-right publication based in Ireland)

  • by xigoi on 8/28/23, 12:30 PM

    So if I understand correctly, Community Notes are a feature that allows you to add notes on others' posts and vote on them? How is that different from comments?

  • by bowsamic on 8/28/23, 11:32 AM

    Community Notes seems to be increasing in usage as X becomes more toxic and full of conspiracy theorists. I personally think there is a causation here. The frequency at which damaging content appeared on my timeline (just a few days ago I de-activated my account) has ballooned within the last few months. It got to the point where it was constantly serving me videos of fights or even quite often animal torture. There is also an uptick in hate, e.g. overt racism, accounts that have a lot of followers and openly ask for ethnic cleansing against non-whites, etc. Elon has turned X into "anything" (though, I know more left wing people who are still getting suspended, esp. if they criticise Musk), and Community Notes has become a "release valve" allowing the old liberal twitter to regulate posts on X and prevent the entire platform from becoming the next 8chan.

    If Community Notes are maintained at their current pace, there is the potential for X to continue on a kind of life support for the majority of "normal" people. If Elon starts to limit Community Notes, e.g. if they start being used on his posts too much, then I think that X will surely die. X already smelled bad before Elon bought it, and its stench is intensifying. What was once a nice information platform with subcultures of toxicity and hate, is now perceived as a place where one cannot escape the worst parts of modern internet culture. Community Notes is the only thing stopping it from collapsing into fanatic esotericism.

  • by yankput on 8/28/23, 1:59 PM

    > But what interests me most about Community Notes is how, despite not being a "crypto project", it might be the closest thing to an instantiation of "crypto values" that we have seen in the mainstream world.

    Huh? I didn’t see any MLM or pump-and-dump in the community notes