591–600 of 996 posts
Since it's getting downvoted hard and might be missed, FYI Michael is in the house. I encourage y'all to read the whole article before engaging. https://news.ycombinator.com/item?id=45522396
Reading it, seems like the downvotes are deserved.
This is oddly a case to signify there is value in an AI moderation tools - to avoid bias inherent to human actors.
Do you understand how AI tools are trained?
Earlier quoted context omitted.
I didn't think it should be illegal to say "I wish X were dead" unless you have the means to make it happen like a rabid audience with a track record of killing people you wish were dead. Even then, I think there needs to be coordination or a wink nod of some kind that needs to be proven to muzzle free speech. Freedom of speech we don't like is the true litmus test of free speech. It is trivial to say I support free…
> Freedom of speech we don't like is the true litmus test of free speech. It is trivial to say I support free speech when someone says nice things about me. People are free to say mean things, they just aren't allowed to encourage violence. There is a difference between saying "I hate how Trump runs the country, he is an idiot" and "Can't someone kill Trump already". I have seen a lot of the second kind on reddit. Th…
Like, on Reddit you're technically not allowed to say "Man, I'm so glad that Adolf Hitler guy is dead."
To my knowledge it is not illegal to say something like that (anywhere you want to live anyways)
I think that a world where people are free to talk about how they're glad horrible tyrants are dead is a better world than the one where they aren't free to express those ideas.
You wouldn't want to live in an America where it wasn't legal to condone the death of Osama Bin Laden, would you?
If you suspect something is a commonly held misconception, frequently asking a LLM about it is close to useless, because the abundance of text repeating the misconception (it is common after all) just makes the model repeat the falsehood. Asking the model to apply a more balanced view quite often triggers an iconoclastic antagonism which will just give you the opposite of whatever the commonly held opinion. I have had some success in asking for the divergence between academia and public opinion on particular issues.
Models are still very bad at determining truth from opinion, but these are exactly the times when we need the models to work the best.
There may be an opportunity if there are enough known examples of instances like this story for a dataset to be made where a model can be taught to identify the difference between honest opinion and bad faith actors, or more critically identify confidently asserted statements from those supported by reasoning.
Unfortunately I can see models that can identify such falsehoods being poorly received. When a model can say you are wrong and everybody around you says you are right, what are the chances of people actually considering the possibility that the model is, in fact, correct?
Earlier quoted context omitted.
Hello. It's nice to be able to interact directly with the subject of the article, so thanks for coming on. It's a shame you're being downvoted, because it would definitely be interesting to hear your perspective. This can't be a pleasant experience for you. I have a couple of questions for you. Firstly the article really didn't hold back about you in a way that you don't usually see. But he makes very specific and ve…
This is a wonderfully mature and constructive comment. I appreciate this is off-topic, but I really wanted to highlight/praise what you'd written. It came across to me as very "HN" and the guidelines appear to corroborate this... > Comments should get more thoughtful and substantive, not less, as a topic gets more divisive.
Novati makes two relevant points:
1) Article author did not reach out to Novati for comment.
2) Codesmith ran a misguided marketing campaign that ran afoul of Reddit auto-mods.
These seem like reasonable responses to me. Are they true? I don't know, but I would like reasonable discussion about it.
I really get the impulse to vent when reading about something so outrageous, and in a way this kerfuffle represents the arbitrary unfairness of power leveraged from the shadows, but is that happening here?
I'm french I was banned from the France subreddit for saying Hamas fighters dress as civilians. The problem of Reddit goes beyond astroturfing. It's a judge jury executioner problem. Moderation is the most expensive problem of online platforms.
The irony of your post being downvoted for stating facts (i.e. what personally happened to you).
Example:
If one side says "Hamas dresses as civilians" to excuse killing civilians, they're factually correct and criminal.
If someone then goes online and drops "Hamas dresses as civilians" they're factually correct and reinforcing a criminal's defense.
Anyway, the morality of IDF behavior is apparant to an impartial behavior by just considering the thought experiment that IDF doesn't killing Israeli civilians "just in case" they're Hamas in disguise. They only kill Palestinian civilians "just in case" they're Hamas.
The LLM aspect of this, I think shows both a common weakness and an opportunity. If you suspect something is a commonly held misconception, frequently asking a LLM about it is close to useless, because the abundance of text repeating the misconception (it is common after all) just makes the model repeat the falsehood. Asking the model to apply a more balanced view quite often triggers an iconoclastic antagonism which…
Models are not bad at it. Models are not even trying. As you point out, it is about what is the most common text. It has nothing to do with logic.
Earlier quoted context omitted.
The way HN and public forums work is that people can ask questions and others can answer. The post you are replying to is an answer to a question. You need to scroll a bit up to see the original question.
Pointing at the moderation of an explicitly conservative place for right wingers as a grievance to illustrate how it is only balanced how conservative opinions are getting banned across mainstream subreddits is fairly disingenuous. And clearly this has been a discussion on that angle, rather than an answer to the rhetorical question above.
How do you know the question was rhetorical?
I don't know about this particular case, but, generally... bad actor subreddit moderators have been an occasional thing for well over a decade. And it's also been widely known for that long that Reddit is an influential venue in which to take over a corner -- for marketing or propaganda. What's an equal concern to me is how insufficiently resilient Reddit collectively appears to be, in face of this. A bad actor mod o…
Reddit has a moderation problem, and it's a big one. They've now been asked to appear in front of Congress to address concerns about politically motivated violence being incited through their platform: https://oversight.house.gov/release/chairman-comer-invites-c... Personally I believe I've seen more people in the past few years wish a politically motivated death on somebody else via Reddit, than I have anywhere else…
Funny how for the last 30 years of right wing violence/extremists far exceeding left wing nothing was done at all about it, no questions asked. Hush hush, don't talk about gun control or the real causes of these peoples' actions.
But then the year that left wing violence finally exceeds right wing, they all start crying that it's unacceptable and something needs to be done about it.
Source: https://www.csis.org/analysis/left-wing-terrorism-and-politi... "So far, 2025 marks the first time in more than 30 years that left-wing terrorist attacks outnumber those from the violent far right."