On the second page, it's 9. On the third, 11.
It seems unlikely that many people are downvoting them so quickly, and for most I can't see why they should be dead.
I know that there's websites known to publish only AI slop which are probably blacklisted here, but it seems unlikely that a third of the submissions are about them.
There's some automatic AI filter now?
How does it work, and are people assumed to check all recent dead posts and vouch for them if they don't deserve to be dead?
I imagine that the wide majority of users assume there's a good reason for something to be dead, and ignore it rather than performing that check.
I read /newest almost every day and vouch for submissions that I think have been unfairly killed but that's almost never.
check all recent dead posts and vouch for them if they don't deserve to be dead?
To the extent it is important to someone they will do it. To the extent it is not, they won’t.
How does it work
I suspect using tools, heuristics, and intuitions developed through direct experience within exactly the circumstances of running HN.
Who here knows that it's something you're supposed to do, if you are?
I imagine that each new submission is seen at most by a handful of people, by the way, on average probably too few to resuscitate a dead one.
And I hope we can get an actual answer to how does it work.
> Most pages on the internet are not a good fit for HN
Most pages on the internet are not submitted to HN
Nothing is preventing anyone who feels it is important from doing it to the degree they feel that importance.
And I hope we can get an actual answer to how does it work.
The way it is is probably the best approximation of how it is supposed to work.
> The way it is is probably the best approximation of how it is supposed to work
Sure
Since dead posts are only shown to logged in users we can't even use archive.org to check the reliability of the flagging
Currently on new for dead/flagged: new account self spam / new account zero effort ai slop spam / new account zero effort ai slop spam / show hn ai slop with all default llm design / spam / spam / spam / account that has never posted anything but its own blog / spam (all of dev.to is dead afaik, because it is nothing but a spam source; there are no useful posts on it)
There is nothing that I consider even slightly interesting or reasonable or innovative to hackers. I am also not anti-gen AI but there is a line between "person has used claude to create something" and "literal zero-effort unreviewed trash that is a waste of the environment". >99% of ones I'm see in /new is the latter.
The two submissions from bamei2ai are not working links, so no questions about them.
Most of the others are from new or very low karma accounts, ok, but is it forbidden to submit something before having gained a good karma? Are submissions from such accounts killed automatically?
The only dead post left is https://news.ycombinator.com/item?id=48798889 , I'm not sure what's wrong about it
There’s nothing particularly unusual about this page of /new. I opened up current /new after finishing this post and on page one found a YouTube link, which made me think “that’s odd”, looked more closely, and ended up writing a twenty minute mod email reporting several more sites for spam; the future [dead] count will increase accordingly.
> is it forbidden to submit something before having gained a good karma?
Not precisely, but it’s in really poor taste after the first time, and I believe the mods tightened up the Show HN thresholds a bit recently due oto the codeslop flood. This is a community, not a marketing channel; if they’re not here to submit and/or discuss other people’s work, then it should come as no surprise that we’re not here to discuss theirs either.
> Are submissions from such accounts killed automatically?
Not that I’ve seen. Certainly lots of them end up killed b/c spam bots are detectable, but it’s not uniform. But, as I said a couple weeks ago in a variation of this post: Either genuinely invest time and energy into this community or move on.
> I'm not sure what's wrong about it
#51: Boring submission. Poor composition skills, no effort invested in even basic capitalization. Open admission of AI-generated writing: “The site is a bit slop”. Deserves as much attention as effort invested. This post could have been replaced by a training harness.
User has only <100 karma on an account 5,000+ days old, so this is less a site participant, more a fly-by-night forum spammer as their participation stands right now.
I would flag this if it wasn’t already flagged. Thanks to whatever HN system or user saved me having to do so.
And if it's some Hacker News own bot doing some of the killing, that should be declared.
And a ton of other things, I like Hacker News's interface and in part its user base, but I'd change much of the rest.
Dang said on the AI guidelines post that they’re working on this now.
> And if it's some Hacker News own bot doing some of the killing, that should be declared.
First rule of antispam is do not publish the details of antispam decisions automatically. This post could have been replaced by an email to the mods asking what’s going on here. I’ve learned of a few possible automated systems that could be engaging in parallel while reviewing /new posts and flagging and reporting folks and sites by email to them.
Yes.
> Would knowing that they were flagged for one thing but not another make it easier for for them to identify what it was that let them slip past undetected?
Yes.
It seems unlikely it would help much, to me.
But I'm still trying to understand if an automated flagging system is in place to begin with.
It would help reviewers a lot in salvaging unjustly flagged posts, though.
> I think moderation here works pretty well
I don't
https://news.ycombinator.com/item?id=33272357 https://news.ycombinator.com/item?id=38412074 https://news.ycombinator.com/item?id=28457450
No real answer there though, unless there's someone like https://news.ycombinator.com/item?id=38413375 flagging like there was no tomorrow for subjective reasons.
But that’s just my experience.
And you can't vouch for them until they're dead, by the way, another nice detail (in the meanwhile they are removed from the front pages).
I recommend setting "showdead" to "yes". You see the damndest things.
It's flagged but not dead submissions that I can't find.
The submission queue definitely gets gamified - self promoting articles seem to get a massive surge of upvotes suggesting a kind of bot farm.
Enquiring minds wanna no.
They only allow you to read what they want you to read.