There seems to be a "guilty until proven innocent" auto-moderation in place and I'm guessing other users are expected to vouch for those submissions one-by-one?
To check, toggle your "showdead" attribute. (I have no idea if that allows you to see your own dead submissions but you can at least confirm my observation)
If you have questions like this, you should be sending them to hn@ycombinator.com so we can actually see and answer them, not posting metathreads about it. (This is in the site guidelines: https://news.ycombinator.com/newsguidelines.html.) The only reason I found out about this thread is that a different user did email us.
It's all too easy to look at a limited set of datapoints and misinterpret what you're seeing. For example, some spam rings post legitimate (or legitimate-looking) submissions in order to cover for their spam. If you see a submission like that being auto-killed, you might well think "how can the moderators possibly be so heavy-handed" when in fact we're working hard to prevent spammers from taking over the site.
It's a constant struggle, like a cat-and-mouse game crossed with an arms race, and it's impossible to get the anti-abuse code perfect, so there will always be false positives. When you see those, you can help by vouching for them and/or by letting us know about them at hn@ycombinator.com so we can fix it. Many users do those things every day.
>Posts by new and low-karma users are not automatically killed. Anyone can see that by looking at the /newest page.
>[let us] know about them at hn@ycombinator.com so we can fix it
You’re right to chastise me about the title. I had to remove all the qualifiers to meet the title character limit (also negative-> neg for that reason), so the title was more chicken little than I would have liked, but I did leave the qualifiers in the explanation, where I wasn’t claiming that all noobie posts were getting killed. My primary intention was to warn some of the good faith noobies who looked like they were getting caught up in the spam filter, and a post that noobies could see seemed to be a more appropriate venue for that. I didn't think my post would provide any extra utility to spammers. I apologize if the person that emailed you wasn’t a good-faith noobie poster concerned about their own posts.
But overall, because of the information asymmetry, I do think you underestimate the positive impact on the community of you coming out of the woodwork to provide explanations about specific instances once in awhile, as much as you loath to do so it seems.
e.g. TIL approx +50% of noobie posts are potential spam at a given time.
But there are false positives. If they are interesting, you can vouch them. If they are very interesting and vouching does not work you can send an email to dang hn@ycombinator.com . It's a manual process so use it only for special cases. (Remember to include a link to the post.)
- A Chrome extension for tracking dramas. It appears to have been re-submitted, as the link was previously shared just yesterday (and it was dead, too), but this particular author has a lot of dead links which seems peculiar.
- Questionnaire for a school project. I don't know what this is because it's a Google docs link and I'm not going to open it. But also because the submitter doesn't even bother mentioning what the "chosen subject" is. I don't really care much that this is dead.
- "Trillion Ton Iceberg, Worlds Largest, Crashes into South Georgia" a youtube video which is hardly a video since it's just a bunch of AI generated illustrations and appears to be fairly grandiose but low on actual sources and info. meh
- A link to a tweet from an account that "applies AI" to put your face on a ballerina animation. Yawn.
- Spam for a "IT solutions for healthcare" company.
- A link submitted by someone who has all dead submissions on their profile. Uhmm...
- Spam to a blog about AI, I guess.
All in all, I'm setting showdead back to no again.
P.S. Meanwhile, I counted about 5 or 6 links sent from newly created users which are there happily unflagged.
You can use the "replies" endpoint to see if dang (or one of the other/earlier mods: sctb or pg) have replied directly to the user with a ban message, e.g.:
https://news.ycombinator.com/replies?id=genezeta&by=dang
(This will show any replies by dang to you, genezeta, not necessarily ban notices.)
I've just tried that on the first non-green HN userID for a "[dead]" submission the New queue, with nothing showing, so it's a bit hit-or-miss.
Alternatively you can look at the member's submissions under their profile page. Patterns such as constantly submitting the same set of domains, especially if self-promotional, will stand out. That userID I'd checked above ... seems to have a limited range of sites they're interested in submitting:
<https://news.ycombinator.com/submitted?id=adib_zouiten1>
Some sites are banned due to impenetrable paywalls. 404 Media is one such (with several submission in the queue presently):
[dead] Show HN: Try QwQ-32B for Free: Advanced AI Reasoning Model (qwq-32b.net)
[dead] Show HN: I made an AI platform that combines GPT-4o, Claude, DeepSeek and more (typethinkai.com)
[dead] Some basic primitives for PHP OpenAI Client (github.com/assistantengine)
[dead] Benchmark on Multi-Embodiment Intelligence Normative Data for Robot Manipulation (arxiv.org)
[dead] Using "tools" support (or function calling) with LangchainJS and Ollama (hashnode.dev)
One was killed because the account, while not banned, is breaking this guideline: "Please don't use HN primarily for promotion." (https://news.ycombinator.com/newsguidelines.html)
One was killed because the site is banned on HN (in this case, because it has been the source of too many low-quality posts and too much promotional behavior).
Two look like false positives by spam filters, though it's not so easy to be sure. I've unkilled those now:
Show HN: I made an AI platform that combines GPT-4o, Claude, DeepSeek and more - https://news.ycombinator.com/item?id=43278331
Benchmark on Multi-Embodiment Intelligence Normative Data for Robot Manipulation - https://news.ycombinator.com/item?id=43278119
TypeThinkAI also looks like a graft, but at least they're honest about just being a wrapper around models produced by others?
I vouched for one of the AssistantEngine links because although the author did submit a bit too many links, the project might deserve a chance.
I also vouched for the arXiv paper.
hashnode.dev has been banned for a long time, probably because spammers use it a lot.
Good-faith posters should know that this is what is happening (assuming the auto-killing is intentional and not a bug) so they can go and try to build enough karma from their comments before posting a story.
Some such submissions get killed, mostly for reasons related to spam filters. But there's a huge difference between "some" and "all". I'd appreciate it if you'd stop making dramatic "all" claims about what is very much a "some" situation.
Obviously we have no intention of impeding legit users and legit submissions. Legit new users are what keep HN alive. Why would we cut off our own oxygen? The issue is that it's hard to write spam filters that don't produce any false positives. Perhaps you know how to do that; I don't.
If you want to help, vouching for the false positives (i.e. legit submissions that get killed by mistake) is helpful, as is emailing links to hn@ycombinator.com.