Tell HN: new and non-neg low karma users, your posts appear to be auto-killed

#1

Post by bookaway » Thu, Mar 06, 2025, 9:45 AM UTC

Just an FYI -- I don't know if this has been the case for a while and I'm just noticing it, or a new moderation change but a lot of noob submissions and low karma user submissions (sub 100 points?) seem to be getting auto-killed despite some being non-political legit tech submissions.

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)

Re: Tell HN: new and non-neg low karma users, your posts appear to be auto-killed

#2

Post by not_your_vase » Thu, Mar 06, 2025, 10:48 AM UTC

Well, to be fair, what are the chances that a user who has registered a few minutes ago without email with username "createvideoai" or "username998" (just took random spammers from the new page) will submit a non-spam link? It's arguably naive and simple heuristics, but I guess it's a reasonable baseline...

Re: Tell HN: new and non-neg low karma users, your posts appear to be auto-killed

#3

Post by bookaway » Thu, Mar 06, 2025, 11:11 AM UTC

not_your_vase wrote: post #2

Well, to be fair, what are the chances that a user who has registered a few minutes ago without email with username "createvideoai" or "username998" (just took random spammers from the new page) will submit a non-spam link? It's arguably naive and simple heuristics, but I guess it's a reasonable baseline...

If you go to the "lists" page there is literally a "noobstories" page, because someone deemed that that would be an interesting filter for some people to explore. What's the point of posting if you're a noob if you know your legit post is going to be auto-killed and no-one is going to see it?

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.

Re: Tell HN: new and non-neg low karma users, your posts appear to be auto-killed

#4

Post by genezeta » Thu, Mar 06, 2025, 11:11 AM UTC

Stuff marked [dead] on the first page of /newest right now:

- 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.

Re: Tell HN: new and non-neg low karma users, your posts appear to be auto-killed

#5

Post by bookaway » Thu, Mar 06, 2025, 11:22 AM UTC

genezeta wrote: post #4

Stuff marked [dead] on the first page of /newest right now: - 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…

Your impatience has betrayed you. Some examples from the next pages, that caused me to write this post:

[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)

Re: Tell HN: new and non-neg low karma users, your posts appear to be auto-killed

#6

Post by genezeta » Thu, Mar 06, 2025, 11:30 AM UTC

bookaway wrote: post #5

genezeta wrote: post #4

Stuff marked [dead] on the first page of /newest right now: - 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…

Your impatience has betrayed you. Some examples from the next pages, that caused me to write this post: [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 No…

So... except for the arxiv one, all AI related stuff. Personally I don't flag those but the overabundance of links about that can easily become tiring for people, I'd guess.

Re: Tell HN: new and non-neg low karma users, your posts appear to be auto-killed

#7

Post by bookaway » Thu, Mar 06, 2025, 11:39 AM UTC

genezeta wrote: post #6

bookaway wrote: post #5

Earlier quoted context omitted.

Your impatience has betrayed you. Some examples from the next pages, that caused me to write this post: [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 No…

So... except for the arxiv one, all AI related stuff. Personally I don't flag those but the overabundance of links about that can easily become tiring for people, I'd guess.

[flagged]

Re: Tell HN: new and non-neg low karma users, your posts appear to be auto-killed

#8

Post by genezeta » Thu, Mar 06, 2025, 11:53 AM UTC

bookaway wrote: post #7

genezeta wrote: post #6

Earlier quoted context omitted.

So... except for the arxiv one, all AI related stuff. Personally I don't flag those but the overabundance of links about that can easily become tiring for people, I'd guess.

[flagged]

I don't find it hilarious. It's just an observation that all those are AI related.

I never claimed that people "investigate" the links and kill them. But I do believe that if enough people get tired of a certain subject, they will flag such links without bothering to even click them.

Then again, you did assume that this was done automatically. I think I can be allowed to assume that it's just other users flagging stuff they don't like.

Re: Tell HN: new and non-neg low karma users, your posts appear to be auto-killed

#9

Post by yorwba » Thu, Mar 06, 2025, 12:09 PM UTC

bookaway wrote: post #5

genezeta wrote: post #4

Stuff marked [dead] on the first page of /newest right now: - 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…

Your impatience has betrayed you. Some examples from the next pages, that caused me to write this post: [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 No…

The QwQ one is a graft. (Whenever a new AI model is released, someone will register a related domain name, throw up a templated website and try to milk the hype.)

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.

Re: Tell HN: new and non-neg low karma users, your posts appear to be auto-killed

#10

Post by bookaway » Thu, Mar 06, 2025, 12:12 PM UTC

genezeta wrote: post #8

bookaway wrote: post #7

Earlier quoted context omitted.

[flagged]

I don't find it hilarious . It's just an observation that all those are AI related. I never claimed that people "investigate" the links and kill them. But I do believe that if enough people get tired of a certain subject, they will flag such links without bothering to even click them. Then again, you did assume that this was done automatically. I think I can be allowed to assume that it's just other users flagging st…

>Then again, you did assume that this was done automatically.

If you see multiple [dead] stories that are minutes old by different users that have non-clickbait tech titles it is not unreasonable to think auto-moderation is a more likely explanation than someone manually killing every single post in near real-time by keeping a filter algorithm in their head and manually checking whether each expression in that algorithm is true or false.

>I don't find it hilarious.

If YC was asking for AI startup submissions but HN was auto-killing noobie AI submissions I would find that hilarious in a black comedy sense. Pity we don't share the same absurdist humor.