Hacker News

Top stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Jeff – Jev-compatible 0.8B decision models, trained at home, ~30 ms (github.com/firelex)
    47comments
  2. Who should be held accountable when an AI Agent (accidentally) acts maliciously? (greenpants.net)
    37comments
  3. Pirating the Pirates (mubi.com)
    194comments
  4. 12,000-year-old Göbeklitepe burials explain scattered bones (archaeologymag.com)
    8comments
  5. MicroLLM Lab – Try 7 tiny LLM's in the browser (stateofutopia.com)
    43comments
  6. World Labs Is Joining AMD (worldlabs.ai)
    47comments
  7. Scientists solve 1840s space weather mystery (arstechnica.com)
    21comments
  8. Hijacking the PS5's RTMP stream (yashgarg.dev)
    54comments
  9. Joseph Szabo’s pictures of American adolescents (newyorker.com)
    34comments
  10. Parley: Federated, decentralised chat that speaks plain IRC (mills.io)
    150comments
  11. It's Time to Investigate the AI Labs (calnewport.com)
    53comments
  12. Kids turned low-traffic NPR Spotify comments into a secret group chat (thisamericanlife.org)
    149comments
  13. Best of British Design (best-of-british-design.vercel.app)
    23comments
  14. What reversing, modernising old games tells us about the economic impact of AI (isfine.org)
    8comments
  15. Flock Wants the Most Detailed Map of Its Surveillance Cameras Taken Offline (theintercept.com)
    43comments
  16. 3D necroprinting: Leveraging biotic material as the nozzle for 3D printing (science.org)
    —discuss
  17. Sonnet 5.5 (anthropic.com)
    344comments
  18. Cf: The Agentic CLI for the Cloudflare API (cloudflare.com)
    42comments
  19. Nvidia wants to put a watchdog chip next to every AI agent (cnbc.com)
    127comments
  20. Does Reddit have an astroturfing problem? What the data suggests (petervijeh.com)
    91comments
  21. Show HN: Destroy Any Website with Stickman (spritefusion.com)
    25comments
  22. First Steps of the PLC Organization – Independent Public Ledger of Credentials (plcred.org)
    18comments
  23. Launch HN: Vespper (YC F24) – SOTA Docx MCP (vespper.com)
    8comments
  24. What heraldry and Japanese mon can teach about visual-identity generators (benovermyer.com)
    20comments
  25. Who wrote Elizabeth I's most scathing letters? (smithsonianmag.com)
    18comments
  26. Updated Google Maps shows destruction of the city of Rafah (twitter.com/aliabunimah)
    36comments
  27. MongoDB CEO resigns to join Meta (reuters.com)
    252comments
  28. Pacing the Frontier is not the actual goal for AI labs (lesswrong.com)
    60comments
  29. GrapheneOS – When an app is slow (wirelessmoves.com)
    41comments
  30. When did Google get so weird? (sancho.bearblog.dev)
    1021comments

Ask HN: Why so many dead comments lately?

13 pointsby 5d ago
10 comments
I've noticed a lot more dead comments lately in cases where the person is writing reasonable comments. You'll need showdead turned on in your settings to see the examples.

Take this thread as an example https://news.ycombinator.com/item?id=49818698

The OP has a new account but all his replies are within the guidelines.

Here's another account with a dead comment in that thread. Most of their comments, while brief, seem perfectly reasonable. The account is 2 years old but most comments are dead https://news.ycombinator.com/threads?id=josefritzishere

In this thread (https://news.ycombinator.com/item?id=49574167) all of this user's (https://news.ycombinator.com/user?id=backlava12) comments were marked as dead. The user has an established account and all comments seemed within guidelines.

Has anyone else noticed this more recently? Is it perhaps a result of trying to identify AI-bot accounts?

5d agoHN ↗

HN’s mods have said they have huge amounts of code fighting spam alongside human workers, and they deliberately don’t discuss their tactics in order to avoid an arms race between themselves and the spammers

5d agoHN ↗

I dunno. One of my pet peeves of the last few months is that somebody posts an AI generated blog post and it's not good enough that we can click on it to read it so they also post an AI summary as a comment. Like... we can click on the link and read your post. Haven't you ever used HN before... Oh, I see, you created this account six hours ago.

I think they might have tuned up the heuristics to suppress that kind of thing but that's not what is going on there... like these are real replies in a real conversation. I don't like it that I can't upvote or vouch for them.

5d agoHN ↗

I take it is due to automation, mostly. Every single website that has some sort of user-contributed content, forums, source code repository, social media, anything that has comment or upload capability of any kind whatsoever is being bombarded by a massive torrent of crap, and hn is no exception. The internet has no good defense against it, therefore people have to resort to captchas, geoblocking, shadow banning, proof-of-work portal solutions, cloudflare (eek!) and adjusting detection thresholds "just right" so things feel usable.

Hat's off to the mod staff, things feel "alright" even with the current situation.

5d agoHN ↗

Self promotion posts are one thing, not sure if it applies to the examples you gave, good luck if your hn user name and your github user name is same when you have low karma.

4d agoHN ↗

I'm not sure if commenting on this thread will help since I recently discovered I'm affected by this when a kind stranger replied to one of my recent comments saying they had to "vouch" for me, so you might not see my comment anyways

I think in my account got flagged because of three things

1. Age, it's a pretty recent account, I used to have another one that's like 8yo by now (if not more) but I lost access to that email (and I could've benefited from a username change, so I decided to create a new one)

2. I created this account using a VPN, I have a slight suspicion that it caused me to be soft-banned from the start, but I can't confirm

3. HN probably implemented more strict filtering due to the advance of LLMs

If I were to pinpoint on a exact cause I'd say it's more likely that 2 and 3 are contributing much more than 1

4d agoHN ↗

Your account got caught in a spam filter, probably by mistake. Sorry! I've unbanned it now.

4d agoHN ↗

Thanks for replying. These were just some examples I've seen recently and was able to find them easily in my history. There have been more that I've noticed lately. I imagine this will continue to be the case going forward.