Retr0id

20 ms·

1. 631.▲
   by Retr0id 5mo ago

   I got it in 15 but overshot the last 2, so I'd say 13 at least is possible. But definitely not 6!
2. 632.▲
   by Retr0id 5mo ago

   Feels like there's too much rolling resistance, even with a perfectly lined up shot there's not so far you can go. And definitely not enough to make it in 6! Also if the camera angle is too high, you can't pull back far enoug
3. 633.▲
   by Retr0id 5mo ago

   As you can see here, they've already switched it out for a different command, likely due to incident responders over-indexing on npm as an IOC. https://news.ycombinator.com/item?id=48503258
4. 634.▲
   by Retr0id 5mo ago

   Heh well there was a dashed border in the "original" version and I suppose it just made it the windows-y equivalent. I imagine it wouldn't have done the same if it was prompted from scratch. ("it" being Opus 4.8 btw
5. 635.▲
   by Retr0id 5mo ago

   Agreed. It's not that the designs it produces are bad necessarily, they're just very same-y. People often talk about the bootstrap era, but that wasn't as bad because bootstrap wasn't so strongly associated with low-e
6. 636.▲
   by Retr0id 5mo ago

   Continuing in that vein, it'd be interesting to see a Win98 version. Edit: https://retr0.id/stuff/deslop/
7. 637.▲
   by Retr0id 5mo ago

   I call it claudeslop but I suppose claudish is slightly less inflammatory.
8. 638.▲
   by Retr0id 5mo ago

   They could've pip installed, curl|sh'd or anything else, it's not relevant to the underlying issue.
9. 639.▲
   by Retr0id 5mo ago

   This doesn't really have anything to do with npm.
10. 640.▲
    by Retr0id 5mo ago

    I haven't used Arch for a few years now, but when I did the AUR was my favourite aspect. It was never perfect from a security PoV, but in 2026 this kind of trust model feels increasingly scary.
11. 641.▲
    by Retr0id 5mo ago

    Seems plausible to me, they can get into a very "roleplaying" latent space, especially if the prompt is flowery enough.
12. 642.▲
    by Retr0id 5mo ago

    If such a language existed, it would surely take a human years of study to become proficient at it.
13. 643.▲
    by Retr0id 5mo ago

    I wouldn't be surprised if they encrypt them at rest, but at some point the weights have to be loaded into vram.
14. 644.▲
    by Retr0id 5mo ago

    This sounds backwards, any interrupted conversation becomes less useful for training.
15. 645.▲
    by Retr0id 5mo ago

    Any public research footprint seems to be enough, I applied as an individual and everyone I know who tried got accepted.
16. 646.▲
    by Retr0id 5mo ago

    It seems like they've given up on the idea of the Cyber Verification Program https://support.claude.com/en/articles/14604842-real-time-cy... When Opus 4.7 was introduced it started refusing anything cyber-adj
17. 647.▲
    by Retr0id 5mo ago

    I don't think it's weird that the post made it to the front page, but watching the downvotes roll in on my own mildly critical comment has been intriguing. I saw it go up to +2, down to 0, up to +3, and now it's on +1.
18. 648.▲
    by Retr0id 5mo ago

    The escalating nerfs of "cybersecurity" topics is incredibly frustrating. Opus 4.6 had boundaries that seemed reasonable to me but 4.7+ turned it into a moralizing asshole. It'd be less bad if it just gave an error message, b
19. 649.▲
    by Retr0id 5mo ago

    I made a golang socks5 proxy that routes traffic to different VPSes (or the default gateway) based on hostname, over mutual-TLS tunnels, authenticated using ed25519 keypairs shared out-of-band. The "client" and the "server&qu
20. 650.▲
    by Retr0id 5mo ago

    I just never look in ~
21. 651.▲
    by Retr0id 5mo ago

    It's just a regular hash table with linear probing https://github.com/yon-language/yon/blob/aa4617ced3abc92ac53... The only relation to the Leech group appears to be the number of slots. When a table get
22. 652.▲
    by Retr0id 5mo ago

    I understand that you can hash any object into a 196,560-slot space (that's how regular hash tables work), but I'm not sure why you'd want to do that. Regular hash tables can be resized when they get full (or less-full),
23. 653.▲
    by Retr0id 5mo ago

    I have a strong understanding of content addressing, memory allocation, data structures, and a superficial understanding of lattices. It is not clear how one has anything to do with the other. If there is any meaningful link or benefit, it
24. 654.▲
    by Retr0id 5mo ago

    What is the core novelty?
25. 655.▲
    by Retr0id 5mo ago

    Taking it at face value, you'd spawn a subprocess to handle each web request and exit once it's done.
26. 656.▲
    by Retr0id 5mo ago

    > exactly 196,560 slots per heap > Allocation hashes the bytes; identical content returns the existing slot What happens when two distinct objects hash to the same slot? (This is inevitable by the pigeonhole principle, and highly like
27. 657.▲
    by Retr0id 5mo ago

    The whole premise of what's being described here is to bake the weights into the silicon. That isn't what I'd describe as vendor lock-in, any more than I'd describe a CPU that can only execute ARM instructions as vendor
28. 658.▲
    by Retr0id 5mo ago

    What kind of vendor locking would be possible?
29. 659.▲
    by Retr0id 5mo ago

    If only Anthropic had some kind of automated tool that was good at porting software
30. 660.▲
    by Retr0id 5mo ago

    I never said that.