How much slop is in Linux now? Ehh we decided not to tell you
https://lore.kernel.org/lkml/20261009083137.3-1-sashal@kernel.org/
How much slop is in Linux now? Ehh we decided not to tell you
https://lore.kernel.org/lkml/20261009083137.3-1-sashal@kernel.org/
@davidgerard it's just a tool, bro, you don't need to know what tool i used, bro.
@davidgerard That will make it tricky once all LLM assisted content is ruled as copyright infringement and needs to be stripped out.
@kbm0 @davidgerard i wonder if that's part of the reason for this change
@tmw @kbm0 @davidgerard Yeah, I don't know for sure, but considering Linux is basically a corporate project, this change will also provide welcome ambiguity for lawyers that either want to enforce copyright claim on the kernel or deny copyright infringement. The ambiguity will also be useful for lobbyists when arguing for favorable changes to laws, "It's already deeply embedded, and there's no way to tell what was written by what!"
@kbm0 @davidgerard @zzt resetting the pointer to 2025 is fine by me. I don’t just mean the Linux kernel. I mean everything.
Actually, if we have to go through all this shit again, maybe not..
@davidgerard "We didn't need the thermometer anyway."
@davidgerard@circumstances.run I don't understand. If LLM-code is fine, why go out of your way to obfuscate what is or isn't LLM-generated code? All this does is erode trust.
I am really baffled by the speed of how fast this project goes down the shitter 😥
/e: yet there's still a glimmer of hope that torvalds is going to get over his "it's just a tool" phase. I heard him already complaining about the volume of bogus pull requests etc.
I'm afraid that glimmer is looking more and more like a mirage...
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
His ignorance is, at this point, wholly wilful.
oh damn - reading that mail made me literally facepalm. instantly going ad-hominem and argueing like a 5th grader. 🙄
@spitfire @McCovican What we need is a fork.
New Linux and Slop Linux need to be developed separately.
@yora @spitfire @McCovican There are other, under-resourced operating systems that are more worthy of the attention than this insecure C fossil.
@landelare @spitfire @McCovican Sure. If we make all the Linux software run on it I'd be happy to use it.
@yora @landelare @spitfire @McCovican
Redox OS @redox_os@icosahedron.websit (strong anti AI policy) already runs a ton, including a graphical desktop!
i've been wondering when this would get mentioned. i wish i had the free time to work on such a project, but although i learned C in the early 90's i've never done any kernel work so even the thought of using what little free time i have sounds completely daunting.
@McCovican @spitfire @davidgerard unfortunately, the trend more and more agentic AI contributions to the kernel and less and less humans vetted to make those AI contributions. Linux will stop being a community project, and it will be more like a few key maintainers and companies managing AI agents to mutate the code, "open source" only in the sense that, I guess, you can look at the code with your eyeballs if you want
@McCovican @spitfire @davidgerard
Wait is he arguing that deterministic and non-deterministic are the same thing?
@zedaardv yep - that's what he's doing. The mind boggles!
@spitfire
Damn! what a muppet.
@spitfire @davidgerard It's not often that I'm rooting for the AI bros, but if they could submit even more bogus PRs, that would be very nice of them in this particular instance
@landelare the tried and tested "flood the zone with shit" gambit, yes? 😉
@spitfire I'm not very well-versed in these techniques, but I'll happily watch the wrestling match
@landelare just for the record - me neither 😄
/e don't mind if I save that picture you posted. it's lovely :heart:
@spitfire I have stolen—I mean, obtained it to train my own AI, so that it's legal—it myself, go ahead
@spitfire @davidgerard I’m afraid this will not end in Torvalds getting over whatever this is but us realizing he’s a full blown neonazi or something.
@CynAq @spitfire @davidgerard I also don't think he gets over it. He just seems so ultra utility-focused and small scoped. Who cares it's drowning the reviewers, or leading to widespread problems mental wise. To him it feels faster right now, and probably does in the future when he vibe codes some other random bullshit. Review burden, understanding of systems and code? Who needs it, go pay Anthropic to save them one more day until they burn, bitches.
@dimenzio @CynAq @davidgerard after catching up on all his recent posts I am going to retract my "glimmer of hope" statement. The man is deteriorating fast.
@CynAq @spitfire @davidgerard He did almost complain about "Social Justice Warriors" a while ago, which is an old nazi 4channer phrase with the same meaning that the right uses "woke" today.
He just dropped the "justice" part to not make it too obvious.
@Kiloku @CynAq @spitfire @davidgerard Yeah, when he dropped the part about it never being a social justice project and would never *be* motivated by social justice, that was the end of my enthusiasm.
If social justice isn't a valid motivation in his mind, he can't be trusted.
@davidgerard Problem solved ✨️
@davidgerard And _the guy_ has lost his marbles.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
@davidgerard And as things get less and less stable, they'll deny it being down to the slop code. Comments sections will be packed with slopbros going 'Nuh huh, you have no way to know it was down to AI, you're just guessing! Correlation is not causation! Checkmate, luddites!'
@davidgerard@circumstances.run Sorry, I can't read it, I have thrown all of my computers into the sea and never wish to touch one again.
@paul a wise choice
good thing masto works on an abacus now
@davidgerard if not for explicites audit reason, there is no need to have it. Who commit is the owner and responsible.
@Opmodoro 0 follower spam sockpuppet
@davidgerard I have always perceived these footnotes and tags as free advertisement for multi billion dollar corporations that already make the news every other day with new bullshit hacks and mishaps because plastering the company name across everything there is is the only thing that matters, they want to be everywhere all the time. Therefore I'm happy with less ad spam, in particular due to the unenforcable, voluntary nature of these tags, a generic "assisted by: LLM" would be kinda useless.
@davidgerard Final straw for me. I am, as Emperor Penguin suggested, "forking off"...to #NetBSD
@davidgerard What I don’t understand is why this seems to be the only place the kernel is willing to embrace anything modern. I mean, there was a maintainers’ revolt over Rust and development still happens over email with format-patch, but LLMs come in and suddenly it’s all about accepting the future of software development.
@davidgerard a reasonable conclusion seems to be that unless explicitly stated otherwise, assume by default that it was probably assisted by some LLM.
@davidgerard ITT: not a single person on fedi read any of the responses in the massive thread lol
Also for context: someone made this patch without discussion. There will be discussion. Reading the thread (click next at the bottom) is important before having an emotional response.
@elebertus I read the LWN report. This is consensus.
@davidgerard Might as well go to TempleOS if we are doing faith based development
@davidgerard
I (not a lawyer) will claim that this is against the EU's AI Act, as there all slop-generated text should be marked as slop-generated.
@davidgerard + the irony of the anti-bot page when first accessing this link! 😆