this is the one and only reason that the Linux kernel is going out of its way to hide which commits are LLM generated by abolishing the Assisted-By tag, simultaneously with torvalds fully slopcoding a core “improvement” to the kernel’s memory management
if you don’t believe me, here’s part of what the diff removed from the kernel’s docs:
> -When AI tools contribute to kernel development, proper attribution > -helps track the evolving role of AI in the development process. > -Contributions should include an Assisted-by tag in the following format::
> I asked whether that risked cutting off bright new developers that we would rather hang onto; Torvalds answered that the kernel project only wants serious developers. That includes having access to enough resources to function in this world. > [...] > Torvalds added that he does not want to scare off people who want to write code; his message is that you have to either put in the effort to learn, or put a lot of money into an LLM to participate.
Linus Torvalds prefers if you can’t contribute to the kernel without pouring money into tokens for Claude Code
if you don’t pay a shit ton of money for tokens, you aren’t a serious developer according to him
the “learn” thing here is a distraction: in this context, you’ll be learning to clean up slop until you either burn out (very quickly at that) or pay for tokens and cognitively surrender to the bot
@techokami > Laurent Pinchart said that some developers are frustrated that they have to justify their work against criticisms from Sashiko while other, LLM-generated patches appear to be accepted immediately. Abeni defended the automated reviews, though, saying that the number of patches introducing bugs has dropped considerably. Zijlstra said that, while reading LLM-generated reviews "shuts my brain down", he still thought that the tool is useful; it finds real problems.
@techokami sashiko is their weird anthropomorphizing name for the LLM that runs on their CI and generates lists of “problems” with patches by the way, a thoroughly worthless thing in my long experience
as with all other systems like this, it thinks LLM code is a-ok but nitpicks code written by people to death
this is one of the mechanisms that burns you out fast when a project goes in on LLMs
@zzt@clayote@hipsterelectron@Colman@techokami me when I learn about LSPs for the first time (seriously it's so useful I spent way too long doing stuff like this manually but it's invaluable in languages like Python especially
@Lunaphied@zzt@clayote@Colman@techokami my emacs package makes ripgrep matches editable and it is strictly more powerful and useful in every possible way along with not tying me to a specific compiler or build system and even working for projects i don't have to build in advance as a result
@hipsterelectron@Lunaphied@clayote@Colman@techokami damn, fuck, that sounds excellent. I really like editable matching as an interface but the only emacs package I knew that did it was really barebones and a bit slow
@zzt@Lunaphied@clayote@Colman@techokami 100 stars on github somehow although the byte compiler is weird about my macro bullshit sometimes lmk if you want any features or shit breaks https://codeberg.org/cosmicexplorer/helm-rg i used to recommend deadgrep by wilfred hughes which was a really thoughtful bit of UI i was too lazy to do better than but he's a slopper now. if you can convince me to work on this again i would feel better about actually defeating him in the marketplace of ideas though
@zzt@Lunaphied@clayote@Colman@techokami if you mean vertico for general completion then yes, it will pull in helm as a library and use it for the UI but that won't infect anything else. tbh i'm incredibly pissed off at helm after they tried to tell MELPA to delist my package since they made one themselves. thierry volpiatto can fite me irl
@zzt@Lunaphied@clayote@Colman@techokami helm also tends to just break shit every so often for fun i just need to get my act together and move to something else. never heard of vertico before tbh i'll have to check it out
but anyway yes I’m gonna pull your package in next time I regenerate my session!
(some inside baseball on how my distro currently works, I build emacs and all its packages in one step but the old one gets orphaned so I have to restart my session right after I do it. what I need to do is make sessions user packages and hold a symlink to the currently running session somewhere but I have not done this yet)
@zzt@hipsterelectron tangentially related but I am always confused by the minad suite. I have never been able to figure out/remember what is needed for what and I end up installing all of them and getting further confused.
@hipsterelectron whats the deal with Helm? I have heard of it but I dont know if its meant to be a specific package that cannot be installed with others or something. Like I have heard ido and ivy do similar things
@zardoz so all of these plug into completing-read and shit. the reason this is confusing is because if you M-x helm-mode it takes over all your completions everywhere and it has some incredibly half-assed defaults (i personally find it unusable for C-x C-f, which i use ido for, which also has fantastic LRU ranking for this while otherwise going by lex sort)
@techokami@zzt "LLMs are not the problem, it's the users' fault" is basically the same argument used by pro-gun lobbyists in the US. Tells you just how much they care about fixing the problem.
@zzt@techokami “Linus Torvalds believes that blaming LLMs for the kernel community's problems is wrong, when the problem is LLM users who do not understand how to use them or how the kernel works.”
@cjgibilisco@techokami channeling all the managers shouting over their project being on fire with PIPs for everyone who failed them by not LLMing hard enough
@cjgibilisco@zzt@techokami I’ve noticed this is a common pattern with those who become AI dependent: the AI cannot fail, it is always the humans who have failed the AI. The final form of cognitive surrender.
@zzt@techokami I’m repeating the prediction I made half an hour ago: it won’t be long before Linus goes on a rant that somehow culminates in “too many brown contributors to FOSS”
@zzt@techokami The most fun thing about them at work is that they manage to follow "best practices" better than most humans but do it in service to code that otherwise doesn't work.
The cleanest, most SOLID and DRY implementation of nothing in particular that you've ever seen.
@zzt this whole thing is just a multitude of fuckedness
THIS hurt the most to read: > Torvalds and others repeatedly said low-quality contributions from new developers should receive low priority, and that the community should stop trying to make contribution easy. Many developers said, repeatedly, that trying to mentor those developers is almost always a waste of time; they do not listen.
@zzt also it sounds like if you want to become a contributor now, hope you have lots of money for plane flights and hotel rooms to frequent conferences enough to be noticed
@johnzajac@techokami@zzt This is complete budget related company think of "I don't need slower juniors, someone else will train them!" - but in a context where it makes zero sense unless you accept Linux already being completely undermined by corporate desires (it is).
The NPO world has always been a tax haven, and since the 1980s has been growth-pilled to the point where working in NPOs/Foundations is nearly indistinguishable from working for tech companies, except that because of the earnest nature of people attracted to that sector, there's significantly more abuse and overwork that goes unreported and unaddressed.
@techokami@zzt This is a really bizzare combination of obnoxious elitism and AI psychosis, by the way.
"you have to either put in the effort to learn, or put a lot of money into an LLM to participate."
In other words, we have impossible standards, but they don't apply if you're rich and sucking up to Anthropic like we are.
Linux is fascist. It would be irresponsible to deny, a project entirely taken over by fascist techbros and their fascist corporate backers is fascist and dead walking.
@techokami@zzt makes me glad i never ended up learning linux kernel dev 😬
though hey, this might just break me out of the sort of programming burnout i've been having and get me to contribute to netbsd so... thanks linus, i guess?
@techokami@zzt I won't be surprised if Linus ends up switching Linux to source-avaliable model or even outright closing sources whatsoever and selling licenses to companies directly. Unfortunately my work/life flow (including VR) is only possible on either Windows or Linux, at least for now.
> He has been playing with an LLM on a separate project for several months to learn about the technology, but hadn't used one on the kernel until last week. He took an old patch and had the LLM revise it; the result has made him unwilling to work without an LLM. He plans to post that series at some point. Why bother, he asked, to search for all of the call sites of a function he has just modified, when the tool can fix them all automatically? But, he said, it took six months to set up the tool to not produce bad responses. If you are a competent developer, he said, an LLM makes a great tool, but only after that effort has been put in to set it up. He resisted suggestions to post his LLM-training materials, saying that each developer needs to go through that process separately.
@zzt@techokami "If you are a competent developer, he said, an LLM makes a great tool, but only after that effort has been put in to set it up. He resisted suggestions to post his LLM-training materials, saying that each developer needs to go through that process separately."
Of course, he won’t share his setup. Otherwise, he might lose favorite excuse: “You’re just prompting it wrong.” At this point, Torvalds is acting like a complete clown.
@remivox@zzt@techokami It's the completely natural progression lol, when you outsource everything to the LLM, there is nothing special about you. You have no value as a developer. The software does it all for you. Your particular configuration of the software is all you have left to justify your existence. Of course he's not going to share it.
@tael@zzt@techokami He won’t share it not because it’s some kind of trade secret or requires deep expertise, but because the whole house of cards would come tumbling down. I guarantee there’s nothing special about his setup.
@zzt@techokami yeah reading this, there are some people in the replies fighting the good fight.
I think my opinion now is that Linux is too large, it is insufficiently modular (building out of tree drivers used to be a lot easier) and lkml has long had a reputation for being unfriendly to newcomers. It's a psychological problem more than a technical one, but that barrier exists because the project is huge and has no architectural direction aimed at making it more maintainable.
