Oh my god. No.

That's not what this argument is about. The JIT is non-deterministic based on the hot path yes. But the way that code gets turned into code IS.

The compilation of instructions is deterministic. The control flow and WHEN it gets compiled may not be, and it may smartly choose not to compile certain things based on that hot path but that's not the same.

A compiler and interpreter is a SYSTEM. a LLM is not.

Much of open-source development relies on the honor system. There is likewise no way to tell, just by examining code, that it was not plagiarized.

Honor that needs to be enforced is not honor. It is something less.

AI has fundamentally changed how you do CS in industry, it’s completely unavoidable at this point and not using it in some capacity is doing yourself a massive disservice

If it was in truth inevitable then there would be no need to constantly shriek "It is inevitable!"

There is also evidence that deferring critical thinking to an LLM causes cognitive atrophy.

https://arxiv.org/abs/2506.08872

It relies on legality, not honor. E.G. with GPL, or code with a proprietary license. It's the threat of losing their job, getting sued or whatever that discourages plagarism.

Sure, but, like, you wouldn't.

On that note, OpenBSD appears to be a likely refuge now that Linux has been abandoned to slop.

https://marc.info/?l=openbsd-tech&m=177411620801633&w=2

He might be a genius in the past. He has always been arrogant. Now he's just stupid and arrogant.

OMG, yes! Why didn't I think of that before!? I'm gonna be a psychology doctor and rename all my hallucinating patients as "Claude" and BAM! They become genius now!

Neither does the use of an LLM mean you do not understand thoroughly what you submit. He doesn't give a crap how the code was created, as long as you, the human (I hope), understand what you are submitting and can take accountability for it. You are still responsible for whatever crap comes out of an LLM, the same way that you are responsible for what comes out of the compiler. Doesn't mean you shouldn't use compilers or LLMs.

Just move to NetBSD already. I guarantee you that Linux will become shit in a year. The best outcome could be the core team abandoning Torvals and forking it under a new name.

So Linus does not respect me. Nice.

It’s mutual.

To be clear, I haven't downvoted you.
I think it's a valid reason to hate the technology but well below others like deskilling, proliferation of spam and propaganda, etc. If these data centers were all run on solar power then it'd be less of an issue.

That specifically might not on its own, but there are also a lot of other factors that definitely mean responsible people don't drive. Even local LLMs map neatly onto EVs - they're not as polluting day to day, but still require plenty of environmental damage to be built.

I agree with you here (and also cool project/idea). What I'm trying to get at with my analogy is that, if you understand a programme's source code, you understand and are able to predict its behaviour in a meaningful way (although of course you lose some amount of knowledge of what the processor exactly does with which registers etc). When you only know the prompt used to generate a programme, you can't do that. That's why you have to manually check the code an LLM writes, but not the assembly generated by a compiler.

My first couple of programming classes were in Pascal. Then there was the third class where the professors had written their own pseudo language that existed only on the chalkboard and in their own heads. At least I learned recursion from them. I had to teach myself every language I ever used on the job. I was EE instead of CS, so I suppose I shouldn't complain too hard, perhaps there were more practical classes deeper into the CS tree.

Backdoors through model poisoning is so evil, I love it.

The other practical side of this is enforceability.

There is likewise no way to tell, just by examining code, that it was not plagiarized.

Non-point elements removed to help you focus.

Thanks

The problem with "LLM requires skill to use correctly." is that LLM's are not being sold as a skill but rather the end of skill.

I had the misfortune of working a lot with claude skills for the past few months.
I run the exact same skill, on the exact same codebase and it doesn't yield the same results. Sometimes it straight up ignores parts my prompt. Other times it ignores the skill file entirely. And then I ask Claude, why did it do X thing that it wasn't supposed to do? The answer is "I apologize, I should have blabla". It is extremely unreliable and unpredictable.
Now I'm stuck with a feature that I'm responsible for that occasionally goes rogue. Fun times.

They're bad at dates and age too. That's absolutely true. Like I said: always verify. They're good for finding stuff, not at being correct.

BASED LINUS

He used to be cool, but then he started following some corporate hype trends that make me suspicious, I think he is nearing retirement and doesn't want to fight corpos anymore.

The moment he accepted Rust officially on the kernel was the time I knew he got a bit too old to care anymore.. This is just a continuation.

Also is so curious to see the venn diagram of people supporting Rust and AI being mostly a circle. My theory is that Rust takes every piece of joy out of programming and thus the idea of never programming again becomes so attractive if you are surrounded by Rust.

An unpredictable tool is a useless tool. A hammer does what it says it will: apply force at the head. AI straight up lies, is objectively wrong, or doesnt work. Thats a garbage tool.

Eventually youre going to ask it to do something you dont know how to do, and you wont be able to fact check it. Thats when the harm happens (or just mild inconvenience, but in my professional case, harm).

If the linux kernel breaks due to a bug introduced by AI or built on a pile of code no one truly understands, we're set back anyways. Better to avoid that and make sure all code is human written/understood

I'm finding the allure of AI is particularly strong for people who have risen to positions of power where they had to stop doing the work themselves. Instead, they had to work through other people to get bigger projects done.

The allure is that you can just say what you want done, and it does it. If it's done it "wrong" you can just tell it it's wrong and not have to worry about arguing the technical differences. As such you feel like you get closer to what you wanted, rather than somebody else's take on what you wanted.

The things is, when you are removed from the code, you can only work at a high level. That person you asked to do something is considering 10 other low level concerns that you didn't consider. I think Linus's recent patch series is a good example. He saw a potential optimisation, did a proof of concept but missed a race condition. He's now handed it over to the person that normally works in that area and knows the details. Can the optimisation actually work? Who knows?

Linus accepting AI now says more about how far the technology has come than anything else. AI today is not the same as AI a year ago.

Let me try that again:

if your an accountant and you boss copy your accounting sheets and do not pay you, it is not stealing? You both have the file, for you it won’t make you any money,

The copying part is gone, yet the problem stays the same, so the copying was never the problem.

This is the thing I'm trying to advocate people to understand. It's not that they were wrong about AI a year ago - AI has changed in radical ways people couldn't confidently predict within just a year.

The AI coding agents a year ago DID create slop that would dillute your code base. The AI coding agents today generally don't.

It's becoming increasingly less valuable to combat your LLM and scrutinize it's output. You can provide more market value by assuming it's 95% correct, doing some manual verification, and architecting it in a way that leads to low integration complexity. The OOP principles are still relevant - abstract, encapsulate, etc.

I swear I'm not an AI bro despite the above comment being somewhat pro-AI. I'm just trying to be pragmatic about the state of our field.

It's not 'based' to be pro-AI.

This is the standard position for anyone that got mega-rich on tech and still has high tech market exposure, such as Linus.

Morality of using resources that someone else stole is 'muddy' to say the least but the majority of modern society would not function if we didn't look the other way.

For example, I hope you're not living anywhere in NA or most of Africa or many areas in Europe or some areas in Asia. Because if you are, you're probably standing on stolen resources.

you talked about labor theft, why do you think the boss should pay you? An I am not saying "should" as society should be like that, I agree with no private property and no money -> should as right now is it OK or not to make money from other people works : this is exploitation.

I argue that the "steal" of today's LLM is the same exploitation of worker as usually described, here outside the social contract of librism and free software.

AI is not a tool. Its an addiction

I totally agree. I'm watching people lose their drive to create and it's so sad. They're getting high off the satisfaction of "creating" without the work.

...and then they lose access to their wündertool and they can't do anything.

Yeah dude licking corporate boot is so based

Not to mention it was an issue kernel developers had spent 2 years trying to solve unsuccessfully.

A calculator isn't a sycophant text extruder offered by megacorps via predatory tactics and doesn't require half a little of water to operate last time I checked

Appeal to authority. Linus was always a fucking moron.

I don’t know how you came up with your rust <-> ai theory, that hasn’t been my experience at all.

Multiple Pascal classes for EE thats interesting. You're not much older than I am and when we did Pascal my teacher said something like its not really used as much now but its a great learning language, but the 2nd class just moved onto C.

I had a friend doing EE and they had Perl as their programming class a few years later.

If I had to guess id say its Python now? That'd be practical as well.

I also leaned every language i used outside of school on my own.

Nearly all evaluation steps for an LLM I produce some level of entropy. Yes you can evaluate an LLM model deterministically, but we don't. We even tune the levels to get good results. Model cards have sampling parameters supplied to give "the best" results. For example, Qwen 3.8 flash next gives

We recommend using the following sets of sampling parameters for generation:

Thinking Mode: temperature=1.0, top_p=0.95, top_k=20, min_p=0.0, presence_penalty=0.0, repetition_penalty=1.0

Instruct (or non-thinking) mode: temperature=0.7, top_p=0.80, top_k=20, min_p=0.0, presence_penalty=1.5, repetition_penalty=1.0

Replaying the seed will only get you the same result using a particular engine and a particular random number generator. Run it on llama.cpp Vs Strata and you can get different results. Different libc, different results.

A compiler is systematically deterministic. No calls to rand()

I personally like Rust fine, it's not my favorite or my strongest language but we've suffered stuff like Java for decades.

If you're too much of a black/white thinker to be able to have an intelligent and mature conversation, there is nothing I can say to change your emotional response.

With that being said:
Fuck big tech, fuck how AI is being used by it and large companies. Period. Full stop.

Can't shame someone for being born in a certain place.

I'm not pro-AI.

It's based because he's right. He's looking at this from a tool standpoint. But too many people are lost the emotional sauce, unable to see the engineering and usefulness beyond how terrible it's being used. Linus is an engineer, not a politician, not a fuckin business man... You're all coming at him with the myopic and emotional vigor of evangelists.

We should put the responsibility on the person. Not all funnel into these fucking echo chambers, tone policing everyone's little divergences from whatever stupid fucking camp you're part of.

Linus made linux, wtf have you done?? Why should I put any stock into what you have to say?? What long running successes and judgment do you have to show???

You cant put the shit back in the horse, but have fun trying.

using proprietary tech whose purpose is to hook you and then enshittify after you become reliant on it

(Opens trenchcoat in alley)
"Psst... Hey guy...ya want uncanny waifu images? Ya wanna code faster? Ya wanna write your own kernel? Make a game without thinking too hard? Heh heh...first thousand tokens are free, buddy. Heh heh heh. (Wink)",

Edit: lol wonder the motivations for the downvotes. I was equating LLM corpos with shady drug dealers hooking victims with empty promises. Didn't hit I guess? Lol

This is just one example. I’m not going to keeping punting just for you to move the goal posts.

The answer is certainly not to embrace "AI" in it's current forms or allow for big tech monopolists to continue their wasteful and destructive infrastructure buildout. The oligarchy isn't going to regulate itself to worker self-determination either. Nothing short of the death of capitalism will do.

Thanks for your contribution to the pile.

It's a mix of conquered, purchased and stolen land

Even considering rhetorical goals and context, without "I particularly dislike this guy", it's an enormous leap to twist those words that way.

Stop blaming the pandas.

marketing departments have always lied