just in case you didn’t know that “it’s only for code reviews, I don’t see the problem” is always bad faith horseshit: phoronix.com/news/Linus-Torval…

> Linux creator Linus Torvalds has been known to use AI to help with kernel debugging and outside of the kernel tree has also done some vibe coding with Google's Antigravity. He has previously said he views AI as a useful tool. It turns out he's even now using Claude Code for working on new Linux kernel features.

> "The code, the test programs, the review behind the list above and the first draft of this text were done with an LLM coding assistant (Claude Code) under my direction. The measurements were run on my machine."

so just a full vibe, done with the LLM that can’t make a web browser, C compiler, rsync features that aren’t riddled with bugs, or JavaScript runtime

this is the shit we were warning you about

was there a reason why you didn’t listen?

in reply to [object Object]

I cannot emphasize enough how horrid Claude Code (and every other LLM) is at systems code

it’s horrid at everything, but when torvalds pretends it’s any good for kernel work he’s trying to talk over a clear pattern of failures

I cannot name a single vibed piece of systems code that actually works with anything approaching basic reliability

torvalds’ proof that any of this shit does anything at all is that:
- it shit out a generic python UI for one of his side projects
- the corporations funding him via the Linux Foundation and keeping him fat in tokens say it works great

that’s it. that’s where all his confidence comes from. that and him being a mediocre little fucker.

in reply to [object Object]

surely this time. surely the horrid thing that can’t produce stable software for the xteink x4, to the point where people are abandoning their devices out of frustration, will be fantastic at making features for a much larger, much much more complicated kernel that has to work across a gigantic variety of hardware

surely this time

reshared this

in reply to [object Object]

god, remember all the little fuckers who used torvalds’ “not a social warrior project” proclamation as entryism to push for LLMs in other projects “for reviews only” because that’s what the kernel was doing

all of those little fuckers just got the outcome they were waiting for

This entry was edited (Monday, October 12, 2026, 6:27 PM)

reshared this

in reply to [object Object]

Great fantastic definitely not freaking about what the fuck I am gonna do when the Linux kernel just stops working it's not like there just aren't any alternatives I could switch to right like NetBSD oh wait that doesn't fucking work on my hardware uhhhhhhhhhhhhhh.

reshared this

in reply to Jacob Alexander Tice

And even if alternatives DID exist (which they still don't), uh, well I do many things on my computer. I play games. I make and edit videos. I use it as an amp sim to play guitar and bass. And a lot of my software doesn't work on things that aren't Linux, Windows, or macOS. I don't own a Mac. I am never installing Windows again.

reshared this

in reply to VoidZeroOne

I genuinely have no idea how to do that, what I would even use (you'd think that after existing for nearly 6 years that Zen 3 would have wide ranging support and you'd be dead wrong as outside of Linux everything open source is still various states of fucked on Zen 3, not even to mention the dire GPU situation which Because Nvidia™ is somehow even more embarrassing thanks Nvidia you assholes), how it would even stack, anything.

[object Object] reshared this.

in reply to VoidZeroOne

I dunno. Last time I tried was about a year ago and NetBSD kernel panicked on my 5700X3D every time I tried to boot. FreeBSD might work but they're also drowning in slop. OpenBSD doesn't work for me but for a different, funnier reason: wifi just doesn't work. And yes. I need WiFi to work. This is not up for debate. I do not own this house and do not have permission to run Ethernet up to my bedroom. Not that it matters because it looks like OpenBSD ALSO STILL is drowning in slop too.
in reply to Jacob Alexander Tice

Why the fuck is it so hard for so many projects to just… not use Claude Code or whatever other LLM extrusion machines? Like I don't even care if they use chatbots elsewhere in their lives at this point, I just don't understand why they'd want to let it touch their actual computer. And don't these things cost a lot of money too? And what is gonna happen when they can't keep artificially keeping prices down?
in reply to Duke of Germany 💫

@duke_of_germany CrossPoint is vibed (and so are any alternatives as far as I’ve heard, though I don’t know what those are as I held off on buying one of these) and from what I’ve heard the standard firmware probably(?) isn’t, but is horrid in various other ways

this is very much the type of device that shines once it gets FOSS firmware, but unfortunately it is 2026

in reply to Bredroll

@Bredroll that is the problem. LLM / vibe coding does not require knowledge of how a system works. It will create high level code that may work, but the "coder" has no idea why. When bugs come up then there is no one with the knowledge to fix it. Can one of these coders troubleshoot a memory or stack issue?

Second to your point, I have yet to see these engines create low level code that is required for high speed, close to hardware applications (routers, infrastructure, kernel, drivers, etc.).

Reminds me of a PHP script that is 200 lines and a Perl script that is 5 lines that does the same thing. In the end the Perl script is faster.

We may get there but by that time will there be any programmers left?

in reply to [object Object]

This post about the Zed vibecoded editor is quite informative: goto.liten.app/@krig/statuses/…


So far in the zed (now gram) codebase I have found three separate theme pickers, two separate markdown parsers and renderers, two separate language server installation implementations, one broken capability/permission system that only applies to one of the language server implementations and a system for loading settings from JSON that leveraged monomorphization bloat (convoluted generic async rust code strikes again) to generate 10 MB of code in the binary... and probably even more nonsense that I'm not remembering now. Trying to fix anything is like pulling at a thread in a ball of yarn but the ball of yarn is the ball from Katamari Damacy and out tumbles three cows, a high rise building and a brass orchestra

in reply to Jernej Simončič �

@jernej__s I've been using Gram and it's pretty nice. I don't have really intensive demands for a text editor, I just need it to edit, you know, text. And not be too much of a pain.

Not too surprising about Linus being all-in for AI, given his whole history of Not Playing Well with Others.

in reply to Stu

@tehstu these hallucinated people are really considering claude as a coding slave that does intelligent things on their behalf, like an intern they can coach.

I hate that Torvalds decided to do this, first because it's ridiculous, and more importantly, because he's showing the way to many developers that will take this as an authorization to vibe code the kernel in a large scale.

@Stu

[object Object] reshared this.

in reply to [object Object]

AI abuse

Sensitive content

María Arias de Reyna reshared this.

in reply to Bob Mottram ✅

AI abuse

Sensitive content

This entry was edited (Monday, October 12, 2026, 11:21 PM)

[object Object] reshared this.

in reply to Krampus 🌰

@castanea_jo universeodon.com/@mathew/11739…


Wikimedia signs a business deal with OpenAI to sell the content we built, and OpenAI attacks their web tools?

Ruby goes all in on genAI, and OpenAI inserts malware into RubyGems packages?

Linus invites AI use for kernel code, and kernel.org gets hammered by AI bots?

I'm going to need to stock up on strings for my tiny violin.


in reply to [object Object]

I take some solace in knowing that, while my C chops are not nearly as good as Linus's, I will soon surpass him in skill as he gradually loses any semblance of technical ability by outsourcing it to the bobot and I don't.
This entry was edited (Monday, October 12, 2026, 6:17 PM)

reshared this

in reply to _Έλλεν Εμίλια Ά. Arson ζ.

@fogti it turns out a kernel can do memory management really fast when the LLM strips out all the safety checks and falsifies all of the tests

reshared this

in reply to [object Object]

@fogti

Well, that's perfectly inline with the AI-first mentality.

It's all they want at work.

"More, faster."

Yes, but the more we are generating is bad. It's more bad than it's been and we're making it faster.

"But is faster?"

Well, we're closing old tickets but opening a lot more for each ticket we close. Also... The robot can't figure out how to make good tickets so support is completely busy making ticket for us to _break_ things faster than before.

"Is faster. Do more."

👀

reshared this

in reply to Matt Hall

I gotta be honest, I blame all of this on the prevalence of private equity. The "product" no longer competes in a wide market but just to a few people at a time. In a space where the charisma (or whatever it is these cardboard-cut-out tech CEOs _have_) can overwhelm the logic and reason _over time_ that a marketplace would reveal tech CEOs are filling _their_ pockets, and tying up _all_ the available capital on ponzi scheme after ponzi scheme.

Deb Nam-Krane reshared this.

in reply to Matt Hall

We're told it's inevitable by the fifteen weirdos making the decision to hold the world captive in order to enrich themselves.

...

These people are the literal real-world manifestation of the vampire overlord trope... We all _see_ that, right?

Stakes through the heart for all of them.

reshared this

in reply to [object Object]

I like how "under my direction" is basically a disguised appeal to authority: people are supposed to be specifically impressed that Linus Torvalds, god among sheep, resorted to sludge-coding. I mean, it's special when he does it amirite?
in reply to Mx. Chara Aznable of Pnictogen

@mxchara linus torvalds, the programming god boy genius who either refused or failed to learn basic frontend python for an easy side project so he had “google antigravity”, whatever the fuck that is, generate it, and that experience convinced him LLMs were amazing at kernel code

under the direction of the guy who can’t figure out python

reshared this

in reply to [object Object]

My favorite part about that Python… thing? By far it's that, ok, even assuming Linus Torvalds of all fucking people doesn't know how to write Python (big assumption), he doesn't need to write Python for it? Nobody put a gun up to his head and said "write this in Python?" So what the fuck?
This entry was edited (Monday, October 12, 2026, 8:12 PM)
in reply to [object Object]

It's funny when people say things like, "Even *Linus* is doing [thing]!" to justify problematic stuff.

This is the guy who was a happy BitKeeper user until Tridge blew up the relationship w/ McVoy and the kernel community. He's not some shining beacon of free-softwareness and ethical purity, he's someone who will look the other way on ethical issues for pragmatic reasons. That's just who he is.

This entry was edited (Monday, October 12, 2026, 8:37 PM)

Deb Nam-Krane reshared this.

Unknown parent

@cyberquixote @theorangetheme @fogti IMO the kernel has stayed relatively slim until recently, and old code does not necessarily equal bad code, especially with something as low-level as this - a solid chunk of computer firmware is still code from the 80s
in reply to Taffer 🇨🇦🎃👻

@Taffer @catsalad I tried setting up a laptop with FreeBSD to do everything I do on my LMDE laptop, and the problem is there are some things that only run if I install the Linux subsystem (I forget the official name of that), and even then they're a bit wobbly. I decided it was sort of missing the point of moving away from Linux because of the AI use if I still have to install the Linux kernel and core just to use what I need to on BSD.
I do still have a laptop with GhostBSD, though.
⇧