Bad News Monday: We Are Losing Linux
As if waking up on a Monday wasn't bad enough, today I got the news that the Liux kernel is now1 officially accepting AI slop [LWN]. That's the last thing you wanna hear these days, and the whole report makes for a frankly grim read.
Linux holds up a great portion of the internet; just because you haven't heard of it, doesn't mean it isn't there. Most servers run on Linux, and Android is a Linux derivative. More broadly, Linux is a family of operating systems that are all based on the same "kernel", which is the core component that talks directly to the hardware and support everything that happens downstream. It's a big deal, and an area where the margin of errors are vanishingly small, which is why it has always been built and maintained by people who strictly know what they are doing. Except recently they have been flooded with low-quality submissions to the code made by LLMs, and the solutions they came up with in their yearly summit are... not great.
Linus Torvalds2 said that "The problem [...] is that we are getting increasing numbers of patches from people who do not know the kernel. By pointing the finger at LLMs, maintainers are blaming the wrong thing." [LWN] As if LLMs weren't the reason every clown in the world suddenly thinks they are a veteran developer capable of putting their dirty paws in the kernel. But Torvalds' bizarre answer to this was that "the kernel project only wants serious developers. That includes having access to enough resources to function in this world. [...] you have to either put in the effort to learn, or put a lot of money into an LLM to participate."
The kernel is a very complex piece of code, so I agree that putting in the effort to learn about it should be the minimum before you attempt to modify it, but putting a lot of money into an LLM to be able to contribute??? First of all, I thought we all agreed that money =/= competence in general, but also pouring a lot of money into an LLM is not a substitute for knowing what the fuck you are doing, and never will be. There, I said it and I'm not sorry. Pouring more money into an LLM won't magically make you understand the Linux kernel. Taking the time to learn how it works will.
Oh, and being able to engage with more expert developers and maintainers also will... but the maintainers aren't willing to teach you, thankyouverymuch. The ones present at the meeting agreed that young people nowadays don't wanna learn, so there's no point in mentoring anyone. In their own words: "it is not the maintainers' job to teach humans how to work on the kernel." [LWN] I'm sorry, I know you're all busy, but somebody has to do that job. Otherwise, what do you think will happen when the expert developers who built the current kernel retire, and all you have left is a bunch of AI-generated code that nobody understands what it does? I mean, look at the picture at the bottom of the report, it's not long until most of them retire. What is your plan from then on? You guessed it, no mention of that. The meowgic chatbot will have solved everything by then, surely.
The report provoked some concerns in the wider community, understandably so. Many were quick to point out the obvious, which is that LLMs are what's allowing low-quality contributions to be made in the first place, and that maybe LLM use should be banned. Maybe just Maybe, the kernel is too delicate and crucial to be left in the hands of a chatbot. To which Torvalds had this to say: "This is NOT some kind of “social warrior” project, never has been, and never will be" [here].
Which... what??? I'm sorry, Linus, I don't want to catsplain your own creation to you... but... Linux has always been a social project. It was born as a reaction to proprietary software, created by people who believed that software should be a community effort, where everyone is free to participate and free to benefit from. Now tell me how that's not borne out of dreaming a better society. I don't know what to think of it, except "boomer who's only progressive until being progressive gets in the way of convenience".
So, what do we do know? I don't know, honestly. If you believe in open source software, Linux is still your best bet, and maybe all this is a fluke. Maybe the reaction from the wider community will be strong enough to course-correct. Maybe.
The AI bubble will crash when the venture capital money runs out, but considering that the AI sector is single-handedly holding up the world economy there will be a lot of effort to keep it running for as long as possible. Which means we might have a few years left of this travesty, and that's enough to do serious damage to the Linux kernel in the meantime. And I'm not just talking about introducing bugs and vulnerabilities: the biggest risk is the loss of knowledge that comes when the old developers are substituted by new people who poured a lot of money into LLM instead of taking the time to learn how the kernel works. I can't believe I'm saying this, but Linux might be destined to die when those developers retire, either because they won't let anyone take over or because the people who take over have no idea what the fuck3 they are doing. And that matters, because what we will loose then is choice. The choice to have technology that works for us and that is not a black box closed off from scrutiny. The choice to not be enslaved to yet another greedy corporation. The choice to participate in the development of the most important technology of our time. That's a fucking lot to give up.
=^.^=