“Useful” is not sufficient

The text "Quick note" in my handwriting

“Useful” is not sufficient

Jul 15, 2026

—

So Linus Torvalds, head of the Linux kernel development, put his foot down on the Linux Kernel development mailing list when someone was bringing up criticism of LLMs:

“Linux is not one of those anti-AI projects, and if somebody has issues

with that, they can do the open-source thing and fork it.

Or just walk away.

AI is a tool, just like other tools we use. And it’s clearly a useful one.

It may not have been that “clearly” even just a year ago, but it’s no

longer in question today.

There are other questions around AI (like what the economy of it will

actually look like in the end), but “is it useful” is no longer one of

those questions. Anybody who doubts that clearly hasn’t actually used

it.”

Linus Tovalds

This is an argument we do often hear: That LLMs’ usefulness cannot be debated and therefore they are just “tools” that need to be accepted and that just need some proper governance or social norms and individual vigilance.

This is presented as sort of the killer argument for the use of LLMs: They have proven to be so super useful. As if that was a good argument.

Usefulness is not an objective, fully abstract property: A thing is not useful for everyone or in general but mostly just for specific people in specific contexts. And useful never comes for free.

But let’s take a step back and use a different example: Child labor. Child labor is very “useful” if you want cheap labor. It’s 100% deplorable to use it or accept it. Child labor is a monstrosity. But if “useful” is enough, do we need to accept child labor? Palantir surveillance software is useful to some. Building closed source proprietary software is useful (for the people selling it). Usefulness is a very weak argument in support of something.

Sure people consider LLMs useful, otherwise they literally wouldn’t use it. But that’s not a get out of jail free card.

Especially when it comes to free software, it’s weird when the main characters totally ignore the political and social context of technological objects: Many people use Linux not because it’s the best but for political reasons, for reasons of autonomy or to be part of a community.

But now we know where Linus stands. If something is somehow useful (to you? to Linus? I don’t know who it has to be useful for) it’s okay to use. I wonder what the stance is on using decompilers to generate code from proprietary drivers or operating systems and just paste them into the Linux Kernel? It is basically the same. Code of dubious legal standing being thrown into a codebase built around a free license. But just decompiling and copy pasting is so fucking useful.

In a burning world we need to hold ourselves and each other to a higher standard. “Useful” is far, far from being good enough.

Liked it? Take a second to support tante on Patreon!

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

---

Comments

11 responses to ““Useful” is not sufficient”

1. Emmanuel F. July 15, 2026 Thank you. Technology is never neutral. See Jacques Ellul’s work ( <https://en.wikipedia.org/wiki/The%5FTechnological%5FSociety> ) Reply 2. Ian M. Noone Avatar Ian M. Noone July 15, 2026 @blog @chronohart My b-i-l is a dev for Red Hat and is seeing younger devs come in who only know "vibe-coding", and it's gonna be a real shit show when all the older ones who actually know how to troubleshoot code age out or leave Reply as Reply on the Fediverse #### Original Comment URL Paste the comment URL into the search field of your favorite open social app or platform. Comment URL #### Your Profile Or, if you know your own profile, we can start things that way! Why do I need to enter my profile? This site is part of the ⁂ open social web, a network of interconnected social platforms (like Mastodon, Pixelfed, Friendica, and others). Unlike centralized social media, your account lives on a platform of your choice, and you can interact with people across different platforms. By entering your profile, we can send you to your account where you can complete this action. Your Fediverse profile Reply Loading… Save my profile for future comments.

1. Liam Proven Avatar Liam Proven July 16, 2026 @ianmnoone @blog @chronohart I have bad news for you. That happened a generation ago. In the timeframe from the end of the 1990s to the early 2000s was when the copycat newbies took over and most of the skilled artisans got promoted out of responsibility, or quit, or stopped caring. As it happens, I'm ex RH myself but this isn't a RH local problem. It's an "entire PC industry" problem and since things went 64-bit it's spreading to firmware, embedded, etc as well. Reply as Reply on the Fediverse #### Original Comment URL Paste the comment URL into the search field of your favorite open social app or platform. Comment URL #### Your Profile Or, if you know your own profile, we can start things that way! Why do I need to enter my profile? This site is part of the ⁂ open social web, a network of interconnected social platforms (like Mastodon, Pixelfed, Friendica, and others). Unlike centralized social media, your account lives on a platform of your choice, and you can interact with people across different platforms. By entering your profile, we can send you to your account where you can complete this action. Your Fediverse profile Reply Loading… Save my profile for future comments. 3. tkissing Avatar tkissing July 15, 2026 @blog Linus T. has never been a great public example of being a good person, but this is low even for him Reply as Reply on the Fediverse #### Original Comment URL Paste the comment URL into the search field of your favorite open social app or platform. Comment URL #### Your Profile Or, if you know your own profile, we can start things that way! Why do I need to enter my profile? This site is part of the ⁂ open social web, a network of interconnected social platforms (like Mastodon, Pixelfed, Friendica, and others). Unlike centralized social media, your account lives on a platform of your choice, and you can interact with people across different platforms. By entering your profile, we can send you to your account where you can complete this action. Your Fediverse profile Reply Loading… Save my profile for future comments. 4. eobet July 15, 2026 Asbestos was also very useful, and we’re still cleaning up that mess. Plastics are still very useful and we haven’t even begun cleaning that up (we all have micro plastics in our blood). They’re going to spend a long time cleaning up the mess left behind by AI (not just the code, but also the brain rot), if we haven’t burned the biosphere by then. Reply 5. @990000@mstdn.social Avatar @990000@mstdn.social July 15, 2026 @blog I’ll wager AI coding can be very useful if you’re diligent and responsible especially when using it to cut down on tedium — not necessarily thinking for you, per se — but it will never be ok environmentally at the rate that it’s being abused 🤷🏻‍♂️ Reply as Reply on the Fediverse #### Original Comment URL Paste the comment URL into the search field of your favorite open social app or platform. Comment URL #### Your Profile Or, if you know your own profile, we can start things that way! Why do I need to enter my profile? This site is part of the ⁂ open social web, a network of interconnected social platforms (like Mastodon, Pixelfed, Friendica, and others). Unlike centralized social media, your account lives on a platform of your choice, and you can interact with people across different platforms. By entering your profile, we can send you to your account where you can complete this action. Your Fediverse profile Reply Loading… Save my profile for future comments. 6. Chris July 15, 2026 I think Linus done it right because we will have community free will share and driven LLMs in future. They are ne free will evolution spending things we have to presend so linux and everything have to handle this kind of animals and friends or foes. Reply 7. Mary July 15, 2026 Yes! Every move away from punch cards was a move in the wrong direction! Reply 8. Devin Avatar Devin July 16, 2026 @blog I'm new to programming. Today, Jetbrains' Junie agent, running a Gemini Flash Preview model, helped me put a sun and moon .svg on the dark mode toggle button I made for my Angular app. I'm using Tailwind CSS, so implementing the graphics' opacity changes with 'peer-checked' on the checkbox input element was a bit obscure to find guidance on. I really appreciated agents as a tool today, and I learned something. Anywho, I think using LLMs for important systems programming is a bad idea. lol Reply as Reply on the Fediverse #### Original Comment URL Paste the comment URL into the search field of your favorite open social app or platform. Comment URL #### Your Profile Or, if you know your own profile, we can start things that way! Why do I need to enter my profile? This site is part of the ⁂ open social web, a network of interconnected social platforms (like Mastodon, Pixelfed, Friendica, and others). Unlike centralized social media, your account lives on a platform of your choice, and you can interact with people across different platforms. By entering your profile, we can send you to your account where you can complete this action. Your Fediverse profile Reply Loading… Save my profile for future comments. 9. Abhiseck July 16, 2026 Mr. Torvalds was never one of those people to take ethical or moral stance over topic related to software, unlike GNU Project. He switched to using the term “Open Source” instead of Free software, so that he doesn’t have to think about ethical issues regarding software freedom. For a long time he supported the use proprietary version control system for developing Linux, only when he got f\\\*ed by the company that he decided to build Git. Since 1996, he violated the license of his own software by adding non free peripheral software to the kernel\[1\]. So I’m not at all surprised that he’d take such a stance regarding LLMs. <https://www.fsfla.org/ikiwiki/selibre/linux-libre/> Reply 10. Florian July 16, 2026 Aside from the ethical issues surrounding LLMs themselves, there are many other aspects of Linus’s position that I don’t understand. For example, how are contributors who can’t afford LLM services supposed to have a chance to keep up? Of course, ensuring that isn’t the project’s primary goal, but it’s still a form of indirect gatekeeping. Furthermore, in my professional experience (where we’re pressured to use LLMs), LLMs are particularly prone to errors when there isn’t enough context. With a project as complex and important as the Linux kernel, how can we be sure that the LLM won’t generate code based on incorrect assumptions? Code that might have subtle side effects on other parts of the kernel? LLMs have shown time and again that the code quality looks good at first glance, but the errors are often very subtle because incorrect assumptions were made about the system as a whole. In my eyes Linus is wrong, and I’m a bit worried about the future. Maybe the BSDs are a healthy alternative. Reply