Login
You're viewing the functional.cafe public feed.

Replies

  • Nov 18, 2026, 12:53 PM

    @jackdaniel i've noticed in other places the delusion that LLM-generated code is unencumbered simply because AI companies don't claim the copyright (e.g., the Tmux project scrupulously lists the copyright policies of major AI providers). i suppose one delusion tends to beget another.

    💬 1🔄 1⭐ 1
  • Nov 18, 2026, 12:54 PM

    @spnw there is a lot of wishful thinking in llm circles and the most damning one, yet sadly plausible, is "there is too many of us and money involved to decide that it is illegal".

    💬 3🔄 3⭐ 12
  • Nov 18, 2026, 1:03 PM

    @jackdaniel @spnw I'm not sure it's wishful thinking any more. SQLite *is* too big to fail, bigger than Anthropic or the Linux kernel.

    💬 1🔄 1⭐ 0
  • Nov 18, 2026, 1:11 PM

    @deech @spnw

    I'm not denying that it is a likely outcome, but I don't personally believe in a notion of "too big to fail", only in " big enough to split apart".

    I will hypothesize a little, but it is not that all software vanishes if we take an obvious (to me) take, that llms training is not a fair use given that results compete with original authors. it is not that we will lose every bit of software tainted by copyright violations forbidden fruit -- sqlite may be reset to 2026-05 state, Linux to 2020-something etc. and that would be a compromise.

    the alternative is of course very likely and demoralizing. abiding by rules is for the weak, and law is a shield of the wealthy. it is a cynical line of thought that I don't entertain myself, among other things because it leads to revolutions where everybody loses. so I'd clearly prefer big to fail, or if impossible - split and slowly dissolve.

    💬 1🔄 2⭐ 8
  • Nov 18, 2026, 1:14 PM

    @jackdaniel @deech there's also the possibility that it becomes treated like asbestos (to use a popular analogy). as in we're stuck with the old commits indefinitely but future LLM code generation becomes legally dubious. i doubt even this is a strong possibility, sadly.

    💬 0🔄 0⭐ 1
  • 💬 0🔄 1⭐ 2
  • Nov 18, 2026, 6:00 PM

    @jackdaniel @spnw I was arguing that being a problem on the horizon way back when the SFC was offering clown guidance about LLM legal standings.

    Well lookie what actually happened. Copyleft took it in both kneecaps from a sawed-off twelve gauge. Who, besides me and a multitude of other people with basic common sense, could have foreseen such outcomes from integrating slop into all our software infrastructure, making the plagiarists too big to jail? Inconceivable.

    💬 1🔄 0⭐ 2
  • 💬 0🔄 0⭐ 0
  • Dipoledipolecat@yiff.life
    Nov 18, 2026, 8:36 PM

    @jackdaniel The quoted text doesn't appear in the linked page. I can't find it on older versions of the file either. What is the quote from?

    💬 1🔄 0⭐ 0
  • Nov 18, 2026, 8:39 PM

    @dipolecat

    It was the paraphrase from memory on why sqlite does not accept external contributions. I've looked up the exact quote, it goes like:

    " But SQLite is not open-contribution. In order to keep SQLite in the public domain and ensure that the code does not become contaminated with proprietary or licensed content, the project does not accept patches from people who have not submitted an affidavit dedicating their contribution into the public domain."

    sqlite.org/copyright.html

    💬 1🔄 0⭐ 3
  • 💬 0🔄 0⭐ 0