RE: https://chaos.social/@Foxboron/116170859737134271

That’s one way to piss off all your contributors. I think I’ll do this with books. “Claude, rewrite this book in my own style and using different words from the original, so I can’t be accused of copyright infringement.”

I went looking through the repo to see if there had been any community discussion on changing the license. There wasn’t, but the history of issues concerning the license and ownership of tests, etc. leads me to believe the maintainer doesn’t really know much about licensing, and it was causing them all sorts of problems, so they just nuked it all and went with MIT.
@ramsey I'm sure he weighed the tradeoffs. In 4 days he'd done work he put off for 4 years and in the process made the library better and closed out 50+ issues.

@ralphschindler @ramsey And violated how many people's copyright?

And burned how much electricity computing the next letter in a string, which could have powered dozens of homes?

@Crell @ramsey

> And violated how many people's copyright?

with this commit, I suspect none:

https://github.com/chardet/chardet/commit/7e25bf40bb4ae6884892c094080e011290494947

not sure on electricity usage, but presumably he paid for his tokens to do it

I like to think we gave up incandescent bulbs for led as the sacrifice to power this new revolution of computing.

Wipe it clean · chardet/chardet@7e25bf4

Python character encoding detector. Contribute to chardet/chardet development by creating an account on GitHub.

GitHub

@ralphschindler @ramsey That commit means nothing. I am talking more about the copyrighted code the LLM was trained on in the first place without the consent of the authors.

And "he paid for his tokens" is also irrelevant. We gave up incandescent bulbs for LED to avoid making Earth uninhabitable for humans. We need clean electric everywhere, a decade ago. All these data centers are just ensuring my daughter lives in a refugee camp in a few decades.

https://www.garfieldtech.com/blog/selfish-ai

Selfish AI | GarfieldTech

@Crell @ralphschindler @ramsey LLM training on copyright material with no remuneration sucks, and it's not OK. But it already happened, and it isn't going to unhappen. The best we can do is figure out what happens next at this point.
@josh @Crell @ralphschindler My issue is more about what happened to the contributions made to this specific project. The maintainer wiped them out, yes, but if they pointed Claude at the existing code and said “rewrite this so that it doesn’t use the exact code that was contributed,” then I think that’s a royal dick move. Hence my “rewrite this book” analogy.