Sam StephensonMar 5
Made the mistake of reading the Mark Pilgrim GitHub thread first thing this morning and now I’m in a terrible mood
Show threadSam StephensonMar 5

This issue has it all!
- Original copyright holder makes a straightforward claim of violation
- Sloppers try to defend it using the Oracle v Google ruling
- Armin “Plausibly Deniable” Ronacher shows up to do nothing but sow doubt
- Maintainer in question replies with a response obviously written by Claude, saying he instructed Claude not to reference any copyrighted code

https://github.com/chardet/chardet/issues/327

No right to relicense this project · Issue #327 · chardet/chardet

Hi, I'm Mark Pilgrim. You may remember me from such classics as "Dive Into Python" and "Universal Character Encoding Detector." I am the original author of chardet. First off, I would like to thank...

GitHub
Show threadSam StephensonMar 5
Just… fuck you, dude
farhaven 🇪🇺Mar 5
Show threadSam Stephenson

“I then started in an empty repository with no access to the old source tree, and explicitly instructed Claude not to base anything on LGPL/GPL-licensed code.”

idk, it’s just baffling to me that a professional software developer who’s been maintaining a project for over a decade can so fundamentally misunderstand what LLMs do

Mar 5 at 4:51pmWeb
Show threadCap YbarraMar 5
@sstephenson it's somehow only ever misunderstanding that's advantageous to their rhetorical position 🤔
Show threadFoolishOwlMar 5
@sstephenson It's Enclosure, the foundational maneuver of capitalism. Seize the commons and claim that since they were commons, no one owned them, but now they're yours.
Show threadKasper Timm HansenMar 5
@sstephenson it seems like people have no idea what a training set and how models are built.
Show threadbenMar 5

@sstephenson i told it not to do something that means it definitely won’t!

and if despite my clear instructions it does the thing, i’ll make sure it apologises properly

Show threadJames HoldenMar 5
@sstephenson Models have long been shown to be incapable of not doing something you tell them not to. Try asking an image generator model to draw a picture of a room without a duck in it. They can’t.
Show threadnate berkopecMar 5
@sstephenson using claude to write your comments and the slop readme as well makes me think the author here just likes turning their brain off. letting the shoggoth take the wheel.
Show threadIvan Sagalaev Mar 6
@sstephenson cue that quote about not understanding something when your salary depends on you not understanding it.
Show threadJoeri SebrechtsMar 6
@sstephenson @NfNitLoop
What a Solomon’s choice. Either the original creator is right that this is a derived work and LGPL must be applied. Or, the maintainer is right that this is a pure LLM recreation of the v6 design, in which case no copyright can be applied, as courts have already ruled that AI-produced works that lack human creativity cannot be copyrighted.
Show threadAlex RussellMar 6

@sstephenson ...or how copyright works, or how communities work, or how OSS functions, or how LLMs are trained.

I'm pretty used to software people getting waaaaay over their skis in other domains, but this...*phew*.

Show threaddeividMar 10
@sstephenson Oh wow. Just a few days ago I found https://malus.sh/ and thought that it may be a parody but it won't be too long before people start actually doing it. Well, there you go 😡
MALUS - Clean Room as a Service | Liberation from Open Source Attribution