[nickff]

It seems to be a common view on HN that licenses and conditional access to websites should be ignored (i.e. WRT ad-blockers), but also that licenses on Open-Source Software repositories should be respected (i.e. WRT LLM training). I believe that holding these contradictory views is common, but the conflict would need to be resolved to come to a conclusion on how to proceed with LLM training.

[pseudalopex]

There is no contradiction. Open source software licenses allow use without conditions. Ad blocker use does not distribute the modified web pages.

[nickff]

I have not seen any evidence that LLMs ‘distribute’ modified software, though they do seem capable of replicating it.

[lavela]

I fail to see how mass scale reproduction of copyrighted code isn't a form of distribution.

[nickff]

Replication is not the same as reproduction; I can replicate an API without violating someone's license or copyright (which I would by reproducing their work).

[pseudalopex]

Reproduce is a definition of replicate. And LLMs reproduced code.

[pseudalopex]

The view LLMs should respect open source software licenses is not for replication alone. Models and generated code are derived from training data.

[nickff]

Developers are permitted to learn from open source code with restrictive copyrights, and apply those lessons to developing other software which does not comply with the copyright of their 'example'.

As an aside, I do believe that LLM trainers are ignoring and violating many licenses, but open-source software is not a clear example of a violation.

[necovek]

Depends on how you define "learn": usually, a company wanting to rebuild and publish something under a different license prohibits their developers from having ever looked at original code, to avoid the risk of copying over exact snippets out of their memory accidentally.

Copyright protects only arbitrarily non-trivial parts of the original being reproduced, but that means that you have to be careful with learning from copyrighted material. Programming books will have direct clauses allowing snippet reuse, but not for teaching purposes.

[pseudalopex]

> Sure, but developers are permitted to learn from open source code with restrictive copyrights, and apply those lessons to developing other software which does not comply with the copyright of their 'example'.

This was a different argument. And there is no contradiction to separate LLMs and people.

> As an aside, I do believe that LLM trainers are ignoring and violating many licenses, but open-source software is not a clear example of a violation.

How?

[sumeno]

LLMs are not people. They do not learn the way people do.

Even if they did, if someone memorized copywritten code and then typed it back out that would still be a copywrite violation

[no-name-here]

> Open source software licenses allow use without conditions.

Don't a number of open source licenses notably involve restrictions?

[technothrasher]

You seem to be conflating copyright with access rights. Two very different things. Regardless of your feelings on either, there is no contradiction in holding different views on them.

[nickff]

Copyright is all about gating access, as media rights holders for sports well know.

[SllX]

Well no, it’s about legally gating the ability to copy so the original author doesn’t have to compete in the same market to sell his own book with every other bloke with a printing press and a copy of the book. Everything else is an addendum.

[joquarky]

No, it's to promote the progress of science and the useful arts.

The current implementation has recently become obsolete.

[SllX]

Don’t confuse the social justification with the actual purpose of copyright law just because it’s written into the US Constitution that way. America didn’t invent copyright law.

[nickff]

That may be the reason copyright came to be, but it's much more expansive now.

[SllX]

That is still the meat and potatoes of copyright law.