Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

'gary_0' being happy with the license terms isn't what defines 'open source'.

I'm fairly happy with the license terms too. They're just not open source. We dilute the term open source for the worst if we allow it to apply to build artifacts for some reason.



We were talking about "looking a gift horse in the mouth", as in it's still a positive thing regardless of the semantic quibbles about open source. Nobody would argue that a hypothetical openly licensed Windows binary-only release is "open source" and I'd appreciate it if you read my comments more charitably in future.

Source code licenses are naturally quite clear about what constitutes "source code", but things are murkier when it comes to ML models, training data, and associated software infrastructure, which brings up some interesting questions.


> We were talking about "looking a gift horse in the mouth", as in it's still a positive thing regardless of the semantic quibbles about open source

Your gift horse in the mouth comment was visibly an aside in the greater discussion being enclosed in parenthesis.

> Nobody would argue that a hypothetical openly licensed Windows binary-only release is "open source" and I'd appreciate it if you read my comments more charitably in future.

That's why I'm using it as an example metaphor in my favor. It's clearly not open source even if they released it under Apache 2. It's not what their engineers edit before building it.

> Source code licenses are naturally quite clear about what constitutes "source code", but things are murkier when it comes to ML models, training data, and associated software infrastructure, which brings up some interesting questions.

I don't think they're all that murky here. The generally accepted definition being

> The “source code” for a work means the preferred form of the work for making modifications to it. “Object code” means any non-source form of a work.

Is this the form of the work that Mistral's engineers work in? Or is there another form of the work that they do their job in and used to build these set of files that they're releasing?


You're asking them to release all their training data? very unlikely that's going to happen.


There's a lot of reasons why an org wouldn't want to open source their release. That doesn't make it open source.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: