Are neural nets (the weights file) copyrightable?

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

syzygy
Posts: 5741
Joined: Tue Feb 28, 2012 11:56 pm

Re: Are neural nets (the weights file) copyrightable?

Post by syzygy »

Fulvio wrote: Sun Feb 28, 2021 10:51 am
syzygy wrote: Sun Feb 28, 2021 10:43 am Do you understand "if', i.e. the concept of a hypothetical?
No.
Maybe there is a misunderstanding with the meaning of releasing, which I interpreted as "Remove restrictions or obligations from (someone or something) so that they become available for other activity."

Can you please reword your statement following this guidelines?
https://wordvice.com/grammar-avoid-double-negatives/
I already removed the single negative, but you ignored it.
syzygy
Posts: 5741
Joined: Tue Feb 28, 2012 11:56 pm

Re: Are neural nets (the weights file) copyrightable?

Post by syzygy »

Fulvio wrote: Sun Feb 28, 2021 10:54 am
Dann Corbit wrote: Sun Feb 28, 2021 10:22 am You don't need any permission if you supply the source. It says that very clearly in the license.
People are trying to manufacture a requirement for data that does not exist, so that they can find fault.
Are you claming that the images of a GPL videogame are not part of the source?
If they are not part of the distributed executable, then they are not part of the source.

If they are part of the distributed executable, you need to provide, together with the source code, the material necessary to recreate the executable. (Or at least something close to it -- you don't need to provide a full copy of the compiler.)
Dann Corbit
Posts: 12792
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: Are neural nets (the weights file) copyrightable?

Post by Dann Corbit »

Yes, of course I am claiming that, assuming that the images are loaded into memory by the program.
Data is not source code.
I make no claims as to copyright of the images.
That is a separate matter.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
Michel
Posts: 2292
Joined: Mon Sep 29, 2008 1:50 am

Re: Are neural nets (the weights file) copyrightable?

Post by Michel »

Even if you create a copy without changing one bit you are doing something illegal, unless with permission from the copyright holder. Or unless some exception within copyright law applies. But clearly copyright law restricts the right to copy.
This seems very weird to me (because of the book example). But anyway it seems the modification clause in the GPL does make sense from this point of view.

So it still all boils down to whether dynamic linking creates a derivative work or not. Duh. The FSF says yes (supported by Law professor Eben Moglen) and there are no court precedents one way or the other.

The fact that there are no court cases may indicate that companies have generally chosen to follow the interpretation by the FSF though.
Ideas=science. Simplification=engineering.
Without ideas there is nothing to simplify.
Fulvio
Posts: 396
Joined: Fri Aug 12, 2016 8:43 pm

Re: Are neural nets (the weights file) copyrightable?

Post by Fulvio »

Dann Corbit wrote: Sun Feb 28, 2021 11:12 am Yes, of course I am claiming that, assuming that the images are loaded into memory by the program.
Data is not source code.
We need to start to define some word to avoid misunderstandings.
Everything is data, just 0 or 1, and everything is loaded into memory.
Data is usually used with the meaning of "user's data" like the data you store into a database.

What you are claming is that you can download the source code of a GPL game:
https://wiki.wesnoth.org/Download
replace a few .png images and then you can legally distribute it with a different license?
User avatar
towforce
Posts: 12518
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK
Full name: Graham Laight

Re: Are neural nets (the weights file) copyrightable?

Post by towforce »

Fulvio wrote: Sun Feb 28, 2021 8:16 amI have a reproduction of the 1865 edition of "Alice's Adventures in Wonderland" which was published with illustrations (by John Tenniel).
Could someone have distributed the book with a different story and the same illustrations without permission? No.
Could someone have distributed the book with different illustrations without permission? No.
Can someone distribute stockfish with a different NNUE without permission? No.

+1

I like clear explanations, and that was a very clear explanation!
Human chess is partly about tactics and strategy, but mostly about memory
syzygy
Posts: 5741
Joined: Tue Feb 28, 2012 11:56 pm

Re: Are neural nets (the weights file) copyrightable?

Post by syzygy »

Fulvio wrote: Sun Feb 28, 2021 10:51 am Maybe there is a misunderstanding with the meaning of releasing, which I interpreted as "Remove restrictions or obligations from (someone or something) so that they become available for other activity."
I missed this. No, releasing here just means releasing it to the public, i.e. publishing it, putting it on your website for download, etc.
Windows 95 was released in August 1995. Video games, movies, smart phones, etc. also have a "release date".
syzygy
Posts: 5741
Joined: Tue Feb 28, 2012 11:56 pm

Re: Are neural nets (the weights file) copyrightable?

Post by syzygy »

Michel wrote: Sun Feb 28, 2021 11:32 am
Even if you create a copy without changing one bit you are doing something illegal, unless with permission from the copyright holder. Or unless some exception within copyright law applies. But clearly copyright law restricts the right to copy.
This seems very weird to me (because of the book example). But anyway it seems the modification clause in the GPL does make sense from this point of view.
But surely you are aware that you are not free to make copies of a book that you bought? (There may be an exception in certain jurisdictions for "home copies" but that does not diminish the fact that copying is an act restricted by copyright.)
So it still all boils down to whether dynamic linking creates a derivative work or not. Duh. The FSF says yes (supported by Law professor Eben Moglen) and there are no court precedents one way or the other.
Including a reference to another book does not copy material from that other book into your book. So there is no copying when you include a reference. The FSF is alone in its position, and Eben Moglen isn't exactly impartial.
User avatar
hgm
Posts: 28391
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: Are neural nets (the weights file) copyrightable?

Post by hgm »

Fulvio wrote: Sun Feb 28, 2021 11:43 amData is usually used with the meaning of "user's data" like the data you store into a database.
Computer software is organized hierarchically in 'levels' and what is code at one level can be data as far as the undelying level is concerned. So the distinction is not that clear cut.

E.g. engines like Fairy-Max or Sjaak II come together with a file of game definitions to specify the rules of individual chess variants. The game-description format, when sufficiently powerful, in fact becomes a programming language interpreted by these engines. Although to the engines they are just data files.

The GPL FAQ explicitly addresses this situation, and states that it is legal to distribute a GPL'ed interpreter with an interpreted program that has a different license. As far as I am concerned that settles the matter: linking programs at the same level of organization (e.g. through DLLs) is not allowed, bundling interpreted code with an interpreter is.
Michel
Posts: 2292
Joined: Mon Sep 29, 2008 1:50 am

Re: Are neural nets (the weights file) copyrightable?

Post by Michel »

syzygy wrote: Sun Feb 28, 2021 12:56 pm
Michel wrote: Sun Feb 28, 2021 11:32 am
Even if you create a copy without changing one bit you are doing something illegal, unless with permission from the copyright holder. Or unless some exception within copyright law applies. But clearly copyright law restricts the right to copy.
This seems very weird to me (because of the book example). But anyway it seems the modification clause in the GPL does make sense from this point of view.
But surely you are aware that you are not free to make copies of a book that you bought? (There may be an exception in certain jurisdictions for "home copies" but that does not diminish the fact that copying is an act restricted by copyright.)
So it still all boils down to whether dynamic linking creates a derivative work or not. Duh. The FSF says yes (supported by Law professor Eben Moglen) and there are no court precedents one way or the other.
Including a reference to another book does not copy material from that other book into your book. So there is no copying when you include a reference. The FSF is alone in its position, and Eben Moglen isn't exactly impartial.
But why are there no court cases then? If dynamic linking does not create a derived work then it becomes trivial to hijack large GPL projects. Surely companies will do this if it is possible.
Ideas=science. Simplification=engineering.
Without ideas there is nothing to simplify.