Are neural nets (the weights file) copyrightable?

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

Moderator: Ras

syzygy
Posts: 5801
Joined: Tue Feb 28, 2012 11:56 pm

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

Post by syzygy »

Alayan wrote: Sat Feb 27, 2021 2:11 am The purpose of a video player is to play video files. A specific video file is in no way essential to it. If you don't have it bundled with the video player, it doesn't affect the functionality of the video player in any way.
But you are misguided in thinking that program functionality plays any role here. Copyright is about expression, not about functionality. The copyright on SF is the copyright on the free creativity (= not determined by functionality) used in producing its textual source code. It doesn't matter whether the program produces anything sensible.

Ignore functionality and just look at dumb source files.
gonzochess75
Posts: 208
Joined: Mon Dec 10, 2018 3:29 pm
Full name: Adam Treat

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

Post by gonzochess75 »

syzygy wrote: Sat Feb 27, 2021 2:28 am
Alayan wrote: Sat Feb 27, 2021 2:11 am The purpose of a video player is to play video files. A specific video file is in no way essential to it. If you don't have it bundled with the video player, it doesn't affect the functionality of the video player in any way.
But you are misguided in thinking that program functionality plays any role here. Copyright is about expression, not about functionality. The copyright on SF is the copyright on the free creativity (= not determined by functionality) used in producing its textual source code. It doesn't matter whether the program produces anything sensible.
That's true, but the copyright LICENSE of SF - aka, the terms under which the copyright holder grants others permissions they otherwise would not have - expressly prohibits combining it with a non-GPL'd component to form a larger program.

Could Disney grant ViacomCBS permission to redistribute Star Wars if and only if ViacomCBS granted Disney permission to redistribute Star Trek?

Clearly the copyright of Star Wars has absolutely nothing to do with the copyright of Star Trek. They are two distinct works. But could Disney draft up that copyright LICENSE and could ViacomCBS legally accept?

The point Alayan is making about functionality has to do with the interpretation of the words of the GPL, "not combined with it such as to form a larger program." His point is that it is incredulous to believe that GPL'd engine + NNUE net is "not combined with it such as to form a larger program." OTOH, it is easy to see that a video player playing a random video file does "not combined with it such as to form a larger program."
Last edited by gonzochess75 on Sat Feb 27, 2021 2:41 am, edited 2 times in total.
Alayan
Posts: 550
Joined: Tue Nov 19, 2019 8:48 pm
Full name: Alayan Feh

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

Post by Alayan »

syzygy wrote: Sat Feb 27, 2021 2:28 am
Alayan wrote: Sat Feb 27, 2021 2:11 am The purpose of a video player is to play video files. A specific video file is in no way essential to it. If you don't have it bundled with the video player, it doesn't affect the functionality of the video player in any way.
But you are misguided in thinking that program functionality plays any role here. Copyright is about expression, not about functionality. The copyright on SF is the copyright on the free creativity (= not determined by functionality) used in producing its textual source code. It doesn't matter whether the program produces anything sensible.
We agree that the NNUE net is not covered by SF's copyright. The distribution of the standalone NNUE net cannot be forbidden by SF rights holders because they don't hold right to the net.

But functionality is relevant to the Stockfish licensing conditions.

Distributing the NNUE net without a GPL-compatible license alongside Stockfish violates Stockfish licensing conditions. Therefore, the bundle distribution (even if in two separate files instead of merged in one binary) is forbidden.
gaard
Posts: 463
Joined: Mon Jun 07, 2010 3:13 am
Location: Holland, MI
Full name: Martin W

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

Post by gaard »

Alayan wrote: Sat Feb 27, 2021 2:38 am (snip)
Distributing the NNUE net without a GPL-compatible license alongside Stockfish violates Stockfish licensing conditions. Therefore, the bundle distribution (even if in two separate files instead of merged in one binary) is forbidden.
Where is this written in the GPLv3 license? Forgive me if I am asking you to repeat yourself.
Last edited by gaard on Sat Feb 27, 2021 2:54 am, edited 1 time in total.
Modern Times
Posts: 3770
Joined: Thu Jun 07, 2012 11:02 pm

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

Post by Modern Times »

Alayan wrote: Sat Feb 27, 2021 2:38 am
Distributing the NNUE net without a GPL-compatible license alongside Stockfish violates Stockfish licensing conditions. Therefore, the bundle distribution (even if in two separate files instead of merged in one binary) is forbidden.
So many points of view, it needs to go before a judge in a court of law to properly answer the question, in my opinion. You may have a proper legal opinion to back up your assertion, but people go to court on opposite legal opinions and the judge decides which one is right.
George Sobala
Posts: 44
Joined: Sat Feb 03, 2018 2:42 pm
Location: Yorkshire, England

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

Post by George Sobala »

After pages of off-topic discussion about GPLv3 by two individuals who seem to agree about the answer to the actual topic, let me toss in some further off-topic GPLv3 observations.

Daniel Uranga and CB are separate legal entities. DU has distributed a modified SF by publishing the modified source code on Github and possibly by providing a binary to AS/CB to whom he has also given the source code. He therefore has no on-going obligations to continue distributing on Github and could take this repository down at any time.

CB has distributed binaries by selling them. To date, they have not distributed the source code to anyone, although it is their and not DU’s obligation. Arguably they may or may not have fulfilled their obligation by pointing to DU’s repository, and arguably they have 30 days from binary distribution to fully comply. However unarguably should DU take down his repository after the 30 days have elapsed, then CB will be in potentially permanent breach of the GPL.
gonzochess75
Posts: 208
Joined: Mon Dec 10, 2018 3:29 pm
Full name: Adam Treat

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

Post by gonzochess75 »

gaard wrote: Sat Feb 27, 2021 2:49 am
Alayan wrote: Sat Feb 27, 2021 2:38 am (snip)
Distributing the NNUE net without a GPL-compatible license alongside Stockfish violates Stockfish licensing conditions. Therefore, the bundle distribution (even if in two separate files instead of merged in one binary) is forbidden.
Where is this written in the GPLv3 license? Forgive me if I am asking you to repeat yourself.
Here you go:

"You must license the entire work, as a whole, under this License to anyone who comes into possession of a copy. This License will therefore apply, along with any applicable section 7 additional terms, to the whole of the work, and all its parts, regardless of how they are packaged. This License gives no permission to license the work in any other way, but it does not invalidate such permission if you have separately received it."

Emphasis mine.
https://www.gnu.org/licenses/gpl-3.0.en.html
syzygy
Posts: 5801
Joined: Tue Feb 28, 2012 11:56 pm

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

Post by syzygy »

gonzochess75 wrote: Sat Feb 27, 2021 2:34 am That's true, but the copyright LICENSE of SF - aka, the terms under which the copyright holder grants others permissions they otherwise would not have - expressly prohibits combining it with a non-GPL'd component to form a larger program.
The GPLv3 only requires that the entire work, i.e. FF2-SF and its sources, be distributed under the GPLv3.

And again, there is no relevant difference with the image viewer example. That the functionality is different is irrelevant. FF2-SF is an entire copyrightable work just as much as the modified image viewer.
Could Disney grant ViacomCBS permission to redistribute Star Wars if and only if ViacomCBS granted Disney permission to redistribute Star Trek?
I know how licensing works, thank you.
Michel
Posts: 2292
Joined: Mon Sep 29, 2008 1:50 am

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

Post by Michel »

syzygy wrote: Fri Feb 26, 2021 11:47 pm
syzygy wrote: Fri Feb 26, 2021 11:04 pmThe FSF's FAQ contains many untruths.
The following paper makes for interesting reading and comments on the absurdity of the FSF's position:
https://courses.cs.washington.edu/cours ... rksGPL.pdf

This article is about the GPL 2 and the wording of the GPL 3 is significantly different. In particular the GPL 3 no longer uses the terminology of derivative work.

I don’t see how a license cannot prohibit distribution in a larger package if certain conditions are not met. This is clearly what the GPL 3 attempts to achieve and its application to concrete cases (e.g. dynamic linking which is similar to the use of a NNUE net) is detailed in the FAQ. I agree that it is not always obvious how to extract the claims in the FAQ from the license itself but I rely on the fact that the GPL 3 was written by a Columbia law professor and has undergone substantial review before it was accepted.
Last edited by Michel on Sat Feb 27, 2021 7:15 am, edited 2 times in total.
Ideas=science. Simplification=engineering.
Without ideas there is nothing to simplify.
Modern Times
Posts: 3770
Joined: Thu Jun 07, 2012 11:02 pm

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

Post by Modern Times »

To me it is crystal clear that the NNUE distributed alongside the stockfish binary is in no way covered by the GPL. Sure I'm not a lawyer, and even the lawyers may disagree on it. No matter how many more pages this thread drags on the position won't be any clearer.