Copyright and Machine Learning IP

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

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
hgm
Posts: 23608
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: Copyright and Machine Learning IP

Post by hgm » Sat Aug 11, 2018 1:25 pm

noobpwnftw wrote:
Sat Aug 11, 2018 1:20 pm
In that case whether your creativity is "enough" becomes very subjective.
This is why we have courts and judges, rather than automatic application of the law.

In real life there hardly are any absolute boundaries. The criterion "enough" is already in use, in the context of copyright law. E.g. to determine whether a translation of a copyrighted work generated copyrights for the translator.

noobpwnftw
Posts: 338
Joined: Sun Nov 08, 2015 10:10 pm

Re: Copyright and Machine Learning IP

Post by noobpwnftw » Sat Aug 11, 2018 1:32 pm

AdminX wrote:
Sat Aug 11, 2018 1:13 pm
noobpwnftw wrote:
Sat Aug 11, 2018 1:00 pm
hgm wrote:
Sat Aug 11, 2018 12:54 pm
If I make a much stronger NN by feeding it carefully selected (by me) games, I would own the copyright of the resulting NN.
Leela folks would probably argue that the effort they put into making the supervised NN training code(currently broken) is much more than your effort in "careful selection", so they would in turn own your resulting NN.
Is all this really about the NN or is it about the Engine? At the core I would think it should be about the Engine, maybe I am missing something. Is DeusX using a diifferent executable (lc0.exe) as well? As it stands now, I view the NN as I would a opening book with different moves. Only in this case different weights.
Nobody ever neglected the copyright of the NN driver aka. "engine" part, if you are going to that extent then the same logic applies to if you publish a book, powered by Roman alphabet, only in this case DeusX NN runs on LC0 engine based on CUDA backend of TensorFlow powered by cuDNN built for NVIDIA GPUs manufactured by probably TSMC.

chrisw
Posts: 1999
Joined: Tue Apr 03, 2012 2:28 pm

Re: Copyright and Machine Learning IP

Post by chrisw » Sat Aug 11, 2018 10:28 pm

hgm wrote:
Sat Aug 11, 2018 12:54 pm
Isn't it a bit of a moot point whether there is copyright on the produced NN? If the human interaction was indeed minimal, everyone else could repeat the process, and generate the same result. As stated above, uniqueness was not a criterion, they are perfectly allowed to make exactly the same thing by other processes than copying (i.e. independent of the original instance). If person A was allowed to use software written by X to create (and distribute) product P without effort, a person B could do the same, no matter who holds the copyright on P. De-facto P would always be unprotected against identical products flooding the market.

Only when the use of the software would be restricted in some way it could make a difference who owns the copyright. But it seems to me that this could be satisfactorily solved by the licence agreement for use of X. This could stipulate a condition that anything produced with the aid of X should be considered a 'derivative work' of X (comparable to translations of a book to a different language). Whether A should be considered to be co-owner of the copyrights would depend on the amount of effort he had to put in.

So in practice: if I use LC0 to produce a NN by feeding it 'millionbase', the NN would be public domain, as the authors of LC0 did not put any conditions on its use, and I did not put in any effort. If I make a much stronger NN by feeding it carefully selected (by me) games, I would own the copyright of the resulting NN.
In last case, you would never own it, you might share it with Leela Authors, who would have done the bulk of the creative work, via Leela(Trainer).
Which would make an interesting conundrum if you tried to apply restrictions to the "work" in some kind of commercial scenario.

Not to mention the hoops you'ld need to go through to demonstrate sufficient "creativity" in selection.

User avatar
hgm
Posts: 23608
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: Copyright and Machine Learning IP

Post by hgm » Sun Aug 12, 2018 7:35 am

But the Leela authors would never know that they owned the copyrights, if I would not tell them. They won't even be able to prove I used the same algorithm as Leela, without me telling them what game collection I fed to it. It seems you are advocating a law that is completely unenforcible. Law-makers usually do not like that.

Furthermore, the GPL does not automatically transfer to the output of a GPL'ed program. gcc is an obvious precedent for that.

chrisw
Posts: 1999
Joined: Tue Apr 03, 2012 2:28 pm

Re: Copyright and Machine Learning IP

Post by chrisw » Sun Aug 12, 2018 8:56 am

hgm wrote:
Sun Aug 12, 2018 7:35 am
But the Leela authors would never know that they owned the copyrights, if I would not tell them. They won't even be able to prove I used the same algorithm as Leela, without me telling them what game collection I fed to it. It seems you are advocating a law that is completely unenforcible. Law-makers usually do not like that.

Furthermore, the GPL does not automatically transfer to the output of a GPL'ed program. gcc is an obvious precedent for that.
I didn't realise this was a conversation about how to breach law. I had assumed it was about what was the actual real legal situation where all facts were on table. But ok.

I also wasn't advocating laws, I was trying to have a fruitful conversation about what the law really actually was by putting forward some reasoned arguments about what appeared to be the situation. I found it fascinating how the advent of machines that were able to create "works" by themselves affected the copyright situation. But, ok, again.

syzygy
Posts: 4450
Joined: Tue Feb 28, 2012 10:56 pm

Re: Copyright and Machine Learning IP

Post by syzygy » Sun Aug 12, 2018 9:02 am

noobpwnftw wrote:
Sat Aug 11, 2018 1:20 pm
hgm wrote:
Sat Aug 11, 2018 1:08 pm
The only relevant question is whether I would have put in enough creative work to make the result copyrightable. Not how complex the tools were that I used. Even an electric typewriter would take many times the effort to develop than it would take to write the average book on it.
In that case whether your creativity is "enough" becomes very subjective.
I'd say two things are needed:
(1) creativity must have gone into selecting the training games;
(2) some of that creativity must be perceivable from the end result.

For example, one could be very creative in creating the seed of a random generator (e.g. write an original poem and use the md5 of its ascii representation as the seed), but it would not make the (pseudo)random text created by the random generator copyrightable.

It is a bit hard to imagine for me how one could pass (1) and (2).
The Deus X network in all likelihood does not pass (1).

chrisw
Posts: 1999
Joined: Tue Apr 03, 2012 2:28 pm

Re: Copyright and Machine Learning IP

Post by chrisw » Sun Aug 12, 2018 9:20 am

syzygy wrote:
Sun Aug 12, 2018 9:02 am
noobpwnftw wrote:
Sat Aug 11, 2018 1:20 pm
hgm wrote:
Sat Aug 11, 2018 1:08 pm
The only relevant question is whether I would have put in enough creative work to make the result copyrightable. Not how complex the tools were that I used. Even an electric typewriter would take many times the effort to develop than it would take to write the average book on it.
In that case whether your creativity is "enough" becomes very subjective.
I'd say two things are needed:
(1) creativity must have gone into selecting the training games;
(2) some of that creativity must be perceivable from the end result.

For example, one could be very creative in creating the seed of a random generator (e.g. write an original poem and use the md5 of its ascii representation as the seed), but it would not make the (pseudo)random text created by the random generator copyrightable.

It is a bit hard to imagine for me how one could pass (1) and (2).
The Deus X network in all likelihood does not pass (1).
which generates the farcical situation where the "author" of the chess entity "DeusX" has absolutely no IP in the entity, while Leela Chess Authors own all the IP that there is in "DeusX".

and then the farcical situation that TCEC relationship with "DeusX" is via the team that's own no IP and the not the team that owns all the IP.

You couldn't make this up!

syzygy
Posts: 4450
Joined: Tue Feb 28, 2012 10:56 pm

Re: Copyright and Machine Learning IP

Post by syzygy » Sun Aug 12, 2018 9:50 am

chrisw wrote:
Sun Aug 12, 2018 9:20 am
You couldn't make this up!
I see no (copyright) problem as long as it is acknowledged that Lc0 is the engine behind Deus X. Then its use complies with the GPLv3.

TCEC could have felt cheated but apparently they do not. (But they probably knew it from the start anyway? They can't be that naive.)

TCEC spectators might feel upset, but they are not obliged to be TCEC spectators.

User avatar
hgm
Posts: 23608
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: Copyright and Machine Learning IP

Post by hgm » Sun Aug 12, 2018 10:11 am

chrisw wrote:
Sun Aug 12, 2018 9:20 am
You couldn't make this up!
The most common cause for a conclusion that makes no sense, though, is that the reasoning behind it was flawed.

Copyright law was not written with this particular case in mind, so judgement on this case will have to rely on interpretation and precedent. Your assume the training of a NN will be interpreted as a creative act on the part of the NN, which is rather dubious. An important precedent is this: once a copyright holder releases his work into the public domain, he irrevocably waives the possibility to exert that right. And by releasing Leela under the GPL the authors explicitly placed any Leela output, creative or not, in the public domain.

Branko Radovanovic
Posts: 59
Joined: Sat Sep 13, 2014 2:12 pm

Re: Copyright and Machine Learning IP

Post by Branko Radovanovic » Sun Aug 12, 2018 10:21 am

Just some brief remarks - I could write 5 pages on this issue.

Creativity is not an absolute requirement for copyright; see https://en.wikipedia.org/wiki/Sweat_of_the_brow.

NN weights are intellectual property, but they are not source code. The GPL covers source code only. With the rise of NN, this could be an interesting issue, and personally I'd like to hear the position of the Free Software Foundation on that, but I don't think it's likely they'd ever view NN weights as "code", even if there is a degree of functional equivalence.

The idea that because monkey pressed the shutter button, the photographer does not own the copyright, is perversely stupid and wrong.

The final and the most important thing: copyright law has always been largely utilitarian both in theory and practice - of course, because it serves a purpose, like all laws. And let's be honest, it's not to protect "creativity" or "innovation" per se, but rather to protect the financial interests of creators. You have to create something that hasn't existed before, and you have to put some effort, resources and skill into it: this is your investment, and the law is meant to protect it. Ultimately, this is vital for the benefit of society in general, and the idea of benefit guides the way law is generally interpreted when things are not fully clear.

Post Reply