A lawyer on the GPL and Fruit/Rybka

[BB+]

http://talkchess.com/forum/viewtopic.ph ... 52&t=37817
Nice to see someone with legal qualifications actually speak about this.

For the question:

Does Rybka 1 Beta include a verbatim “portion” or a “modified” “portion” of Fruit 2.1?

My impression is "yes" -- for instance, the evaluation function is a "portion" (indeed, quite a large one), and [for the purposes here] the evaluation function of Rybka can be considered a "modification" of that of Fruit (I realise the extent of this claim would need to be carefully delineated, but trying to do so here would reduce the discussion to one of a more technical nature). Others might say that the "modification" as meant by the GPL should be construed largely in the sense of copy/paste (this is a common armchair opinion about GPL issues, though I can't speak to its validity). In any event, whether or not such a "modification" can be construed to have been done in such a way (e.g. "clean room", as per CW) so as to skirt the GPL issue is yet another question.

Here is a random opinion: http://www.iusmentis.com/computerprogra ... ty-threat/

Basing a product on open source software

The GPL permits distribution of works that are based on a work under the GPL only if such distribution is done under the terms of the GPL as well. There is unfortunately no clear definition of what exactly constitutes "based on". The straightforward case is when source code from a work under the GPL is copied into another application. Statically linking several modules together to create an application would also seem to be covered. Dynamically linking two modules together is already a quite controversial case and by using technologies such as CORBA, RPCs, sockets or plug-in modules the situations can grow quite complex very quickly.

There is no clear consensus on the interpretation of the GPL. An often heard compromise is to assume that any form of linking invokes the sublicensing clause, but using inter-process communication does not. It should however be noted that using such technologies merely to avoid the consequences of the GPL will not be appreciated by the courts.

The Library GPL and the MPL are less problematic. It is permitted to create a "larger work" by combining the work under the MPL or LGPL with other software. This larger work may then be distributed under any licensing terms, as long as the conditions of the MPL or LGPL are adhered to for the portions of the larger work which were originally obtained under those licenses. This means that the source code of these parts have to be made available, including any modifications, and the authors of these parts must be credited in the documentation of the larger work.

[kingliveson]

Though programming background is not necessarily required to understand the report for which he bases his opinion, it does not hurt. The following quote does give me a little pause:

(i) although I am a qualified lawyer, I don’t specialise in or know anything about copyright law and reviewing licence agreements is not part of my job, (ii) I am wholly ignorant of computer programming.

Nevertheless, he does have the qualifications to read and interpret legal documents.

[orgfert]

http://talkchess.com/forum/viewtopic.ph ... 52&t=37817
Nice to see someone with legal qualifications actually speak about this.

But it looks like GCP is taking the lawyer to school in that thread.

http://talkchess.com/forum/viewtopic.ph ... 831#390831

His view boils down to the concept that the GPL depends upon the specifics of local copyright law.

[BB+]

If nothing else, lawyers are good at boiling down difficult situations into specific questions, here it being the meaning of "derivative work". The dispute with GCP seems to be over the explanatory gloss (in the Gnu GPL) on what that term means. I agree with Manis that a court would be likely to take the GPL's definition(s) if there were no other guidance, though GCP (and HGM) is correct that local copyright law would have precedence if there is clarity down such a path (and as to "modification" and "portion").

[hyatt]

If nothing else, lawyers are good at boiling down difficult situations into specific questions, here it being the meaning of "derivative work". The dispute with GCP seems to be over the explanatory gloss (in the Gnu GPL) on what that term means. I agree with Manis that a court would be likely to take the GPL's definition(s) if there were no other guidance, though GCP (and HGM) is correct that local copyright law would have precedence if there is clarity down such a path (and as to "modification" and "portion").

The basic idea here is that copyright law is not exactly well-suited with respect to computer programs. In a book, if you copy parts, you are guilty of copyright infringement. If you rewrite everything so that the similarity is minimal, then you are not. Software is different, since it is more of a black box that accepts input and produces output. But there are many implementations of this black box. And the implementations might look completely different when compared as text (one uses while loops, other uses for, one uses lots of gotos, the other doesn't, one uses arrays, one uses pointers. The list goes on. That might ruin a good work of fiction, but for software, the code itself is not so important compared to the output it produces. In a book, one might get by stealing the plot and making enough changes to make the comparison a little cloudy. For a program, where the source implementation is irrelevant, the issue is more complex. And current copyright law simply does not address that specifically.

There are tools used in university CS programs (to detect copying) that compare programs semantically as opposed to syntactically. But they are very limited and primarily catch the things I mentioned above. There are still lots of ways to fool them because we have so much freedom syntactically in a programming language like C. There are an infinite number of ways to syntactically express a particular semantic idea. That is a problem, to say the least...

[marcelk]

The basic idea here is that copyright law is not exactly well-suited with respect to computer programs. In a book, if you copy parts, you are guilty of copyright infringement. If you rewrite everything so that the similarity is minimal, then you are not. Software is different, since it is more of a black box that accepts input and produces output. But there are many implementations of this black box. And the implementations might look completely different when compared as text (one uses while loops, other uses for, one uses lots of gotos, the other doesn't, one uses arrays, one uses pointers. The list goes on. [...] And current copyright law simply does not address that specifically.

Copyright law doesn't have to address what you describe because patent law already does exactly that.

[BB+]

Seems that Alan Sassler brought up this article back when Rybka/Strelka was in vogue: http://www.btlj.org/data/articles/21_04_04.pdf
Or in a nicer format (to me): http://www.cs.berkeley.edu/~tlavian/pub ... 060403.pdf

Dangerous Liaisons—Software Combinations as Derivative Works?
Distribution, Installation and Execution of Linked Programs under Copyright Law, Commdercial Licenses and the GPL
By Lothar Determann

[hyatt]

The basic idea here is that copyright law is not exactly well-suited with respect to computer programs. In a book, if you copy parts, you are guilty of copyright infringement. If you rewrite everything so that the similarity is minimal, then you are not. Software is different, since it is more of a black box that accepts input and produces output. But there are many implementations of this black box. And the implementations might look completely different when compared as text (one uses while loops, other uses for, one uses lots of gotos, the other doesn't, one uses arrays, one uses pointers. The list goes on. [...] And current copyright law simply does not address that specifically.

Copyright law doesn't have to address what you describe because patent law already does exactly that.

Actually, it does not. The US Patent Office will not issue a patent to a computer algorithm today. They did years ago and found themselves in a foolish position. For example, they allowed someone to patent a program that simply translates a piece of music from one key to another. Which is similar to patenting an idea to compute Pi or something else.

I can't patent anything without a description detailed enough so that someone else could implement the thing given the necessary skill and equipment. In a casino, the continuous shuffling machines sometimes used to thwart card counters at the blackjack games has been patented. If you go look at the patent, the idea, and the implementation are clearly defined. So that it is specific enough that an infringement can be clearly identified. You can patent a device. Or a process, such as some new way of extracting iron from iron ore, etc. But you can't just say "I want to patent the idea of faster-than-light space travel." Nor can you patent basic things like a screwdriver, or a screw itself, unless there is something unique to focus on...

Software lies outside of that...

[Hood]

A programming language is a language. . It is a way of expressing ideas in a way computers understand.

I will be not surprised if the rules applicable to a written text are applied to written programms.

That case only syntactical check is possible and in my opinion real to use and execute. It will eliminate clones.

Semantics checking is like an algorithm checking and the algorithm is an expression of an idea. It is no to prove a patent law breaking.

That noise concerning derivatives not allowed to be on rating list was business controlled action to block competition- black advertisement.

Rgds Hood

[hyatt]

A programming language is a language. . It is a way of expressing ideas in a way computers understand.

I will be not surprised if the rules applicable to a written text are applied to written programms.

That case only syntactical check is possible and in my opinion real to use and execute. It will eliminate clones.

Semantics checking is like an algorithm checking and the algorithm is an expression of an idea. It is no to prove a patent law breaking.

That noise concerning derivatives not allowed to be on rating list was business controlled action to block competition- black advertisement.

Rgds Hood

They do, but that's a problem. If you take a book, and completely rewrite it, so that the characters, locations, actions, etc are all different, even though the story is the same, copyright claims are still possible based on the plot of the story taken in context.

With programs, things are different. There are an infinite number of ways to express the same simple algorithm, yet copyright applies to the written text, not what the program does. And that is a real problem.

Copyright law really wasn't intended to deal with computer software, that was for patent law. But then patent law was considered vague enough in this case that it doesn't work well either...