Crafting a special kind of license for a very special standard.
matthew.flaschen at gatech.edu
Sat Apr 14 23:33:09 UTC 2007
Grant Robertson wrote:
> Isn't that what the GPL does every day? It controls what can be done
> with code that is copyrighted but not patented.
An implementation of a standard is not derived from the standard in the
copyright law sense. Thus, even if you GPLed the standard, proprietary
software could still implement it.
> The GPL even claims the right to force a secondary
> user to release ALL the code that they themselves wrote separately just
> because they used one piece of GPLed code.
That's not true. This only applies if the new work the distribute can
be considered a derivative work of the original.
> Can I not similarly say that if someone wants to use part of a
> standard which is copyrighted that they have to abide by all the terms of the copyright license. How is
> that different from the GPL?
No, because there are no consequences if they ignore these rules. It is
not copyright infringement to implement the methods described in a
copyrighted work. If I buy a book on mousetraps, I don't need the
author's permission to build and sell their new clever mousetrap. If
the mousetrap invention were patented, then I would. The same applies
for a standard. If I acquire a copy of the standard, I'm free to
implement it without special authorization, unless there are relevant
I realized there /may/ be (remember, IANAL TINLA, etc.) is one way to
stop unauthorized implementations without patents. You could register
the standard as a trade secret, and require everyone to sign a contract.
They could agree:
a. not to describe the standard to anyone.
b. only implement it in the ways you describe
c. pay any relevant fees
Then, anyone who implemented the standard in an undesired way would be
in breach of contract; the same is true if they described the standard
However, allowing your standard to be implemented probably invalidates
the trade secret, especially if you allow open source implementations.
I would *not* recommend trying this. It is unethical (and clearly not a
real open standard) and legally dubious. The only reason I mention it
is that someone else is bound to bring up contracts, and this is the
most effective contract I can think of.
> True. Just as I can describe in my own words some code that has been
> released under the GPL and distribute that description. However, you
> can't compile that description and have an executable program.
No, but you can learn from reading GPL code and make a separate e.g.
proprietary implementation that does the same thing. It's a close call,
but that separate implementation need not infringe copyright if you do
> Similarly, one could describe a standard in their own words and
> publish that. However, A) that description would not be a working
But it could completely describe the requirements of the actual standard.
, B) if their description was close enough to the original
> standard and they were trying to claim that it were a standard then
> they could be sued for copyright violation just as someone can't write
> a "really detailed book report" that just happens to read almost like
> the original book and publish that.
It wouldn't be necessary to make it read like the original. If it was
written very differently, I don't think it would be a copyright
infringement. As an example, lots of people make explanations
(sometimes entire books) based on W3C recommendations and the W3C has no
copyright claim against them.
> It is called plagiarism.
Arguably, yes. But plagiarism is not a crime. It is not copyright
infringement, which is.
>> Only patents can control implementation.
> If that were true then the entire Open Source Initiative is a moot
Your analogy is really breaking down. Open source licenses are
necessary because by default I can't make a new version of a copyrighted
program; that's copyright infringement. Thus, if the author wants to
let me, they must grant permission. Every open source license (with the
exception of the thus-far unpopular contract-based ones) is such a
However, I can make a new program B that does *exactly* the same thing
as A. If I don't look at A's code (and even sometimes then), B is never
a copyright infringement. Similarly, if I implement your standard
without reading it, it is not copyright infringement. There are various
ways I could do this, such as reverse-engineering (same way B legally
mimics A) and reading a legal third-party description.
> I am going to have to give this careful consideration. Everything I
> have read so far seems to indicate that the experts in open standards
> believe you can prevent unauthorized implementations through a
> copyright license.
I don't think so. You haven't quoted any experts. The W3C link I gave
clearly implies patents are the real issue.
> Technically, one can only patent the method of
> implementation itself and not the idea the method implements.
In theory maybe (it depends what you mean by idea). In practice, the
USPTO has a bad habit of completely ignoring such distinctions.
> Meaning, I could patent my exact standard but I could not patent all possible
> standards for marking up educational material in a hierarchical
Actually, there's a good chance with the right patent lawyer you could.
It might not hold up in court, but as long as you intimidate your
opponents you won't need to sue, and it could work out fine.
> It seems to me that any software written to work with a standard could be
> considered to be "derived" from that standard. Although, I see how
> that is a stretch the courts may not agree with.
They shouldn't. This is precisely the point of patents (though you
shouldn't really be able to patent software, because it is ultimately a
mathematical algorithm, and those are not patentable) .
> If the OSI refuses to sanction the license
> agreement because it expects big corporations to give back a little of
> what they take then so be it.
This seems outside of OSI's scope since you're trying to license a
standard, not software.
> Finally, I will also have trademark control. I can simply refuse to
> allow anyone to use the trademark unless they abide by the copyright
Yes, that is a good point. That is for instance how Sun (err.. I mean
the totally independent JCP) has been keeping third-party Java
reimplementations in line (they do offer to certify correct third-party
implementations, but none are perfect and I believe the official testing
is expensive). However, there's a limit to how effective the trademark
method is. Note that there /are/ unauthorized third-party Java
implementations, and they are all incorrect to varying degrees (usually
not on purpose). They obviously don't call themselves Java, but they
sometimes say things like "Java reimplementation", "third party Java
implementation", "Java-compatible", and get away with it. This is
actually a really good example, because the Java standard and API is
copyrighted, but Sun couldn't sue unauthorized implementors for
> Heck, even if they do, at least I will have revolutionized education.
> I might not have anything to show for it, but at least I will have
> started something that has helped the whole world and that will be
> worth it.
I wish you the best of luck; try not to let the legal issues bog you
down too much, but do think them out.
More information about the License-discuss