Ted Gould’s debate with Bradley Kuhn and others about the Canonical Copyright Assignment Agreement (CAA) is quite illustrative and one of Ted’s remarks provides a good launching pad for me to express why I find the CAA so objectionable.
I would say that those who don’t trust Canonical to be good stewards of the project with their patch, shouldn’t assign copyright. (reference)
The ability to trust Canonical is indeed one of the problems with the CAA. Canonical does a lot of good work in the free software domain, and its work on Ubuntu is funded by philanthropic donations from its founder. These are great things, without which Ubuntu would not exist. But Canonical is not a charity, it is a company with aspirations of profitability, and contracts frequently with companies who may not have the same commitment to free software as it does. Furthermore Canonical is not only a free software company. It has non-free software in its portfolio, some of which has been placed by it into the Ubuntu namespace. For these reasons I can definitely see why some people would have trouble blindly trusting that Canonical will never use their code in proprietary software. It seems unlikely, even extremely unlikely, but not impossible. That’s particularly the case given that Canonical expressly reserves the right to do it in this clause of the CAA:
Canonical will ordinarily make the Assigned Contributions available to the public under a “Free Software Licence”, according to the definition of that term published by the Free Software Foundation from time to time. Canonical may also, in its discretion, make the Assigned Contributions available to the public under other license terms. (emphasis added)
But this is not the main problem with the CAA. The main problem is that people should not have to decide whether they are prepared to give this trust to Canonical. It is unnecessary, and inappropriate.
- Unnecessary because there is a perfectly simple alternative solution – to include within the CAA a provision which guarantees that the assigned code will not be used in proprietary software. Other CAAs do this.
- Also unnecessary because it is very hard to see why Canonical needs to own copyright in affected code. The stated justification on the Canonical FAQ is “Canonical both uses and distributes software around the world. We need to make sure we are legally entitled to do so with contributed code, in a way that will hold up everywhere.” Whatever other reason Canonical may have for requiring copyright assignment, that one is nonsense. Canonical doesn’t have copyright to a lot of the code which it ships in Ubuntu (because the majority of Ubuntu’s code comes from upstream) and this doesn’t affect Canonical’s ability to redistribute that code, and it does so freely. I have not seen any response which addresses this issue on any Ubuntu mailing list or on Planet Ubuntu, despite having raised it on Ubuntu’s development mailing list in January 2010.
- Inappropriate because the individual is giving up legal rights by signing the CAA. That action means that it is for Canonical, as the recipient of those rights, to satisfy the individual that the rights will be used in the appropriate way, not the other way around.
The CAA is a legal document. As a lawyer drafting agreements on a daily basis (albeit not in this sector), I know that in such documents, it is unacceptable to leave eventualities to trust or hope. If you have a desired consequence in a specific situation, you include a provision dealing with it. I am sure that Canonical would not sign any commercial agreement which left a particular consequence down to trust.
Until these issues are resolved, I won’t be signing the Canonical CAA.
I signed the CAA because Ubiquity KDE needs someone to look after it, and Ubiquity’s Accessibility stuff could really use more than one person, but that quoted bit of it *really* annoyed me. “Ordinarily”? Not “always”? Gah! If I worked for Canonical, I’d be ok with it as a “work for hire” clause (I’ve worked a few places that had “work for hire… except anything under an open source license” in their contracts), but as a volunteer it’s frustrating that I have to sign that so that Canonical can have an installer that works with a screenreader and doesn’t crash on KDE.
Matthew, you and Mackenzie might do well to read Aaron Siego’s excellent analysis of the many reasons he found for not signing the CCA
http://aseigo.blogspot.com/2010/09/copyright-assignments-gone-wild-or-why.html
Mackenzie might be particularly interested in the legal ramifications Aaron spotted.
Matthew, you and Mackenzie might be interested to read Aaron Siego’s excellent analysis of the CCA and his several reasons for not signing it.
http://aseigo.blogspot.com/2010/09/copyright-assignments-gone-wild-or-why.html
Mackenzie might do well to take note of the legal ramifications of the document.
As to the 2nd “unnecessary” point,
1. If it is truly unnecessary, why are so many organizations part of the harmony project? Seems like others think its important too. Not just Canonical.
2. Just because its not being done now doesn’t mean its not necessary.
3. More than likely, the US is the big problem, less so than the rest of the world. As the US coerces other countries to adopt our patent and copyright mess, the issues will get more complicated. Isn’t it better to get out in front of a potential problem?
Dan
David:
Basically, I decided that making Ubuntu accessible was more important to me than the super-freedom bit. The accessibility in Ubiquity is bad enough that it just recites variable names at the user. I care too much about accessibility to just leave it like that.
Hi Matthew,
Related to point 2, I did some research and thought about it and I think that the reason Canonical might want to own complete ownership of the codebase of it’s own projects might be because of two reasons
1) Changing the license terms, without need to contact all the stakeholder. Canonical holds the copyright over libzeitgeist library and it was re-licensed from LGPLv3 to LGPLv 2.1+
I do agree that codebase can also be relicensed to make it non-free, but I doubt Canonical might ever do it (still fingers crossed, I can’t read minds)
2) Having complete ownership of the codebase might improves the market value of the company. Take the example of Sun. It owned OpenOffice, so Oracle knew that it has complete control over the codebase of OO.o
If someone acquire the company and sees that the company does not have a single product over which that company had complete control, they might not want to acquire it or might want to pay lesser.
I am not saying Canonical might be acquired. Neither, I am saying that market valuation will surely be higher, but there are chances of being so. Economies is a complex thing anyway.
P.S: This was my neutral commentary. I am not on Canonical payroll nor had been.
Manish, interesting thoughts, thanks. Neither of these are the stated reasons given in the FAQ. I think that item 1 is a reason often given for copyright assignment, but in this particular case, it hasn’t been explained why Canonical needs to own copyright in libzeitgeist but is happy not to own copyright in other components used in Ubuntu, for example.
But the first and third bullet points are more important to me than the second one. If Canonical came up with a proper justification of their need to own copyright in certain projects, rather than the nonsense one currently given, then I wouldn’t necessarily have an objection to signing a CAA, if phrased in appropriate terms that guaranteed that the relevant code would remain free. But Canonical’s CAA doesn’t do that.
Dan – those aren’t really answers to the point.
David – yes, I was interested, thanks for the link.
If you do not like the CAA and still want to benefit the progress of Ubuntu then you simply submit your contribution upstream, end of story. Everyone benefits, in just the way that most people envisage the FOSS ecosystem.
This little island of trademarkery and proprietorialism needs no support from the community, because Canonical pays its lawyers and analysts plenty to do that job. Let them buy or develop whatever goes in their island.
Stuart,
I agree that contributing upstream is a valuable thing to do. However, if your message is that if I don’t like something in the Ubuntu community, I should just do something else, then I fundamentally disagree. As an Ubuntu member and member of the Ubuntu Community Council I feel that it is my role to try to improve issues in the community which could affect volunteers’ ability to participate, and this is one such area.
Apart from anything else, my contributions to Ubuntu fall primarily in areas where signature of the CAA isn’t (yet) needed: the Ubuntu Documentation Project doesn’t currently assign copyright to Canonical, partly because many of the senior contributors feel the same way as I do about it. But if that project were to contribute documentation to the Unity project, rather than keeping documentation relevant to Unity in its own namespace, it would need to assign copyright. That is unfortunate.
simple solution:
write some great patches for ubiquity or unity or whatever. dont assign the copyright, just label them GPL 3+ (or your favourite compatible licence). maintain a superior branch of the project in your own bzr branch.
anyone who wants to use them under the terms of the GPL can. If canonical want to use them, they have to change their policy. if canonical are too rigid in there demands of CA, then they loose out.
I’ve always wondered why you have to assign anyone your copyright, instead of just granting them an indefinite licence. Surely that’s the other way around the problem?
Canonical asks for copyright assignment only in projects which they start and not in projects started by others.
libzeitgeist development was funded by Canonical, so it came under CAA.
I think this is the biggest gripe with CAA. If they gave the promise not to release it under other licenses, then I think many people would be satisfied and sign CAA.
Start, that does not solve the problem. Canonical has only those projects under CAA which they started and not upstream projects. So the upstream for those projects is Canonical itself.
Two things:
1) Make sure the license you choose is compatible with the license used by ubiquity/unity
2) You can maintain your superior branch, but maintaining is really tough. Chances are less it would make it into official repos. You would have to rely on PPAs for distribution.
This solution isn’t as simple as it looks.
“2) You can maintain your superior branch, but maintaining is really tough. Chances are less it would make it into official repos. You would have to rely on PPAs for distribution.” — Manish Sinha
Go-oo was a project that ran roughly as i described. it was the default openoffice in most linux distros. so if you do it well enough it will be used (though maybe not by ubuntu).
Yes. Actually what we saw as OpenOffice.org was actually Go-oo but, as you mentioned in this case – it would be tough to get your fork into Ubuntu or replace the official one.
You can get it in other distros, but the point being Ubuntu has a huge userbase. You would miss a large amount of new ubuntu users with this approach. This looks like a workaround and not a solution
Dan: the Harmony project provides a range of possible CLAs, many of which do not include the most problematic aspects of the Ubuntu / Canonical CLA. Although Canonical is the prime mover behind Harmony, you can’t treat it and the specific Ubuntu CLA as the same thing. But there’s a lot of people who are not too happy about Harmony, even so.
Somewhat of a moot point isn’t it? I mean, upstart’s been around for a while, how many distro’s have adopted it?
And Unity, any other distro’s even showing interest?
How about the libapp indicator stuff… have those been picked up by any other distro?
And apparmor? Is anyone else using it?
At a certain point you have to wonder if they’ll be able to maintain all this in-house stuff without sourcing it back to the communities for adoption / control.
I’m genuinely curious if anyone knows of counter-examples… where some distro (Ubuntu or Red Hat or Suse) have developed something in house (under CAA or otherwise) that has done really well and been adopted widely. All that come’s to mind is GNome-menu by Suse I think it was, and it looks like LMDE might have adopted that as well.