LWN: Comments on "Canonical copyright assignment policy 'same as others' (ITWire)"

Perceived value

Baylink — Sun, 07 Feb 2010 19:47:37 +0000

On the latter point, sure.

On the former, yeah, if you don't require copyright assignments, and license under GPL, then sure you're giving up at least *some* control over the project: you can't relicense it to something proprietary without major (probably impossible) effort -- that is precisely the bargain under which lots of people contribute to GPLd project.

It's sort of a Faustian bargain, but in reverse.

Perceived value

dmag — Sun, 07 Feb 2010 04:31:25 +0000

>> I will cede *control*, in exchange for getting to take
>> advantage of lots of other people's development work.

> Copyright assignments break that model.

Two thoughts: First, there's nothing in the GPL that requires the company to cede "control". Second, that implies a loophole: a company can only "control" a dual license (with assignment) project as long as they are the "major player".

I'll bet if you hired a bunch of programmers, you could wrest exclusive copyright control from any GPL project. Just add a lot of neat features, post it widely and refuse to assign copyright (do it privately by just ignoring their emails). The users will complain that the "official" version is missing those features. The first few times this happens, the company will re-implement your features. But eventually, they will be unable to keep up because you only have to add features, while they also have to do bugfixes, documentation, etc..

I predict this will sort-of happen with MySQL: The Drizzle fork will become wildly successful. But unfortunately the old MySQL fork will stay valuable for a many years because there are a lot of conservative companies that will want tot stay with the "safe and proven" MySQL fork. (Heck, the fact that MySQL was worth $1B proves my point -- businesses are too used to paying for software.)

if copyright assignment is so evil, why don't you oppose FSF requiring it?

dlang — Wed, 03 Feb 2010 03:21:10 +0000

if the kernel copyrights had been signed over to the FSF, that code would definitely be GPLv3 right now, even over the objections of the developers. Given how opposed many of those developers are, it would probably have triggered a fork in the kernel.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

jmm82 — Wed, 03 Feb 2010 03:15:39 +0000

If the Linux kernel DID want to change its license it would an interesting
scene to say the least.

If copyright assignment of the Linux kernel had been given to the FSF then
the Linux kernel would most likely be gplv3 now. So non-profit or profit
assignment is assignment. There wouldn't be assignment if some rights were
not lost.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

rahulsundaram — Tue, 02 Feb 2010 13:55:42 +0000

No you have category misunderstood what I have been saying I never used
the word evil but it is important to note the clear differences between a
non profit foundation like FSF which offers counter guarantees and a for
profit company and the nature of differences between the copyright
licensing agreements

Whether individual kernel contributors have licensed it under GPLv2 only
or not may or may not be a problem That depends on the specific code in
question which might not even be copyrightable individually There has
been legal arguments placed which have evaluated workaround for such
things discussed in LWN as well before I encourage people to look those
up

Nobody claimed Free software was merely GPL but it is important to note
that permissively licensed code is equally permissive to everybody but
copyright assignments create special privileges to one entity which is a
different case and potential contributors shouldn't be fooled by merely
looking at the license without carefully evaluating whether the terms of
the copyright license agreements are ok with them

if copyright assignment is so evil, why don't you oppose FSF requiring it?

hppnq — Tue, 02 Feb 2010 13:45:13 +0000

That's ... ironic. The angle of Evil Company's profit from Innocent Hacker's hard work does not really make sense, because it assumes the conditions under which patches were contributed were not understood. Anyone who chooses a license should do so carefully, whatever the license says.

Some kernel contributors have explicitly stipulated that their code should only be distributed under the GPLv2 and NOT any later version, and you seem to be saying that this is not only a wrong interpretation, but also not a problem. Maybe I misunderstood.

But more importantly: free software really is a bit more than just the GPL. It allows for proprietary software. If you don't like where a particular piece of software is going, you can fork it. This is true for all free software projects.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

rahulsundaram — Tue, 02 Feb 2010 09:32:08 +0000

I dont see any intersection and I have yet to see a single lawyer claim that
copyright assignment is actually necessary in the case of Linux kernel even
assuming kernel developers want to shift to GPLv3 which they dont and also
copyright assignment to a corporation is different to a non profit foundation
with counter guarantees within the legal agreement to not make it proprietary

if copyright assignment is so evil, why don't you oppose FSF requiring it?

jmm82 — Tue, 02 Feb 2010 05:52:42 +0000

"But what's the relevance of the licencing policies of the kernel development
community to a statement of Canonical regarding copyright assignments ?"

I believe the point was the same people that are mad that the Linux kernel
can not change its license to gplv3 due to the lackof single point of
copyright ownership, are the same people complaining when some company asks
for assignment.

Canonical copyright assignment policy 'same as others' (ITWire)

jmm82 — Tue, 02 Feb 2010 05:25:34 +0000

It's all about the PageRank ;)

if copyright assignment is so evil, why don't you oppose FSF requiring it?

SEJeff — Tue, 02 Feb 2010 05:19:58 +0000

fwiw, between pcc from the bsd camp and llvm, we have healthy competition for
gcc. Granted gcc is still light years ahead of both, but we have competition.
In the end, competition is good for all free or not.

Canonical copyright assignment policy 'same as others' (ITWire)

Baylink — Tue, 02 Feb 2010 02:02:36 +0000

Further thoughts occurred to me after reading your post:

The point of the GPL is to *take the asymmetricality out* of the power imbalance between large organizations and individual contributors.

Perceived value

Baylink — Tue, 02 Feb 2010 02:00:25 +0000

I find it insightful, but I don't buy it.

The bargain, if you're a company, considering open source licensing, whether it's on new code, or someone else's project you're adopting, is very simple:

"""
I will cede *control*, in exchange for getting to take advantage of lots of other people's development work.
"""

Perceived value

SLi — Mon, 01 Feb 2010 23:48:43 +0000

Thank you. I found that quite insightful.

Canonical copyright assignment policy 'same as others' (ITWire)

Pc5Y9sbv — Mon, 01 Feb 2010 23:24:47 +0000

On a largish, NSF-funded, university-based project I worked on several years ago, there was a similar problem. We actually were using a more BSD-style license on our project, but our university lawyers told us we needed to get what was almost as strong as copyright assignment from outside contributors. Their reasoning was that they needed legally executed contracts from those contributors in order to know that we could really use the code as the license supposedly claimed.

It was very frustrating for many contributors who felt insulted that we needed such a contract executed by their employer. But the difficulty in getting such things signed was a good demonstration of the problem: developers thought they could slap a license text on a file and send it to us, but their work is actually work-for-hire for their employer, and the employer is the only one capable of giving us the legal right to use the code in our project. We were not interested in helping to "launder" what could in effect be stolen code.

It's about proprietary forks, of course

rickmoen — Mon, 01 Feb 2010 22:56:51 +0000

Baylink wrote:

Mark's missing it. Copyright assignment policies are an end run around the GPL: They're an attempt to make it practical to *take the GPL off* of code which currently carries it.

It might be pointed out that Shuttleworth rather flagrantly ignores the very question he himself posed: "The most common complaint I've heard is 'why can't a company accept my patches to them under the same licence that they give me the original code?" He asks that, and then changes the subject to what "that suggests".

Why can't a company do that, then? Correct answer: "Of course we can, but would rather not, because we want to retain he option of proprietary licensing." Shuttleworth is forthright about that, higher up in the article: ""All copyright assignment agreements empower dual licensing, and relicensing, and as a result such projects attract more investment than other projects which don't create the same opportunities for underwriting companies."

Rick Moen
rick@linuxmafia.com

if copyright assignment is so evil, why don't you oppose FSF requiring it?

cmccabe — Mon, 01 Feb 2010 19:39:30 +0000

I'm against copyright assignment, whether the copyright goes to Mark Shuttleworth or Richard Stallman-- and for the same reasons.

Someone who contributes to a project should be able to count on that project's license staying the same. When there is a single copyright holder who controls everything, the license could change at any time. A project that started as (say) an image editor could turn into a crusade against some arbitrary perceived social evil. Or it could become another piece of proprietary software. Developers should know what they're signing up for ahead of time, rather than being pawns on a chessboard.

Corporations in general avoid projects with capricious and unpredictable licensing terms, because they're a business risk. If the project leader decides he doesn't like your business model, or the other things your company does he could easily rewrite the license so that you couldn't use the code.

Canonical copyright assignment policy...

butlerm — Mon, 01 Feb 2010 18:42:05 +0000

Does that mean if you later implement the same solution as the patch that you've violated copyright of the original patch's author?

In general, no.

If a third party, such as a patch reviewer, views such a patch does it taint them from contributing to the project further?

In general, no. If the patch is non-trivial, it may preclude the reviewer from writing a patch that is "substantially similar" to the one that was submitted, however. Most patches are too trivial to be so protected, however.

And certainly it is extraordinarily bad faith to submit a patch for review that one does not intend to have merged under the licensing and copyright assignment policies of the project concerned. (This is my understanding, consult a lawyer about real situations, etc.)

The FSF is a special case

dlang — Mon, 01 Feb 2010 17:30:00 +0000

if they promised to always keep the codebase under the license it was originally under, but then wanted the ability to add additional licenses to that I think it would be much better than the current situation

Ship of Theseus

hppnq — Mon, 01 Feb 2010 16:56:40 +0000

I guess the people who contribute to Launchpad already profit from it in some way. In any case, they should know what a copyright assignment entails. There is nothing really sleazy about it. So you don't send in a patch, fair enough.

But more importantly, what you seem to forget is the quite inevitable and considerable spin-off. Free software remains free, even if it's dual licensed (which does not automatically mean proprietary). There are many projects, individuals and companies who in any case profit from the work being done on such projects -- in other ways.

Canonical copyright assignment policy 'same as others' (ITWire)

MattPerry — Mon, 01 Feb 2010 16:44:41 +0000

What happens if a user submits a patch but then refuses to turn over copyright? Does that mean if you later implement the same solution as the patch that you've violated copyright of the original patch's author? If a third party, such as a patch reviewer, views such a patch does it taint them from contributing to the project further?

Canonical copyright assignment policy 'same as others' (ITWire)

SEMW — Mon, 01 Feb 2010 16:17:38 +0000

>>It only does so if you plan to release that code under some other license.

> Clearly you cannot support a product if you can't sell it under a non-open-source license </sarcasm>

Oh, come on. You read LWN, you've read the recurring stories on the problems caused to various projects by incompatibilities between different free software licences. What was the latest one, LGPLv3 and GPLv2, I think? Anyway, the point is that it clearly isn't as simple as "I've released it as Open Source, the only reason anyone could want to change license is to make it proprietary!". Sometimes changing to a different open-source license is desirable, or even necessary.

Of course, copyright assignment is not always necessary for license change. If you're a project the size of Wikipedia, you can avoid having copyright assignment and still change license if it becomes necessary. It involves begging Richard Stallman to, as a strictly one-time-only exception, make a special version of the license designed to apply to only your project that gives you an escape route. This is, to put it mildly, not an option open to most open source projects.

(Since e.g. the MIT license is relicensable to pretty much everything, you could ask that contributions be under that and solve the above problem just as well (though, of course, all of the previous posters objections about their code being taken proprietary would apply just the same as it would in the case of copyright assignment)).

Ship of Theseus

did447 — Mon, 01 Feb 2010 16:05:07 +0000

> but who's to say that will be the case forever?

For contributions from individuals? I'd say the taxman.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

andrel — Mon, 01 Feb 2010 15:45:00 +0000

In practice private corporations are more answerable to shareholders than public companies. Canonical's shareholder runs the company as he sees fit.

Ship of Theseus

magnuson — Mon, 01 Feb 2010 15:06:18 +0000

An interesting position that all of the value of the codebase is present in the originally posted version and all subsequent patches are of "no value" apart from the main branch.

What happens when as a result of these valueless (on their own) patches the entirety of the code base has been replaced by contributed code? All of the "value" of the original code is now gone, or at least transformed, into the current code. A vanishingly small percentage of Linux these days was actually authored by Linus himself and an assignment policy like this would enable him to take it proprietary at any time. Profiting off of the backs of countless contributors.

I expect the Canonical codebase is considerable more insular in that a far smaller percentage of it is coded by people not employed by Canonical, but who's to say that will be the case forever?

Canonical copyright assignment policy 'same as others' (ITWire)

dlang — Mon, 01 Feb 2010 14:24:30 +0000

there's nothing inherent in being a non-profit organization that makes it 'good'

there are good non-profit organizations, and bad non-profit organizations.

it wouldn't surprise me to learn that the RIAA and BSA on non-profits for example.

Companies have a purpose, specified in their initial paperwork...

tialaramex — Mon, 01 Feb 2010 14:14:23 +0000

Actually a company can be set up with arbitrary rules. It's unlikely that you'd manage to IPO a company with rules that made commercial success very difficult to achieve, but nothing stops a private company (and Canonical is a private company) from having such rules.

It is very common for people (even supposedly "smart" stock market people) to imagine that all for-profits are somehow interchangeable profit hunting outfits, mercenaries willing to do anything to make a buck and differentiated only in where they're starting from in that hunt. In fact nothing could be further from the truth. Nearly all companies have specific goals outside of "make a profit".

Of course the owners (in total) of the company can wind it up and start over, but that's uncommon and cannot be forced by a single greedy shareholder in a public company. It's also very easy to engineer things so that a radical change would trigger "change of control" conditions, break contracts with partners and release key employees under favourable conditions. An unpopular change in this case would be corporate "suicide".

Canonical copyright assignment policy 'same as others' (ITWire)

bug1 — Mon, 01 Feb 2010 14:13:30 +0000

Whether thats a good or bad thing depends on the future actions of the person/company.

Assigning copyrights to a non-profit organization seems pretty reasonable to me.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

skvidal — Mon, 01 Feb 2010 13:47:57 +0000

Canonical is a privately held company. They aren't beholding to anyone necessarily.

One of the virtues of being privately held is doing what you want/think is right BEFORE answering to shareholders.

Canonical copyright assignment policy 'same as others' (ITWire)

rahulsundaram — Mon, 01 Feb 2010 11:29:20 +0000

In that case both parties are getting special privileges and not all
copyright assignments are the same and amount of contributions should be
under consideration as well

Canonical copyright assignment policy 'same as others' (ITWire)

epa — Mon, 01 Feb 2010 10:21:17 +0000

The key point is that while Canonical requires copyright assignment, it is also happy to grant copyright assignment when Canonical employees send patches to projects run by others. So it doesn't appear that any one party is getting special privileges.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

nim-nim — Mon, 01 Feb 2010 10:08:33 +0000

You can't oppose the FSF and Cygnus.

Cygnus (and now Red Hat) have largely been successful because they've embraced the GPL and worked with the FSF instead of trying to cross it as many others did and still try to do.

Some people are so set in their no-FSF or no-RMS world they do not realise that in practice working with the FSF or RMS is not the disaster they paint.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

Zack — Mon, 01 Feb 2010 08:37:11 +0000

>for all of you stating or implying that Mark and Cononical are evil for requiring copyright assignment, are you equally opposed to the FSF requiring copyright assignment?

I see no comments on this article (so far) that state or imply that Mark or Canonical is evil for requiring copyright assignments. There are posts questioning the reasoning Mark gives and posts where people explain their point of view on the difference.

>do you also decry the fact that the linux kernel isn't switching to GPLv3 (and probably never will since they don't require copyright assignment)?

Not switching to GPLv3 does underline how clumsily copyrights are generally handled in the kernel development community.
But what's the relevance of the licencing policies of the kernel development community to a statement of Canonical regarding copyright assignments ?

>the comments in this thread reek of hypocrisy.

Again, which comments ? Posters here have a tendency to not only state their opinion, but also elaborate on how they reached the conclusion that led to that opinion. If they are incorrect, many would not object to having explained to them why they are incorrect.
Just lumping them all together and stating what basically amounts to "You're all hypocrites", isn't helpful.

The only ones that can be considered negative are the ones wondering whether the copyright assignment will result in software becoming proprietary, which is not an invalid concern.

Understanding RMS

emk — Mon, 01 Feb 2010 07:05:43 +0000

When dealing with RMS, it helps to understand his goals:

1) He wants to provide as many users as possible with the rights to
examine, modify and redistribute software.

2) He has zero interest in helping developers of proprietary software,
Tivoized embedded systems, or DRM. Under certain very limited
circumstances he will occasionally choose a non-copyleft license, but only
when it has a direct and clear advantage for (1).

The GPLv3 may not be what Linus wanted, but it is absolutely consistent
with what RMS wants: Allowing users to modify the software they use. In
the future, for example, you can expect RMS to become increasingly
bothered by Apple-style "App Stores" and by hosted web applications.

You may disagree with RMS (I often have in the past), but his actions are
not terribly hard to predict.

Perceived value

robla — Mon, 01 Feb 2010 07:01:43 +0000

The reasoning he gives, comparing the relative value of the two contributions, may arguably be a case for why an upstream project shouldn't have to accept a patch under the same licence terms as they make the whole codebase available, but it doesn't at all address the question of why cannot do so.

The problem that big companies face when taking patches to projects they initiate+sponsor is sort of a nasty game theory problem. When they create a new project, they're adding an asset to their balance sheet that they've got complete freedom to do with what they want. One thing they can do is release the source code under an open source license, which we all celebrate as a great thing. They lose resale value, but they don't lose the ability to relicense it. Even if there is no intention to ever relicense that code, the mere ability to do so adds to the balance sheet of a company. For example, there is no way MySQL Inc would have sold for $1 billion had the company not had exclusive ownership of the MySQL database.

Open source projects of all stripes rarely get big, tangible features they truly lust after as outside contributions. Those big features are generally developed as separate projects. Instead, they get lots of small patches and improvements as outside contributions.

Taken individually, each patch is valuable, but not a game changer. So, for any one of these patches, as a sponsor of the project, the tradeoff between taking that one patch versus forever losing the ability to relicense the source code is no brainer. As in "no, don't take the patch". If the contributor can be convinced to relicense the patch, great, but otherwise, forget it.

While it may be possible to argue that the cumulative value of the delta between those patches one gets with a contributor agreement versus those that one would have gotten without requiring one is worth it over time, it's very difficult to know what that delta is. Since that number is almost certainly unknowable, so is the benefit. And therein lies the rub.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

AndreE — Mon, 01 Feb 2010 05:39:32 +0000

I never claimed that it was solution to solve all problems.

Ultimately though, the FSF didn't just decide to make new releases of GCC under a less free license, like any commercial entity might consider. In fact, their zeal is appealing to me. You may consider it "arbitrary" freedom, but having entities that place the freedom of software as their primary concern are obviously the entities mostly like to always protect these freedoms.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

nevyn — Mon, 01 Feb 2010 05:26:04 +0000

> Without the FSF, we wouldn't have a high-quality free compiler at all.

That's obviously false, it's like saying without the FSF we wouldn't have a
high-quality free editor. Yes, RMS and the FSF helped GCC along and it is
still the de-facto free compiler but other people have written compilers and
Cygnus (and others) have put a huge amount of effort into it.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

quotemstr — Mon, 01 Feb 2010 05:06:28 +0000

Without the FSF, we wouldn't have a high-quality free compiler at all.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

elanthis — Mon, 01 Feb 2010 04:06:12 +0000

RMS and the FSF have actively held back the improvement of developers'
lives out of an insane need for arbitrary Freedom (look how long it took for
GCC to get plugin support), and that alone makes them wholly untrustworthy
to me.

if copyright assignment is so evil, why don't you oppose FSF requiring it?

AndreE — Mon, 01 Feb 2010 03:32:29 +0000

Canonical is a for-profit company. Their comittment is to shareholders, not to any community or ideology.

The Free Software Foundation is ostensibly committed to the ideologies around the Free Software Movement.

It's all a question of trust. I trust a promiment and zealous advocacy group way more than a commercial entity whose interest in protecting freedom is mostly defined commerically. Even when RMS dies there are many individuals fully committed to libre software movement and I doubt any sort of subversive attempt to make the FSF adopt a more proprietary-friendly stance would go unchecked.

This doesn't just go for Canonical. Any company in the Linux space is ultimately less trust worthy than the FSF.

The FSF is a special case

corbet — Mon, 01 Feb 2010 03:00:00 +0000

FWIW, the FSF commits itself to something rather stronger than "similar in spirit." They say exactly which rights will remain with the code. One might still not want to sign code over to the FSF, but that doesn't change the fact that it is a very different deal than signing it over to a corporation which intends to release proprietary versions.