EUPLv1.1 vs. GPLv3: What kind of freedom would you like?

Ernest Park recently wrote a comparison on Linux.com between two software licenses, the European Union Public License v1.1 (EUPLv1.1), authored by the European Union and the GNU General Public License v3 (GPLv3) authored by the Free Software Foundation. Although his comparison is quite extensive, I cannot agree with his conclusions, most of which may be summed up by his insinuation that “The EUPL is unique in its encouragement of interoperability, freedom, and lack of license lock-in upon redistribution.” Instead of creating a new FOSS license, I suggest that the EU throw their weight behind an existing license such as the GPLv3 or Apache License v2.

First off, let’s get some notes out of the way:

Mr. Park appears to be a specialist in FOSS licenses. His profile on FOSSbazaar.org states that he is “[a] specialist in license and software usage and proliferation; Expert in creation and management of FOSS usage policy.” As such, Mr. Park can be reasonably expected to understand the nuances of and differences between Open Source Software and Free Software.

The EUPLv1.1 and the GPLv3 are both Open Source Software licenses and Free Software licenses (says Wikipedia). I’m rather familiar with the GPL v2 and v3, as are most FOSS developers reading this post.

The EUPL is probably unknown to most of you. From my preliminary glances at the EUPL it is unlike other FOSS licenses I’ve seen before. It’s like someone took the idea of multi-licensing (as the Mozilla foundation does with Firefox) and wrote an explicit license that allows distribution under the EUPL as well as specialization (i.e. choosing just one of the “compatible” licenses), with a hard-coded list of GPLv2, OSL 2.1 & 3.0, CPL v1.0, EclipsePL v1.0, CeCILL v2.0.

Now let’s take a look at what Mr. Park wrote:

The European Union Public License (EUPL) version 1.1 was released…A major milestone in the evolution of open source licensing, the EUPL contains some important distinctions that set it apart from other popular open source licenses: it’s written with native language support for all EU member countries, it’s compatible by design with a number of OSI-approved licenses and — also by design — incompatible with the GNU General Public License version 3.0 (GPL v3).

This sentence is largely a statement of facts. That Mr. Park calls the EUPL “a major milestone in the evolution of open source licensing” tells us two things:

1. Mr. Park believes that the EUPLv1.1 provides a significant (“major milestone”) advantage (“evolution”) over previous licenses.
2. Mr. Park wants to compare these licenses on their merits as Open Source Software licenses.

Because Mr. Park believes that the EUPL provides a big advantage, I’m going to be looking for support of this argument throughout his comparison.

Because I believe in Software Freedom, I like to talk about Free Software and Free Software licenses. I’m fine with people using the term Open Source Software, but I think that in the community today OSS concentrates on business economics while Free Software talks about user freedom. Whenever I hear someone talk about “freedom” in relation to software licensing I think about Free Software.

[The EUPL is] compatible by design with a number of OSI-approved licenses and — also by design — incompatible with [GPLv3]. Could this possibly be a message to the Free Software Foundation (FSF) and the open source community that there really is a vital, innovative world outside the borders of the United States?

What does writing a different license have to do with indicating that there is a “vital, innovative” non-US world? I think it’s great that the EU is not only thinking about FOSS but is actually doing something to further it, and while I do wish that the EU had chosen a GPLv3-compatible license for several reasons, it’s their right to make a new license if they so wish. What’s Mr. Price’s angle here? Is he trying to start a fight between FSF and the EU?

I like to avoid speculation whenever possible, but if I had to make a guess as to why the EU wrote this license I would guess that they didn’t like what existing licenses offered (or else why write a new one?), and that they specifically did not like the FSF improvements in GPLv3 (as they include GPLv2 as compatible but not GPLv3).

While members of the open source community often bristle at the mention of yet another license, many people have welcomed this one.

Who has welcomed this license? Has any prominent FOSS project moved to it or has any bigwig like Torvalds, ESR, or a present/past DPL spoken favorably about it? The first time I heard about the EUPL was when Mr. Price’s review was mentioned on Identi.ca.

The EUPL is unique in its encouragement of interoperability, freedom, and lack of license lock-in upon redistribution.

I’m not sure that the EUPL is any better than other FOSS licenses at these aims. Specifically:

interoperability – How does it encourage more file format or protocol interoperability?

freedom – Compared to the EUPL, a 3-clause BSD license is more permissive and the GPLv3 provides greater protection of Software Freedoms. What is the benefit of using the EUPL?

lack of license lock-in upon redistribution – The EUPL is a multi-license and as such code licensed under it could be “specialized” down to one of its “compatible” licenses. While some might read this as a lack of license lock-in, the EUPL only lists 5 licenses. Would a hypothetical new “Qubit Public License” that allows specialization to any OSI-compatible license be better than the EUPL? No — I think it would just be silly.

By making specific distinctions, [the EUPL] represents a divergence from the typical freedoms afforded by whatever the latest iteration of the GPL may be.

The “typical freedoms” afforded by the GPL family of licenses are the freedoms necessary to read, run, and share Free Software. If the EUPL offers a different set of freedoms, what are these freedoms and how are they better than those offered by, say, the GPLv3?

Mr. Price then quotes Wikipedia:

The European Union also wishes to dispel legal uncertainties, real or perceived…by creating a software license which takes due account of European Union Law… [T]o dispel fears of license proliferation, the license [also] …specifically authorizes [re-release under the] GNU General Public License (GPL) v. 2 …[and other licenses].

Let me get this straight: In order to dispel legal issues with licenses like the GPLv2, the EU made a new license…which explicitly allows relicensing of code under the GPLv2? That’s a little confusing. It’s also unfortunate that they chose the GPLv2, seeing as how the language in the GPLv3 was specifically rewritten so as to be more applicable internationally.

At the same time, the EUPL tries to take a more international view on the enforceability of an agreement that requires oversight in many jurisdictions. Licenses originating in the United States are derivatives of US copyright law, so as far as they’re concerned international considerations are non-existent.

As I mentioned above, the GPLv3 (and perhaps other FOSS licenses) have definitely taken into account issues of international law. Sure, FOSS licenses are probably influenced by the laws of their country of origin, but that doesn’t mean that their authors aren’t sensitive to the needs of their users.

While translations of existing FSF licenses are informative and referential, the US English employed is the only contract language.

Some might see this as a bonus. Who can afford the legal staff necessary to review a license in 22 different languages? That could be a HUGE burden.

Choosing the correct license could be as simple as using the GPL v3 if the software is originated in the US, and the EUPL v1.1 if the software is from Europe.

I think this was supposed to be a joke, but it highlights an important point about FOSS licenses: There are a lot of licenses and a lot of confusion over which license to use.

However, the freedoms provided by the GPL v3 may be limited to the US, and so may not be “free” enough for the global community. For example, is it truly freedom if you’re protected from intellectual property claims, but then forced to use a specific license — one not of your own choosing — to distribute your original work?

Anyone can choose any license they like to distribute their own, original, non-derivative works. I don’t think anyone is disputing that. However, if someone wants to re-use someone else’s work then they have to follow the terms of the upstream licensing.

The only “license” (and it’s not really a license) that does not restrict a downstream developer’s choice of distribution license is putting code in the public domain. People are free to put their code in the public domain, but it means that code can reused in proprietary software and improved source code does not need to be made accessible to users.

This may be one of the problems that the EU has with the GPL v3. Once a work is created under the GPL v3, all subsequent works will be licensed under it as well. In contrast, works licensed under the EUPL v1.1 can be re-licensed under any of the different licenses on their compatibility list.

The argument being made here is that the EUPL offers more choice than the GPLv3 to developers reusing existing code, so it’s better. But by that argument my ficticious “Qubit Public License” offers more choice than the EUPL, so we should use it instead. If a developer wants to give downstream recipients the ability to reuse code under 5 different licenses, then just multi-license the code under those 5 licenses and be done with it.

Perhaps the EU does not want software created under the EUPL v1.1 to be trapped within the confines of the GPL v3 later.

To put that a different way, perhaps the EU wants to be the one in control of the license definition, thus giving them power when writing subsequent versions. Perhaps the EU does not want the FSF to be the entity maintaining these Free Software licenses.

Here’s something else to consider. The GPL was originally derived from US copyright law, which creates a license bias toward issues unique to the United States. It may very well be that the FSF created their license to be specific to the US software community, but that distances it from the global community.

I already addressed this above. The GPLv3 was specifically written to be more applicable internationally. Given that there was an open call for comments on the GPLv3 drafting process, why didn’t the EU step up at that time and express their concerns?

There’s a large world of software developers and users outside of the US, and their concerns are different those of US-based developers and users. It makes sense that they want a license that addresses those differences. Many OSI-approved licenses (including but not limited to the GPL v3) do not address a global audience, and it’s a fair assumption that current efforts like the EUPL are just a hint of what is to come in global or regional licensing outside the US.

I think that the concerns of Free Software developers the world ’round are quite the same: Give me the ability to read, use, and share the software I write on whatever hardware I please. While the EUPL may be an attempt to address licensing from a multi-language, multi-country angle, I am unconvinced that it is the right approach.

By excluding the GPL v3 from compatibility with the EUPL, the European Commission is indirectly pushing out the FSF as a presence in open source licenses in Europe…we think it’s pretty significant that the GPL v3 cannot be merged with any code under the EUPL v1.1.

I agree wholeheartedly. I’m concerned that this is a misguided power play to wrest control from the GPL family of licenses.

The EUPL does not contain language on time-shifting restrictions or DCMA-specific language, and it doesn’t address issues of tivoization.

These issues are topical for Free Software hackers today — they’re currently affecting the use of Free Software on certain devices, such as Tivo’s ability to prevent people from running alternative Linux kernels (GPLv2) on their Tivo hardware by the use of hardware signing keys. The GPLv3 clearly offers greater Software Freedom than the EUPL in these respects.

The EU’s license contains none of the GPL’s explicit (and US-biased) patent language.

I didn’t write the GPLv3, but I believe that explicit patent language was used to make sure that it was as lawyer-proof as possible. Again, I believe that the GPLv3 was written with an international audience in mind , and I think it is unfair to call the license “US-biased” without backing up that claim with examples of how the language is more applicable to US law than the law of other countries.

The EUPL doesn’t enforce a time limit (like 3 years) for the required distribution of source code — it’s more open-ended. As long as the program is being distributed, the source code will be available.

I see that as a detraction of the EUPL. If I download a binary under the GPLv3 today, that means I can get the source code anytime in the next 3 years. I can even write to the distributor and ask for it via snail mail. If I download a binary under the EUPL today, the person could shut down their repository and unplug their server tonight, and I’d have no legal way to request the source code.

The EUPL appears to be simple, clear, and less restrictive than the GPL v3 license.

By being less restrictive on certain topics (e.g. Tivoization) the EUPL does less to protect the rights of developers and users of the software than the GPLv3. Certain developers of new code may see this as a benefit as it means they retain more rights over the code, however proponents of Free Software probably see this as a flaw in the EUPL.

It reads more like the Apache license than the GPL v3. It’s simply formatted, with 15 explicit sections that say what needs to be said, and it contains little apparent excess.

The EUPL does seem shorter. The GPLv3 includes nontrivial sections on how to use the license and on the meaning and importance of Free Software. RMS would say that these sections are important and are not “excess.”

With the ever-increasing popularity of open source software and the need to protect such works and their associated freedoms, many open source licenses need to evolve in order to address global — or, as in the case of the EUPL, regional — concerns.

I agree that licenses may need to update their language to retain their intent in courts across the world. Part of the work in the GPLv3 drafting process was addressing such concerns.

The EUPL seems to have adapted to this open source climate and has the potential to challenge the GPL by differentiating on a few important key points:
* Some people feel that the FSF is too political,

The EUPL retains the flavor of a dry, business-oriented Open Source license. It does not have the vigor and dedication to user freedom contained in Free Software licenses such as the GPLv3. While bureaucracies such as the EU and big businesses may like this tone, I find it lackluster and disinterested with user rights.

and regional users clearly feel that the organization is overly focused on the United States.

I’m still struggling with what this is supposed to mean — is this still the issue of language-localized versions of the license?

The EUPL stands up to what some view as prohibitive and confusing constraints of the GPL licenses. Here is a license that appears more friendly to the intention of software openness, without the many constraints (patent, tivioization, US jurisdiction, etc.) that can make licenses too localized, and thus difficult to embrace and support on a broad scale.

A FOSS license that does not contain provisions about Tivoization, patents, and the application of rights-crippling laws such as the DMCA is a license that is ignoring clear and present threats to the use and distribution of FOSS by the community. If the EU believes that these issues do not affect people outside of the US then they are deluding themselves.

As far as I know the GPLv3 is legal in all jurisdictions in the world. I don’t believe that localization problems are the chief reason why the GPLv3 is not being used more widely; I believe the issue is that people are choosing not to “embrace and support” the freedoms guaranteed by the license.

Lastly, in the EUPL there is no mention at all of money or freedom. Freedom means never having to be told you are free.

It’s an undisputed fact that the GPL has a nontrivial section talking about Software Freedom. I think it’s arguably healthy for users to be reminded about their freedoms from time to time. Here’s a short snippet of what the GPLv3 has to say about freedom:

When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for them if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs, and that you know you can do these things.

To protect your rights, we need to prevent others from denying you these rights or asking you to surrender the rights. Therefore, you have certain responsibilities if you distribute copies of the software, or if you modify it: responsibilities to respect the freedom of others.

Protecting your rights…the freedom to distribute…change…use… and, perhaps most critically of all, the fact that you know you can do these things.  Isn’t it marvelous? Free Software is all about empowering people.

Mr. Price continues, underscoring his view of the EUPL as an Open Source Software license.

The EUPL v1.1 is a legal instrument, simple and clear to interpret, with less baggage than the GPL v3, and with the intent to enforce locally while inter-operating with certain key licenses.

When he speaks of “less baggage,” he is referring to the removal of both the definition of and the protection of key Software Freedoms.

As the world increasingly sees the value of transparency in code development, it can be argued that licensing needs to be less about a “movement” and more about the proper identification of rights, restrictions, and obligations for use of third party software.

Transparency in code development is a benefit shared by both Open Source Software and Free Software. Proper identification of rights, restrictions, and obligations of accepting a software license is important when authoring any software license, FOSS or not. If you want to take the “movement” out of Free Software, I think you get Open Source Software.

[Open Source] Software use is governed by just a few things…there are only so many ways that software can be used; the Apache license distinctly describes four, and the GPL v2 does not even reference personal use.

The GPLv2 has this to say about the use of software: “5. You are not required to accept this License, since you have not signed it. That seems like it gives you implicit permission to use the software, even without accepting the license.

It seems that software licensing needs to be simplified into a “representative language” that spans the globe, is legally enforceable, and is clear.

I believe that the GPL licenses are designed to be clear and legally enforceable. The GPLv3 has been updated and lengthened in response to attacks against Free Software such as Tivoization. As far as is possible, I believe that the language of the GPL has been crafted to be applicable in legal systems throughout the world, while the unofficial translations of the GPL licenses provide easy access to the basic intention and gist of the license.

Open source licensing needs to surpass the US English language barrier and address a global audience of users. Real freedom will happen when original works can be shared, with constraints and conditions, in a way that can be understood by users and enforced by courts in all parts of the world.

I agree that better global understanding of software licenses is good, but I don’t know of an effective way to write legal text in a cross-language manner without going insane. Even the Creative Commons licenses have legal text written in English and non-legally-binding translations of the key points of the license available in several languages.

For the global future of open source software, fewer licenses are better, but one license best.

The Free Software Foundation uses at least three different licenses on a regular basis — GPLv3 for big chunks of code, GNU All-permissive license, and the GNU Verbatim Copying License. I agree that keeping the total number of FOSS licenses to a minimum makes things easy and means that FOSS developers can have a working knowledge of all of the licenses they’re using, however I don’t think that we should reduce the number of licenses to one. That’s going way too far.

The EUPL shows that by using compatible licenses, a simple message can be communicated, a broader scope of freedoms can be encouraged, and developers can be provided an “out”. It’s our guess that more OSI-approved licenses will eventually evolve to include a compatibility clause like that in the EUPL v1.1, which allows developers to take their code and get away from the current license if they’re unhappy with the way things are going.

First of all, I strongly doubt that other licenses will move to this model. As I’ve said before, it adds an unnecessary layer of complexity onto what licenses that can already stand on their own. Second, why does the EUPL want to serve as this kind of “broker” of pick-your-own FOSS license? Sure, someone can take some 3-clause-BSD-licensed code, make some modifications, and then release it under the GPL, but if the changes are small then most developers would just retain the original BSD license.

Developers: Make up your mind! Choose a license when you start development on a project and stick with it for a while. If you retain control of all of the copyrights to the code, then you can decide to relicense it under a different FOSS license later, but try to choose the right license from day one if possible.

It remains to be seen if the FSF will embrace this degree of freedom in the future.

The FSF will probably argue that the EUPL type of license is more confusing and complicated than it needs to be. I think that the GPL lineage of licenses protect the Software Freedoms of downstream recipients of the code very well right now, and I have not yet seen a clear need for licenses such as the EUPL.

Is freedom just another word? No, freedom is choice. And while licensing choice is baked into the EUPL v1.1, it’s not currently afforded to developers modifying GPL v2 and GPL v3 licensed software.

When I talk about Software Freedom I interpret it as the choice to do certain things, like play Blu-ray discs using a Free Software OS, or play Flash movies using Free Software, or examine the code that runs your browser and blog and tinker with them. As long as I am legally free to write an alternative to Adobe Flash Player or Microsoft Silverlight, and as long as I won’t get sued for writing encoders and decoders for the newest, hottest media file formats, then I retain key Software Freedoms.

The “freedom” that Mr. Price is talking about (let’s call it P-freedom) is something very different. P-freedom appears to be the ability for someone to take some existing code licensed under a license like the EUPL and “specialize” the license on it to one of a specific set “approved” licenses (e.g. the 5 licenses authorized by the EUPL). While I appreciate the idea of trying to foster compatibility between existing licenses by kind of “gluing” the EUPL on top of a set of them, I must reiterate that I think that this is the wrong approach to take.

The fact that developers and their organizations are free to choose the license(s) under which they release their code is marvelous. The EUPLv1.1 is a very interesting FOSS license, and while I personally have some concerns about it (e.g. multiple translations with legal status) and think that there are better methods of achieving some of the goals of the EUPL (e.g. multi-licensing), I look forward to reading more critiques of it and hearing from developers who have chosen it for their code. For now I think that the GPLv3 offers the best protection of my Software Freedoms, and that’s going to be my license of choice for new code.