This paper identifies an often overlooked dimension of open source licensing, that of users being able to run distributed software freely and without discrimination, and suggests additions that would further secure freedom 0 to end users - the freedom for the user to use software however they wish. While, traditionally, open source licenses have left developers free to work with their code in any way they wish, and only secured rights for end-users through code modifications, new licenses rethink this paradigm. Furthermore, modern mobile application infrastructure is more restricted by design, and has changed the landscape of modified versions of applications (simultaneously widening the digital divide), and thus a new analysis is needed on how (and which) freedoms should be secured to ensure that free software remains free.
Copyright © 2021 for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0) 1.1
Rights of ownership inherent in copyright can be granted to others using licensing. Open source (OS) software refers to licensing that allows licensees to freely distribute software code and redistribute it along with any changes. Open source is based on the idea of free software — an ideology of openness and inclusiveness, where required freedoms for using the software are secured for the public.
The freedom in free software refers to liberty, not cost, and this is often
emphasized with the name FOSS or FLOSS — Free/Libre and Open Source Software.
As with every kind of liberty in law, liberties secured through licensing come with
opposing duties to others
Thus licensing liberty is a question of who holds what rights, and how this affects others. There is no ultimate universal liberty for everyone, as each liberty imposes some duties to others or signs away other rights. Practically, there are three somewhat overlapping parties to these license agreements: authors (original licencers), contributors (those who relicense/rerelease or sublicense their derivative works) and public (end-user licensees). The rights of these three parties are weighed against each other depending on what the license contains.
As choosing what rights to sign away with licensing is both a very personal choice and depends a lot on the type of project in question, it is a good thing that there is a variety of licenses to choose from: A for-profit endeavor can open their code to the public without losing rights to make a profit with their software; an ideological programmer can release code secure in the knowledge that the fruit of his labour will be free to all future end-users and not commercialized by others.
The degree of freedom and its recipients can be freely chosen, but while it is
technically possible for everyone to write their own license, not everyone has the necessary
legal skills to do so. It also takes a substantial amount of effort and skill to read through
new licenses and understand their legal ramifications. A large enough diversity in
licenses will create confusion, as it is difficult to grasp the totality of offered possibilities
Free software ideology values the freedom of users. GNU GPL is founded on the four basic freedoms, which are an attempt to codify these values. The four basic freedoms are: • the freedom to use the software for any purpose, (Freedom 0) • the freedom to change the software to suit your needs, (Freedom 1) • the freedom to share the software with your friends and neighbors, and (Freedom 2) • the freedom to share the changes you make. (Freedom 3)
These four freedoms form the definition of free software, according to GNU Project (2001). Of particular interest to the topic of this paper is Freedom 0, which can be seen to be at the foundation of free software. 1.4
These two main considerations, as well as author attribution, the inclusion
informational texts (e.g. copyright and license notices) and DRM (Digital Rights Management)
or patent usage rights in the license, are commonly used in mapping the different
dimensions or aspects of rights granted by OS licenses
One dimension not yet explicitly covered by previous examinations of the wide
array of rights granted to licensees is the right to use the software as they wish (i.e.
freely) without the need to alter source-code. Even though the importance of Freedom
0 has been discussed
While the four freedoms should secure the public the right to use software as they
wish, there is no clear consensus about whether these freedoms are secured merely
by sharing the source-code. An effective example of this can be seen in cases where
distributed programs include hard-coded (embedded in the source code) functions that
explicitly block users based on their affilitation
As censorship on the code level (e.g. domain blocking in source code) is a new
phenomenon, it is difficult to say how prevalent these practices will become. Comments on
implementations of domain blocking reveal that there is confusion on whether such an
implementation of rights management is in compliance with OS licenses, free software
philosophy or both/neither
The technological industry has been moving towards permissive licenses, that allow
for parts of the source code to be republished under closed licenses.
The digital divide is defined as inequality between those who have access and means to
use technology and those who do not. While these inequalities are not absolute, in that
they can be bridged, nor the division clear-cut, it is a phenomenon that clearly exists
and should be addressed
The effects of digital divide become more pronounced on second-and third-level
digital divide, where the use and outcomes of differing skill-sets and possibilities are
weighed
If code is introduced in an open source project with the goal of blocking certain groups from making use of said software, it is not necessarily enough that the code can be copied and changed to remove the changes. If it is indeed an important goal of (some of) the developers to consciously block such use, it is possible that the blocking code will be obfuscated and made difficult to remove, or that other types of difficult-toremove DRM will be introduced. In the current era of centralized platforms, it is even possible to release open-source software that depends on a centralized server network that doesn’t allow forks on it. Both these methods would further exasperate the digital divide, and make securing freedom 0 difficult or impossible even for experienced programmers.
Much of open source ideology hinges on the fact that development is done according to good programming practices and technology is not influenced by outside forces such as politics. This is perhaps not quite as true as it was a a couple of decades ago. In a more uncertain environment, it is completely conceivable that a developer might want the option to protect the freedom of the public to use their software freely, and licensing is the correct tool for guaranteeing such freedoms.
Digital divide makes it imperative for IT specialists to pay closer attention to public liberties, as the majority of users do not have the requisite skillset to modify programming code, rebuild it to new binaries and distribute their fork of the software. This majority is silent due to not being represented on the discussion forums of code development platforms. Moreover, the walled gardens of modern mobile platforms make it even more complicated to distribute your own version of an applications. 2.3
In past years, there has been more interest in open source licenses based on ethical
values
Notably, such licenses cannot be considered open source, as restricting the use of software by licensing is against the open-source ideology and Open Source Initiatives rules for open source licenses (2019), which state that open-source licenses must comply with ”No Discrimination Against Persons or Groups” and ”No Discrimination Against Fields of Endeavor”. Such license is not open. This new type of licensing nevertheless serves as an example of more ethical consideration being put in licensing, as well as the cultural drive for more restrictive measures in licensing — more restrictive for the public, securing more rights for the licensers (i.e. limiting what others do). 2.4
GNU GPL is founded on the four basic freedoms listed in the introduction. The license has been formulated to secure these freedoms through the sharing of source code. 1https://ethicalsource.dev/licenses/ 2https://github.com/EthicalSource/ethicalsource.dev Distribution agreements have been consciously left outside this license, presumably because user freedom is secured through the availability of the code, even in the case of closed or paid binary distribution.
GPLv2 was updated to GPLv3 in 2007 to include new freedoms that were previously not secured for end-users, and includes protections against ”Tivoization”, that prevent users from freely using the software by using hardware-based limits to code modification; and laws prohibiting free software, that prevent using DRM laws to prevent software distribution and discriminatory patent deals. In other words, GPL-licensed software can include DRM as long as its removal is not made impossible with special hardware or laws.
GPLv3
GPL and open-source licenses currently do not include protections for users to run the distributed binaries without impediments. It must be understood that this is not necessarily a bad thing. Not securing these liberties ensures more freedom to the licensers to do as they wish (e.g. implement DRM), and this can be a good thing.
I argue that the following changes are in the spirit of GNU’s four basic freedoms, but should not necessarily be at the core of all GPL licenses. In a similar fashion as LGPL has a slightly different emphasis on the open source ideals compared to GPLv3, these proposed additions would constitute another take on how the license weighs liberties between authors, contributors and the public, granting some freedom away from authors and contributors in favor of the public. I’m not arguing that this should happen instead of what e.g. GPLv3 does — I’m only arguing it should be an explicitly stated option. 3
e) The work must not include any methods or additional protections for restricting the use of the work by particular persons, groups or fields of endeavor.
This addition would disallow the use of DRM (digital rights management) that targets persons or groups. Censorship systems such as domain blocking and networkbased DRM would also fall under this category.
1. Any object code conveyed must not include extra blocking methods aimed at discriminating agains persons, groups or fields of endeavor specified in Chapter 5.
The object of this addition would be to secure Freedom 0 when using conveyed non-source forms (i.e. binaries). This would (most importantly) include mobile apps downloaded from official app stores. This addition might be redundant with the previous addition blocking the use of said blocking methods in the source code. It is included here to facilitate discussion, as the main part of the addressed problem is caused by distributed binaries, especially in the walled-gardens of mobile app stores. Perhaps it would be best to only target the source code and not widen the license to touch this part of distribution. Or perhaps it would be better to keep total freedom for developers when it comes to the source code, and only include this addition or allow distribution as long as there is also a version available without restrictions. 4 4.1
The proposals would sign away more rights from contributors, should they wish to restrict the right of usage by the public. It could be argued that such licensing is no longer open source, or that it is less open, because use of the source code is restricted in some manner (especially in the first proposal). But the restrictions are comparable to the duties of needing to provide the source-code along with distributed binaries or not being allowed to relicense the work and choose their own license — the contributors are given duties that correspond to ensuring the freedoms of the public.
Perhaps such a license would not be open source, but rather part of the ethical licenses family. Nevertheless, such a distinction is somewhat semantic in nature. The cure issue is whether more weight is put on the rights of the developers (both original licencers and new licensees) or the public (end-users, whether developers or not). 4.2
GNU GPL is based on the four basic freedoms of users, and as articulated by
The digital divide mentioned in chapter 2.2 makes it imperative that user freedom be guaranteed, moreso when development and distribution options get more difficult. Mobile application distributions systems are perhaps a special case when it comes to software distribution, as they mostly rely on ”walled gardens”, a centralized distribution system in the form of app stores3. Installing your own version of an application requires at the very least a development environment dedicated to the chosen platform and a high level of technical know-how; possibly also payment to the required distribution platform, if one wants to share the application or install it without resorting to developer options. In such a case, Freedom 0, the freedom to use the software for any purpose, is not secured for all users without additions such as the ones proposed.
In comparison, the proposed changes would be clearly in conflict with the values of some other ethics-based licenses that explicitly restrict the use of software, e.g. the 3While Android allows the installation of apk files outside Play Store, the option is off by default and can be difficult to do without technical skills. NoHarm license4 that forbids use related to things such as slavery, gambling, tobacco, hate speech or discrimination. This is not a problem, as a wider selection of licenses ensures that people can choose how their software is shared and used. The proposed additions would not affect those who choose not to use a license that includes them.
Arguably, the proposed changes would go well together with a new version of GNU GPL. Just as GPLv2 was updated to GPLv3 to protect user freedom from tivoization, laws prohibiting free software and discriminatory patent deals, these new additions would further protect users from new discriminatory restriction methods. As such, they should not be a replacement license but a complementary one or a completely new license. 4.3
GNU GPL explicitly allows selling copies of free software5. This provides a good example of how ’free’ in free software refers to liberty (of code) and not cost (of distribution). The proposed changes do limit freedom to work with the code, which could bear further examination to ensure that this does not create unforeseen problems.
Perhaps the closest thing to the proposed changes is the use of DRM, which is explicitly allowed under GPLv3 licensing6. But even DRM is disallowed when protected by tivoization or law, and a case could be made that similar protection is provided by modern mobile app distribution platforms, which should be included in similar fashion.
The GPL FAQ gives an example of voting machine users not being allowed to change the code running on voting machines. Such abilities to control running software are not required by GPLv3 licensing.
The FAQ states that such liberty is not required, as the voter doesn’t get posession of the object code, even temporarily, and so the voter also doesn’t get possession of the binary software. The FAQ also states that ”voting is a very special case”, so the applicability of these thoughts on software in general and distribution of mobile applications specifically is arguable. 4.4
The proposed changes should be scrutinized by the open-source community and, specifically, legal experts, to make sure that their scope is as-intended. Without sufficient legal expertise, it is possible that there are large unintended consequences, and these should be found out and eliminated before releasing a full license.
4https://github.com/raisely/NoHarm/ 5https://www.gnu.org/philosophy/selling.html 6https://www.gnu.org/licenses/gpl-faq.htmlDRMProhibited The proposed changes do limit the freedoms of contributors significantly. It might be possible to limit these freedoms less, while still protecting the public freedom of use, by e.g. requiring the conveying of non-source forms without the listed protection methods. This would allow contributors to release code and binaries with discriminating restriction methods as long as they also release versions without restricting methods.
Arguably, a more clear-cut division of licenses makes more sense: if you wish to restrict the use of your software by code, choose a license that clearly allows such restrictions; if you wish for your program to be open and freely usable, choose a license that disallows restricting methods. 5
This paper identified user freedom to run a software as they wish as a dimension of open source licensing that has so far been left undefined in commonly used OS licenses. This new dimension should perhaps be considered in scientific literature along with the more common dimensions of copyleft/permissiveness and the securing of patent rights.
In practice, while there are already many licenses to choose from, and there is
nothing inherently wrong in the current formulation of GPLv3, there is some confusion
on whether blocking particular persons or groups is compliant with it
Functionality overriding Freedom 0 should be clearly allowed or clearly disallowed in OS licenses, in order for licensers to make informed choices on the matter. If the question is made explicit in the form of new licenses with the additions proposed in this paper, licensees could choose between licenses based on their own wishes, and contributors using freedom 0 securing licenses could be secure in the knowledge that the project they are working on will stay free.