It Matters Who Owns Your Copylefted Copyrights
by on June 30, 2021
Throughout the history of Free and Open Source software (FOSS), copyright assignment has simultaneously been controversial and accepted as the norm in our FOSS communities. This paradox, I believe, stems entirely from some key misunderstandings that perpetuate. This issue requires urgent discussion, as two of the most important FOSS projects in history (GCC and glibc) are right now considering substantial and swift changes to long-standing copyright policies that date back to the 1980s. This event, and other recent events over the last few years in the area of GPL compliance and corporate FOSS adoption, point to long-term problems for projects. This essay works through these nuances, and will hopefully assist FOSS contributors as they make difficult decisions about copyright ownership for their projects. At the end, I provide a summary list of issues to consider when creating copyright ownership policies for FOSS.
The Default Is: You Give It Away To Your Boss
I was at one time bemused, but now am mostly aghast, at the true paradox of common wisdom about copyright assignment for FOSS projects. I don't believe anyone has done a statistically valid study, but anecdotally it's rare to find a FOSS contributor who enjoys assigning copyright to another entity. FOSS contributors who do assign copyright to a non-profit charity that collects copyrights (such as the FSF or Conservancy) often complain about the complexity and annoyance of the paperwork to get started as a contributor. Even more commonly, contributors express disgust or annoyance at the pressure from the project to give away something that should be rightfully theirs.
I am sympathetic on both points, and have worked over my career to design, implement, and improve copyright assignment processes for projects — in hopes to address that first complaint. However, the second concern — the preference of FOSS contributors to keep their own copyrights, is an ironic complaint. Here's why:
Almost no contributors to larger FOSS projects hold their own copyrights. If they have not gone through a process to assign them to a charity, the most likely scenario is that their employers own those copyrights. My colleagues at Conservancy have been working on the Contract Patch project — which seeks to educate FOSS contributors about their inherent right to employment agreement negotiation, and seek more favorable terms in those employment contracts. However, over the years of our Contract Patch work, we still find that only dozens (among the thousands of FOSS contributors) have insisted that their company allow them to keep their own copyrights. If you work for a company, check your employment agreement. I'll bet that your employer owns your copyrights for everything you do at work — including your contributions to FOSS — unless there's a separate agreement that gives those copyrights to a charity like Conservancy or FSF.
As a result, in debates about copyright ownership, discussions of what policy contributors want regarding the fruits of their labor is sadly moot. Without a clear, organized mitigation strategy to assure that FOSS contributors keep their own copyrights, a project (such as GCC or glibc) that switches from a standing “(nearly) all copyrights assigned to a charity” model to a plain Developer Certificate of Origin (DCO) or naked inbound=outbound contributor arrangement will, after a period of years, mostly likely have copyrights that are primarily held by the employers of the most prolific contributors, rather than by the contributors themselves.
The Faustian bargain that causes this default is the “work-for-hire” doctrine in the USA (and similar arrangements that exist around the world). When you take a job, in most places in the world, by default, your employer owns and/or effectively controls all your copyrights. They argue that they're paying you handsomely for this and as such they have every right to own it all. Perhaps that argument has merit with regard to software that will ultimately be proprietarized, but for copylefted software, the value proposition is different and history shows that the arrangement leads to surprising results.
Copyleft Is Weakened When Most Copyright Holders Oppose Enforcement
Copyleft licenses are not magic pixie dust. Copyleft is not a legislatively-mandated regulation (e.g., pollution regulations) — which are enforced by government staff. Ultimately, an entity — most commonly the copyright holders themselves — must proactively enforce GPL. This entity can be an organization, an individual, a group of individuals, a group of organizations, or a mix of all of those. Someone must enforce the rules; so-called “spontaneous compliance” is a myth promulgated by those who oppose copyleft.
Yet that myth's apparent unexamined truthiness has taken hold; many FOSS contributors reasonably think that it does not matter who owns the copyright of their copylefted works. If it's GPL'd, they surmise, surely someone out there will enforce the GPL when it's violated. Or, perhaps the contributors think that GPL violations on their particular codebase are rare. Either way, most contributors avoid the effort required to figure out who owns their copyrights and ponder whether that entity will uphold copyleft in a reasonable way. Nevertheless, I really urge FOSS contributors to think through this issue with care and healthy skepticism, as they may be surprised at the outcome. What lurks in the backchannels may well surprise you:
Once upon a time, copyleft enforcement was not considered controversial by many otherwise-pro-FOSS companies. In recent years, companies that prefer non-copylefted FOSS have succeeded in making even the most mundane GPL enforcement appear as controversial and industry-destabilizing. Many developers who contribute to GCC and glibc — even ones that work for historically FOSS-friendly companies — will be surprised to learn that their employers (or at least their legal departments) are opposed to any GPL enforcement. Conservancy, as the main organization actively doing GPL enforcement, hears this kind of pressure regularly. Our colleagues at FSF historically told us that they hear the same. It's rarely done publicly — but, when public, criticisms are delivered by proxies in a subtle, “you don't really need copyleft to be enforced anyway, do you?” manner. The public arguments are usually couched in a careful “dog whistle” style. While the message comes through loud and clear to executives, lawyers, companies and their trade associations, it's easily missed by contributors — who understandably have a much more important task to do: writing great FOSS code!
We expect that you may be surprised by this, but we can tell you that: in private, companies that contribute to key copylefted projects are not so subtle. They have communicated to those of us in the GPL enforcement community on several occasions — a message delivered by many executives over a period of many years — that they would prefer both Conservancy and FSF to curtail, or outright end, GPL enforcement. Their view, as communicated to us, is that GPL enforcement causes customers to think copyleft is problematic. As one executive once said to me: “We consider it a failure if any customer ever even asks us if they have to worry about GPL compliance”. These companies usually say: “well, we will always comply with the license, but we'd prefer if anyone who is our customer (or potential customer) simply is just left alone if they violate”. Once, an executive actually claimed to me that his company, a huge multinational software company, would: “cease to build any products on Linux and other GPL'd software at all if Conservancy and FSF don't cease their enforcement”. That statement was fortunately bluster and bluff; the company in question is still heavily invested in copylefted software, including Linux, GCC, and glibc, but the sentiment is a common one that we hear privately often. And, that company has since funded work to discredit GPL enforcement. Conservancy faces constant pressure and threats regarding its GPL enforcement efforts, and we're aware that FSF has faced to similar pressure.
In parallel, these companies have also steadily worked to hold more of the copyrights in key copylefted works — primarily by hiring the key developers that contribute to key copylefted projects. While I oppose this outcome, it's undeniably a rational approach: if companies hold the copyrights, they can decide when, how, where and most important if copyleft licenses are ever enforced. This process is well underway for copylefted works that don't have any policies about copyright ownership. For example, the “Who Contributes to Linux?” reports by LWN show starkly over recent years: fewer contributions coming copyrighted by individuals, and more contributions copyrighted by companies. While corporate involvement is always welcome in FOSS projects, we (as a FOSS community) have not responded with the necessary care to question why the companies, rather than the individual contributors, should own our FOSS. Only very few developers have even questioned this issue; we thank them for their efforts, but such questions are simply not raised enough.
As such, the most important consideration in ending copyright assignment to a charity is to consider what the employers of most of the developers will do with those copyrights, and whether they will enforce the copyleft in the best interest of the community. Even worse, might they eventually sell those copyrights to someone who will use enforcement in a manner that a charity never would — for their own avarice rather than the good of the community?
While the remainder of this essay addresses at length other secondary considerations and nuances regarding copyright ownership in copyleft projects, this point is the absolutely most important one:
Most for-profit copyright holders prefer copyleft to function much like non-copylefted FOSS. Namely, “it's certainly nice when folks voluntarily release their changes, improvements and ‘scripts used to control compilation and installation’, but if they don't, their failure to do so should be begrudgingly tolerated, and compliance should never be compelled for those who refuse”.
Violations Are More Common Than You Think
During this recent upheaval in the GCC community, I spoke at length with a GCC developer who has contributed to the codebase regularly for decades. After I'd explained some of issues above, the developer stated: “Well, is any of this really an issue for GCC? When was the last time there was a GPL violation on GCC?”. I glibly answered: “Well, I haven't heard about one for about a week, but I'll surely hear about another one soon.” The developer was flummoxed to learn that part-and-parcel to nearly every embedded Linux GPL violation (which have been, for years, innumerable) is a companion violation on GCC. Specifically, most system-on-chip (SoC) vendors not only have a stock Linux build given to the OEMs, but also include a toolchain. More often than not, a GPL violator who has what we call an “incomplete Complete, Corresponding Source (CCS)” violation on Linux will also have a “no-source-or-offer” or “incomplete CCS” violation on GCC. The usual reason is that part of the SDK given to OEMs must include an appropriate binary toolchain (sometimes with vendor-specific patches, and almost always with a complex configuration that's difficult to reverse engineer) so that the SoC integrators can compile other software for the system. This scenario is so common that we in the GPL enforcement community coined the shorthand term “toolchain violation” to refer to the scenario where the GCC violation “tags along” with a Linux violation. While glibc violations of this nature are admittedly less common than GCC, glibc violations do occur relatively often in the same manner.
Upstream developers are mostly unaware of how bad the compliance problem
is regarding their code for one simple reason: the folks impacted most by
GPL violations are downstream users, not upstream developers. Furthermore,
correction of compliance problems usually produces code releases that
“work” for the given device on which the original violation
occurred, but rarely is that
The Power of Collective Action
Regarding my views on FOSS copyright assignment, I often quote the famous gaffe of (former USA presidential candidate and senator) John Kerry, “I voted for it before I was against it” when I'm talking about copyright assignment. But I don't quote it purely for self-deprecation; I never felt the press was all that fair to Kerry, because I think policy makers should be willing to change their positions over time as new information comes to light, or when policies that we thought were beneficial in theory prove problematic in practice. I once thought universal copyright assignment for a copylefted work to a single charity was an absolutely necessity. Then, Conservancy's successful and continuing work with various BusyBox copyright holders, and our collaboration with Christoph Hellwig, showed me that there was huge value in individuals having copyrights and making their own choices about enforcing copyleft. Still later, I realized that individuals like Erik Andresen and Christoph Hellwig are quite rare, as most FOSS contributors — even if they've kept their own copyrights — ultimately have to take on such political and career risk to enforce copyleft that they must often chose not to because they could easily get blacklisted from major employers for doing so.
The central thread here is collective action by principled people who will use copyleft primarily as a tool for rights of users and for the improvement of copylefted projects. Ultimately, copyleft functions best when leaders of a project have the agency, wherewithal, commitment, and collective consensus to either ensure copyleft functions to protect their users' rights, or enable another entity to do that job for them in a principled way that serves the public good (usually in contrast to the interests of for-profit industry). (Note that even if that interest is “vendor-neutral”, as that means the interests served are the common business interests of the vendors, but not their customers.)
These issues are complex. Every policy change, or lack of policy change, will have many unintended consequences. An apparent “simple move” away from mandated assignment to a centralized organization could well be a decision to ultimately move copyright holding to for-profit corporations and their trade associations — all of whom are unlikely to stand up for the GPL and user rights. Beyond all else, I urge caution and slow deliberation before any policy changes about copyright control. I ask FOSS contributors to do the substantial and necessary homework of not only reading and considering the points of this essay, but those points raised by all parties. As you do, keep in mind that every party, including me, has a policy goal in mind for copyleft. I and Conservancy are very transparent that our policy goal is to see copyleft licenses enforced regularly on behalf of end users who buy products on the open market that contain copylefted code. We believe, as a policy matter, that copylefted copyrights are best held by entities that have a bona-fide track record of serving the public good, act in a principled and ethical manner in doing so — the most important principles being to never prioritize financial gain over users' rights — and who won't back down and give up when faced with the inevitable anti-copyleft backlash. Others, probably including your employer and the trade associations to which they pay membership dues, probably disagree with this position, but you should ask questions and listen carefully to see if you're getting a straight answer. After you've heard everyone's position, decide for yourself who deserves to have your copyrights: your employer, a charity, or yourself. Once you've decided, you'll have hard work ahead to change the default (which will likely be your employer once projects drop their copyright assignment mandate).
A Word on FSF's GPL Enforcement
The elephant in the room that I've not yet mentioned is the current status of the FSF as an organization. There is no question, given the recent mass resignation of their management, and other rapid leadership change surprises, that the FSF faces serious challenges in the next few years. I personally was previously affiliated with the FSF. That affiliation ended in 2019, so I have no real information about the internal status of the FSF since then. What I do know is that FSF currently lacks sufficient capacity to follow up on any GPL violations on GCC and glibc that we've reported for years. That said, if I have to chose between the strict dichotomy of: (a) copyrights held by the FSF (which has the possibility of recovery in the future and restoration to an organization that actively enforces the GPL) vs. (b) copyrights held by companies — many of whom are known to oppose enforcement of the GPL, I would still pick the FSF. Remember that copyright does not expire for a very, very long time. GCC and glibc are both codebases that prove the half-life of well-written software is very long, and that software stays relevant and useful for much longer than any of us usually admit. We have to think long term about the copyright ownership of copylefted works. While there are serious short-term and medium-term problems at the FSF, we have to consider carefully who is the best long-term steward of copyrights: companies or charities. Most importantly, changing 30 years of careful planning about the copyright inventory of important codebases is certainly not a decision to be finalized in a matter of just a few weeks or even months. Thus, most of all, I ask FOSS contributors to take care and ample time in their deliberations on such matters.
A Summary of Recommendations and Considerations
As promised at the start of this essay, here is a laundry list of recommendations and considerations that any copylefted project should consider regarding how to structure its copyright ownership rules. This list is provided as a quick reference only, and readers should keep in mind the complexity of the topic before jumping to a conclusion about the right policies. Not every issue below was addressed in detail in the essay above, but if there is interest from the community, I will publish further essays on other topics.
- Without your active work to avoid it (such as by modifying your contract or demanding assignment to another entity), for-profit employers will control your copyrights. You typically have no say into how or whether the license of your project is enforced if your employer holds your copyrights.
- Modern copyright assignments to charities such as the FSF or Conservancy usually take two forms: either your employer has a blanket copyright assignment, or you must individually assign copyright. In the latter case, there are typically two components: a disclaimer from copyright from your employer, and an assignment from you. In either case, you should speak at length with your employer's legal department to figure out what form, if any, is in place, and what will happen to those systems if your project changes its copyright aggregation policies.
- FOSS contributors have safety in numbers. A strong consensus that your project wants to see copyleft enforced for your users can create leverage that mitigates risk of “blacklisting” of those who chose to enforce copyleft licenses. When policies like copyright assignment to a charity or copyright ownership only by individuals are set project-wide, companies historically have simply accepted it. (The Samba project is an excellent example of a project where the developers have control.) By contrast, each contributor faces a steep climb in negotiating their ownership of the copyrights in the absence of such policies.
- While assignment to a charity can be annoying and feel problematic, this is often the most expedient way to assure that the copyleft license of your project is upheld. Conservancy, for example, will accept one-off assignment from FOSS contributors to strategically important FOSS projects. Even if you (collectively) chose to not mandate assignment for your project, it's valuable to assist and encourage your contributors to either develop together a collective individual copyright ownership plan, or assign to different charities, or recommend a mixture of the two.
- As an upstream developer, you're likely going to be unaware of most copyleft violations on your code. If possible, work with an entity that has the time and resources to investigate violations for you, and empower that entity to act on your behalf if possible.
- Pay attention to the details of Contributor Licensing Agreements (CLAs). Often, these require that you give up almost as many rights as a copyright assignment would require you to give up anyway. Avoid asking your contributors to agree to a CLA.
- Developer Certificate of Origins (DCOs) and other inbound=outbound contribution regimes are very good, useful and recommended. However ultimately such systems are entirely orthogonal to who holds the copyright. DCOs and contribution terms are general statements that attest to your right and permission to make contributions under the project's license, but are, by design, silent on the details. A DCO is absolutely compatible with a project that separately requires copyright assignment, or with a project that has other recommendations about how developers should handle copyright ownership.
- Seek and consider advice from all sources. Everyone with an opinion on these topics has a policy agenda. Ask what their policy agenda is when they advise you. Ask them their view on copyleft enforcement. Ask them if they've ever investigated copyleft violations on your software, and if so what they found and what their opinion is. If you're considering assigning your copyrights to someone, most notably your employer (who, again, is likely to by default receive your copyrights) be sure you've fully understood their policy and views on these matters. Insist that they put their policy and views in writing for you for future reference. Ask for strong commitments, and be skeptical if you cannot receive them.
Please email any comments on this entry to info@sfconservancy.org.
