The fog of copyleft
First Monday

The fog of copyleft by Aaron Krowne and Raymond Puzio

In this paper we question the notion of using copyleft, in its current form, as the principle contractual framework of free culture. We argue that there are many flaws in the copyleft concept and its execution which are beginning to immerse free culture activity in a “fog” of risk and uncertainty. The main feature of copyleft which is causing problems is its “license lock” aspect, while the main problem with the execution of copyleft is the exercise of too much control by license authors. We give examples from our own experiences with PlanetMath, as well as the free culture community in general, to illustrate. Finally, we provide a wide array of solutions to the problem, ranging from immediately usable legalistic ones to long–term societal transformations.


What is copyleft, and what is its purpose?
Entering the fog
Copyleft’s own fog
“Chilling effects”
License, contract and copyleft
Lifting the fog




“Copyleft”–style licenses have been invented and refined over the last decade and a half, their genesis inspired to carve out from the space of intangible content what we now understand to be a critical “bubble” of free culture. This movement has been in response to the gradual compression of this bubble (which has always existed to some extent) to a veritable dot, as copyright law has become dramatically less liberal. This change in the legal and cultural environment has been exposed, examined, and challenged by others, with Lawrence Lessig and Richard Stallman figuring centrally.

But while the world is just beginning to warm up to the notion of copyleft as the scaffolding of free culture and a bulwark against its elimination, those of us at collaborative projects like PlanetMath (Krowne and Egge, 2001) and Wikipedia (Wales and Sanger, 2001) have discovered that the ecosystem of copyleft licenses is an infirm one. Because of the form copyleft–style licenses take today, the characteristics of the broader free culture community, and copyright law, a great amount of the promise of these “free” licenses has not and cannot be delivered.

In this paper, we will first explicate copyleft, discussing its nature and purpose. Next we will discuss how copyleft, in its current manifestations, falls short of what is needed to best foster and support free culture, and outline the damage it is currently doing. We will draw connections when available to our experiences with PlanetMath and connected projects (like Wikipedia), as well as to other examples from the free culture landscape. We will then discuss the logical nature of the problems and their origin. Next, we will put forth some suggestions for how the situation can be improved, allowing free culture to flourish as it should. We will discuss how the problem promises to grow even more severe in the future, in a number of ways, if it is not solved soon.



What is copyleft, and what is its purpose?

Copyleft is a contractual edifice built on top of copyright law, much in the same way that any traditional copyright license between an author and their publisher would be. It is merely the stipulations of this contract which make copyleft distinct and worthy of a special name. Thus, any folk impression that copyleft is somehow diametrically opposed to or outside of copyright law is incorrect. This realization is critical in order to fully understand copyleft and its drawbacks (as well as its advantages).

Copyleft may be defined as a license that (1) gives the ability to derive subsequent works from an original work without the permission of the copyright holder; and, (2) grants this ability also for derived works via a clause which requires that these works also be licensed under the original copyleft license. Thus, copyleft licenses recursively ensure a kind of “freeness” (and for computer code, necessarily openness as well).

Any folk impression that copyleft is somehow diametrically opposed to or outside of copyright law is incorrect.

Because these clauses only form the functional backbone of an infinitude of specific, situationally–nuanced licenses, we use the term “copyleft” to refer to the broad family of licenses having the above two essential features.

The purpose of copyleft is to provide a tool for content creators to achieve one or more of the following social objectives: (a) protect their own rights to the work while still widely disseminating it; (b) protect against a situation whereby access to the work is unintentionally restricted, beyond what the author deems necessary to secure recompense; (c) insure that they will not be vulnerable to ruinous (and potentially unpredictable) legal actions; and, (d) create a milieux of works which are free to circulate and build openly (free culture).

Others have shown elsewhere (Berry, 2001; Lessig, 2004; Vaidhyanathan, 2001) how all of these objectives have been threatened by the extension of copyright law to its present form, so we will not rehash them. Here instead we will deal with the subject of how copyleft still has a long way to go in meeting all of the above needs, especially (c) and (d).

While copyleft has succeeded in making modern free culture possible, unfortunately this success is becoming limited by its drawbacks. As we shall argue later, these drawbacks are not unfortunate accidents, but arise from copyleft’s basic premises. We believe this calls into question the whole enterprise of copyleft, and will suggest that advocates of free culture would do well to lessen their reliance on copyleft as it stands — either dramatically redefining it, or instead employing other means to create an environment in which free culture can fully flourish.



Entering the fog

Because copyleft is founded upon copyright, it is not surprising that it would inherit some of the problems of copyright. In this section, we shall consider some of the unfortunate consequences of the complexity of copyright law by briefly examining its purpose and history, then placing these in the contemporary context.

When copyright was primarily the concern of publishers, complex, sometimes counterintuitive, regulations and lengthy licenses with hairsplitting definitions, exceptions, and disclaimers served a useful purpose; they helped prevent misunderstandings by clarifying in detail the rights and obligations of those involved, and made it possible to adapt to nuanced circumstances. There was little or no danger that this complexity could lead to confusion because publishers could afford to hire experts who had experience with these subtleties and their practical implications.

However, with the advent of desktop publishing and the Internet, the situation has changed dramatically. Now authors found themselves in the position of being their own publishers online and, concomitantly, were required to comply with copyright law. To the average person who has neither training nor experience with law, lengthy, detailed contracts and complex regulation have the effect, not of clarifying rights and obligations, but of confusing the reader.

This confusion can have devastating consequences. Author–publishers are likely to make contractual or procedural mistakes because of this complexity. Because of the high price of making a mistake (damages and court costs), people become reluctant to take on the concomitant risk, and thus err on the side of caution. Even when it may be theoretically possible to do something within the bounds of the law, these people may not be able to because they do not have the necessary legal know–how.

Further, whilst the budget of a legal department may represent a small part of a publisher’s total budget, for most individuals or collaborative commons–based ventures which publish on the Internet, hiring a copyright lawyer on a regular basis would not be economically feasible, and it would undermine the basic economic benefits of online publishing.

As if this were not bad enough, there is the issue of how to interpret copyright law in a digital context. Since copyright law was originally designed with print media in mind, it is not always clear how to apply it to digital media, especially when one encounters situations which have no real analog in print. Given that experts continue to have legitimate disagreement, it is inevitable that non–experts will only be confused further [1].

Topping all of this off, there is the rise of commons–based peer production (CBPP) projects, such as Linux, Wikipedia, or our own PlanetMath. These productive, Internet–based efforts are characterized not only by their digital dissemination, but by their massively distributed productive nature. That is, not only is dissemination newly scaled up (digital copying), but authoring (digital deriving) is as well. This ability to economically build upon digital artifacts has been enabled by a new class of CBPP coordination systems built atop the Internet. Some examples are code management systems (e.g. CVS, SVN), wikis (e.g. mediawiki), or specialized systems such as Noosphere (the software behind PlanetMath, PlanetPhysics, and our other planned “knowledge planets”).

CBPP efforts result in a massive, continuous re–deriving of digital works, within or without some “out–of–band” socio–political organizing shell. Further, the set of author/contributors tends to be gigantic, and the portion of this set which is actively participating is distinct from the whole set. This dynamic was not at all anticipated by copyright law, which is based on a simplistic model of publication once at a localized, cleanly demarcated instant, brought about by some single author–entity (which could be a relatively small set of co–authors). While copyleft seems to improve on this by anticipating continual re–derivation of the core work, it seems to poorly model a shifting and diverse contributor–set.

When copyright interacts with this landscape of digital production in the commons, we call the resulting situation the “fog of copyright.” While we are being metaphorical with this label, we think a fog serves as an apt metaphor. A fog is something which prevents you from seeing what’s next “down the road” — even things which are quite near. The fog retards progress (or, equivalently, upholds the status quo) and can cause accidents and disasters. It can cause trips to be delayed, postponed, or cancelled altogether. It causes anxiety, disorientation, and makes the effected feel “lost.” And finally, all of these things increase the “cost” of a journey, in the most comprehensive sense of the word.

When copyright interacts with this landscape of digital production in the commons, we call the resulting situation the “fog of copyright.”

Thus, when it comes to copyright, most people in these new situations feel that they are enveloped in a cloud of uncertainty and confusion. The alternative seems to be ignorance of the underlying problems, which is often (in our observation) intentionally chosen. Unfortunately, this head–in–the–sand approach simply opens the door to greater consequences down the road. In the remainder of the paper, we will focus on how copyleft interacts with this general “fog,” creating a kind of fog of its own.



Copyleft’s own fog

The problem of copyright “fog” was explicitly or intuitively known to those who drafted the current copyleft licenses (or those who simply supported free culture in general). Copyleft was invented both to create new capabilities and to lift this fog. And in much of this it succeeded. But, because copyleft operates within the framework of copyright law, not only did it fail to eliminate all of the fog, it actually created its own “fog of copyleft,” with some distinctly new properties.

As discussed above, copyleft uses copyright law in a peculiar way — it requires the user to waive rights and perform certain actions in order to assure that the user cannot restrict the freedoms of future users. This is rather counter–intuitive, and someone who has not studied a copyleft license carefully is apt to confuse the end of ensuring freedom with the means (which is restricting freedom), and thereby misunderstand the nature of one’s obligations under the license.

Far from being a mild annoyance, the fog of copyleft drastically limits the potential of the commons–based productive activity that is just now reaching wide impact in society. Further, some of this fog is even worse than the original copyright fog. Currently, the illusion that there is no fog at all with copyleft is the primary threat, which we hope to dispense with in this paper.

The problems are not just theoretical; participants in the community of free culture are encountering the fog of copyleft — particularly those who are part of commons–based peer-production (CBPP) projects, such as Linux, Wikipedia, or our own PlanetMath. In the following sections, we will describe the threatening interactions and give examples for their occurrence in the free culture world.

License lock and problems with copyleft

Copyleft sets up de facto “law” through contract, which of necessity requires a high level of complexity. This is certainly part of the fog — and initiatives like Creative Commons have sought to remedy it by providing licenses in vernacular as well as the underlying “legalese” — an effort which we applaud.

However, a greater amount of fog is produced by a phenomenon we call “license lock.” This is the clause in current copyleft licenses that dictates all derived works adopt the same license (see Figure 1). This characteristic of free licenses has been considered necessary for them to achieve their end. As we will show later, this is not a correct assumption [2].

License lock has the effect of setting up incompatible “colleges” of content (and content creators) who cannot share material and build on each other’s works. This is thickest part of the fog of copyleft. Our contention is that these colleges unnecessarily isolate individuals, communities, and creative works that would otherwise come together to become greater than the sum of their parts, and further, the fog surrounding the existence of these many colleges is a disincentive to participate in free culture and retards its success.

The situation is exacerbated by the socio–technological changes which we are experiencing in wider culture today, foremost of which are that technology now makes it almost as easy to adapt as to copy, and that creation–in–collaboration has also been made very simple and commonplace. Suddenly, the treatment of derived works is of as practical import as the issue of copying was before computers.


Figure 1 The copyleft situation, and license lock. Derived works can only be licensed under the same license as the original work. This is how copyleft is implemented today.
Figure 1: The copyleft situation, and license lock.
Derived works can only be licensed under the same license as the original work.
This is how copyleft is implemented today.


These new abilities drastically multiply the potential for collaboration, and therefore increase the need to make the ecosystem of copyleft licenses more compatible. Yet, incompatible copyleft–based “free culture” license often impede the necessary sharing of content.

Users of copyleft licenses basically “trust” copyleft to jibe with their intuitive notions of “free” — invariant of bugs or changing particularities of the productive situation. However, copyleft’s complexity and inflexibility betrays this trust, threatening to turn many away from free culture.

Figure 2 illustrates the licensing situation for free culture in general: there is a need for conversion between licenses for various purposes, which can take place outside of copyleft, when there is no license lock.

In the following sections, we discuss copyright dynamics which interact with copyleft to exacerbate the problems.

Copyleft is forever

Copyright originally was designed to expire after a few years, after which time works would revert to the public domain. While it has been widely recognized that abandoning this model has had tragic consequences on culture as a whole (due to near–ubiquity of all–rights–reserved works), there has been little, if any, discussion of the interaction between copyright terms and the license–lock of copyleft. The results are, as it turns out, surprisingly contrary to the spirit of free culture.

Consider just the original of a copylefted work, which will remain accessible under the terms of the original copyleft license and only these terms for as long a time as the underlying copyright is valid — currently, the life time of the author plus 75 years (by U.S. rules). If one wants to use an ostensibly “free” work in ways not permitted or anticipated by the copyleft license, one runs into exactly the same problems encountered today with all–rights–reserved copyright: the copyright holder could be difficult to determine, difficult to reach, or simply say “no” unreasonably.


Figure 2 The license situation in free culture in general. When works can be moved between licenses upon modification, the potential 'universe' of target
licenses, and hence uses, is expanded.
Figure 2: The license situation in free culture in general.
When works can be moved between licenses upon modification,
the potential “universe” of target licenses, and hence uses, is expanded.


Fans of copyleft will doubtless reply that such uses will be explicitly spelled out in the license. Putting aside for a minute the dubious assumption that the license author can a priori think of all legitimate modes of use, there remain problems even when a copyleft license permits the same general activities third parties will want to perform.

With license lock, derived works are subject to the same license (and therefore conditions) as the original work. Any flaws that were present in the original are carried forward to the derived work. Yet at the same time, the “copyright clock” for the work resets with the act of derivation, because the work has changed. This often adds a new contributor, who then jointly holds copyright. This treatment makes sense, prima facie, as the derived version is certainly a new work. However, should this work be further extended into a third work, the copyright clock resets once again.

There is no limit to how many times this can occur, and in fact, heavy re–use of works is precisely what we want to foster in free culture. But the worrisome consequence of this phenomenon is that it effectively casts the original copyleft license in stone. Thus the license may attain, in practice, an “immortality” that even all–rights–reserved copyright does not have! [3]

What is so bad about this kind of rigidity in practice? Surely, if more of a good thing is also good, then keeping that good thing around forever is even better. Indeed, this mindset probably motivated authors of copyleft licenses and continues to motivate their philosophical adherents.

In any of these situations, the “free” license may suddenly seem to become as “unfree” as a straightjacket.

But even if a copyleft license were somehow “perfect,” its interpretation by content creators still would not be. Further, the situations under which the license might be used may change, as new kinds of efforts build upon old content, or the character of the same commons–based project shifts over time. Even if the “escape hatch” of future version re–licensing is left open by the license author (as discussed in the next section), there is no guarantee that the license will be modified to fix what are perceived as problems or undue limitations by its users.

In any of these situations, the “free” license may suddenly seem to become as “unfree” as a straightjacket.

Undue control of free meta–culture

At this point we must introduce a new concept to more fully understand another set of problems with copyleft and free culture licenses. We call this concept “free meta–culture,” and define it as the social rules, practices, and conventions which have to do with creating the rules of free culture itself. Free meta–culture, then, refers to free culture, and exists one logical level “above” it.

Issues of creating and modifying licenses and copyright law itself are free meta–culture issues, as are social practices surrounding the use of licenses.

We have found that there exists considerable undue control in the domain of free meta–culture. By “undue” we mean that this control is not optimal or natural; we find that it causes problems or offends the sensibilities with respect to free culture.

A catalog of some of these exercises of control follows:

  • Free culture licenses which themselves are not free culture — that is, free to modify, translate, or alter in any sense. These licenses have the surprising effect of limiting the ability of third parties to create new free culture licenses, because deriving new licenses from them is prohibited! This problem accentuates the flaws in the extant licenses, and is further amplified by the difficulty of creating new licenses from scratch.

    For example, the GNU FDL license is itself licensed under terms that forbid modification. Technically speaking, this means any license derived from the FDL — whether or not substantially similar — is prohibited. Practically speaking, any license which is suspected by the copyright holders of the FDL of being derived from it opens the author of the new license (or worse, users of it) to legal action (possibly punitive). This is the fog creeping in again, casting a shadow over the creation of licenses — a fundamental free meta–culture activity.

    To illustrate this point, consider the Choral Public Domain License [4]. This is a copyleft license based upon the GNU GPL which has been adapted to use with musical scores rather than computer programs. There is a large digital library, the Choral Public Domain Library [5], the contents of which are provided under this license. Because of the copyright on the original GNU license, this license itself is only legal if its authors have obtained proper permission from the Free Software Foundation, which owns the copyright to the GNU licenses. Should it turn out that the authors of the derived license had not obtained proper permission, one would encounter a rather perplexing legal conundrum: would the license still be valid if the text of the license itself stood in violation of copyright law? Would it be impossible to distribute copies of works in this library because, on the one hand, their license requires that a copy of the derived license be distributed along with the work but, on the other hand distributing a copy of the derived license would infringe on the copyright to the original license?

    We can find no legitimate reason for this kind of restriction. In fact, we think it would be good to have a law that puts all legal codes and contracts in the public domain [6].

  • Free culture licenses which provide “escape” clauses allowing the license author to bind users to future, not–yet–existent versions of the license. Anyone who agrees to such a license is, in effect, agreeing to ad hoc law which they have not yet read. Indeed, who is to ensure that the license copyright holder will even be of the same philosophical mindset in the future?

    While we know of no license which literally allows a spontaneous update of an extant work actioned by the license author, the danger is that third parties may choose to derive new works under new versions of the license unbeknownst to the original author/licensor, thus allowing the license author to exercise control over the meta–culture in perhaps the most important way.

    All of the GNU licenses (GNU GPL, FDL, and lGPL) are examples of licenses with this kind of clause. In fact, the FDL states that if you do not explicitly reference a license version with your use of the license, the work will automatically be covered under whatever the latest version of the FDL says.

  • Licenses which are based on the transfer of copyright from the author to a second party, thus passing control to that second party — the “sign your soul over to us” problem.

    For example, the Creative Commons “Founder’s License” requires creators to transfer their copyright to the Creative Commons organization, which then acts as a benevolent steward of the copyright. While we do not doubt their altruistic intentions, we cannot see why there is any need to give a second party the essentially unlimited powers of copyright ownership when a license which effects the same change but keeps copyright with the author will do just fine.

    If a second party owns copyright, they have additional power in the sense of being able to re–license the content under arbitrary licenses to third parties, regardless of the agreement between the original content author and the copyright–holding organization.

The origin of these problems seems to be that the founders of modern free culture (particularly, those associated with the GNU project) recognized the existence of free culture, but not meta–free culture. Without this concept, those with a particular perspective on free culture cannot even begin to understand how just one type of license is not enough, why license lock can be bad, why restrictively licensing licenses is censorial, and in general, why asserting any control over meta–free culture diminishes free culture.

It would, of course, be wrong–headed to fault the free culture pioneers for this. Back when GNU was founded, the “GNU vs. Hoarder” dichotomy looked to be a complete world–view, and thus a unitary conception of free culture seemed more appropriate (Stallman, 1999). Rather, we are suggesting here that it would be wrong to continue with this mindset into the future.



“Chilling effects”

A copyleft license which turns out to be a poor fit for a project (as the project takes shape over time) is largely the situation with PlanetMath and the GNU FDL license. The GNU FDL license was designed for computer software manuals, but it was adopted by PlanetMath anyway (and also of note, Wikipedia) because (1) there simply was no such thing as a collaborative, “encyclopedia–like” Web site at the time and therefore no special license for it; and, (2) developing licenses ab initio is an incredibly daunting task, especially to legal novices with few resources.

We are now examining the prospect of freeing ourselves and our contributors from the mismatched specifics of this license, not only because of the difficulty in meeting its literal terms, but also because of the inconvenience to downstream adapters of the content. In general, a conceptual fog has been created by the mismatch of the license with the project, which continues to confound those who would work with our free culture content.

Unfortunately, to do this, we face the prospect of a massive copyright–holder inquiry–and–release process: we would need to labor intensively track down thousands of contributors (some of whom have certainly faded from activity in the project) in order to get permission to re–license. The copyright on most entries is now jointly held because of collaboration, further complicating the decision–making. Practically speaking, this re–licensing operation would be impossible for our administratively small project, and content will likely have to be left out to even partially succeed.

While these complexities could be alleviated by future version of the FDL which could be compatible with a simpler and more appropriate license, we have reason to believe that this will not be the case [7]. Even if it were possible or highly likely that the next version of the FDL would solve all of our problems, we suggest this sort of uncertainty and risk constitutes an entirely unnecessary sort of fog.

In fact, the free culture community is faced currently with the specter of functionally identical “free” licenses being kept deliberately incompatible just because they are informed by divergent philosophies (or, worse, ones the license authors feel are divergent). This suggests to us that straightening out incompatible licenses is not simply a matter of technicalities; it is a process of diplomacy between communities and value systems that must be negotiated. Unfortunately, the legal technicalities which arise because of a dearth of this negotiation threaten are already limiting progress in free culture.

Ultimately, there is no reason to cling to any particular license or set of licenses or, indeed, to the very notion of copyleft licenses at all, if it turns out they are doing as much harm as good. It is the principle of fostering “hackability” and “remixability” that ultimately provides for the viability and vibrancy of free culture.

An unhealthy competition between “colleges”

Proponents of incompatible copyleft licenses may argue that it doesn’t matter that license lock creates mutually exclusive “colleges” of content — after all, isn’t value only being added by all of the licenses?

While there is some truth to this point, it also presumes a narrow definition of value and a near–term analysis. It is clear that relative to the potential of compatible licenses that share the same functional objectives (and the need for change–in–licenses through time), vastly more mutual exclusion is occurring than is necessary.

Figure 3 diagrams portability between specific licenses and license categories. Also inserted is a hypothetical “openness license” category, which we discuss later. Moving downward in this diagram “traps” users and prevents them from moving back up. Most tragically, at the bottom level, no lateral movement is permitted at all, despite the fact that the licenses here are philosophically very similar (if not identical).

The restriction of sharing to license colleges severely undermines the fundamental free–culture principle that people should be free to share intellectual content in order to build new content. Again, this is not merely a theoretical point, as practical consequences can already be seen.

For example, the introduction to Linux and the code samples on the e-Theist wiki are licensed under a Creative Commons license, whilst GNU/Linux itself is licensed under the GNU GPL license. So, it would not be possible to share content or code between the two projects without obtaining the explicit consent of the copyright owners.


Figure 3 How portability between free culture licenses works. License categories are horizontal layers and specific licenses are the individual 'bricks.'
Figure 3: How portability between free culture licenses works.
License categories are horizontal layers and
specific licenses are the individual “bricks.”


As another example,WikiTravel cannot share content with Wikipedia, because Wikipedia is provided under the GNU FDL and WikiTravel is under the Creative Commons Attribution-ShareAlike license. The WikiTravel project opted for a different license because the baroque procedural requirements of the GNU FDL would have placed an undue, if not terminal burden on the project. As the WikiTravel copyright page explains [8], the FDL would make difficult typical and legitimate use of WikiTravel content (such as circulating excerpts), thus making WikiTravel somewhat unattractive.

Counter–intuitively, the extremely detailed and exhaustive nature of the FDL license would end up chilling fair use. WikiTravel had to eschew the FDL in order to survive and succeed, but this came at the cost of becoming isolated from all the content and productive efforts of the sister project, Wikipedia.

What is particularly tragic about instances like this is that the incompatible licenses are philosophically the same — the GNU FDL is basically a “share and share alike” license, but the Creative Commons version is less restrictive beyond the basic nature of the license. However, both of these licenses, in being based upon license lock, would technically forbid mixing of content under the exact same terms by a license of a different name.

Figure 4 illustrates the damage done in situations like this, when license colleges create content colleges in kind, fragmenting free culture [9].


Figure 4: How license colleges harm free culture. Two alternate universes,
top and bottom. In the top situation (the present one), commons-based
productive activity is all-too-often fragmented based on individual licenses, even when the licenses are philosophically the same. Individual projects are shown as the bases of pyramids, with stacks representing the value added through free culture activities. In the bottom situation, all content which is to be licensed under terms that are philosophically coincident can be intermixed, resulting in a broader base for the same free culture value-add. The upshot is surplus value which is greater than the sum of the parts - the very dynamic sought in fostering free culture. This value will always be lost as long as license-locked copyleft predominates.
Figure 4: How license colleges harm free culture. Two alternate universes, top and bottom. In the top situation (the present one), commons–based productive activity is all–too–often fragmented based on individual licenses, even when the licenses are philosophically the same. Individual projects are shown as the bases of pyramids, with stacks representing the value added through free culture activities. In the bottom situation, all content which is to be licensed under terms that are philosophically coincident can be intermixed, resulting in a broader base for the same free culture value-add. The upshot is surplus value which is greater than the sum of the parts — the very dynamic sought in fostering free culture. This value will always be lost as long as license–locked copyleft predominates.


As the authors have learned from experience maintaining PlanetMath, it is not reasonable to expect that the average contributor will have a sufficient grasp of copyright law to even navigate fair use and attaining permission with legal correctness. On top of this, having to deal with an ecosystem of free licenses which are not all mutually compatible in surprising ways, or which only permit copying provided that certain intricate protocols are followed, exacerbates the underlying hazards of copyright. A likely consequence is that authors may shy away from participating in free culture because they may not wish to deal with these confusing complexities.

This issue of incompatible licenses is not without social ramifications. Insofar as the act of sharing content is an act which strengthens the bonds that hold together communities, any barriers to sharing are likely to either produce schisms in knowledge communities, aggravate existing divisions, or simply be ignored (however illegally). Either outcome would be unfortunate — the former runs contrary to the philosophy of sharing which is fundamental to free culture whilst the latter calls into question the practicality and validity of free licenses.

Perhaps worst of all, rank–and–file users of copyleft licenses may feel that they are being treated as pawns in ideological squabbles between proponents of different schemes of copyleft.



License, contract and copyleft

It is worth reflecting on what is going on here, and how we, the free culture community, got to this point.

Licenses are contracts, and contracts essentially do two things. Firstly, they spell out the rights and responsibilities of the signing parties, so that they may understand clearly how to proceed. Secondly, they provide a legal impetus for the parties to stick to these conditions, under penalty of law.

This second purpose is actually the weakest motivation for a contract. In general, both parties are voluntarily undertaking their contractual obligations — otherwise they would not be agreeing to the contract! A contract being used by one party to deceive the other is not legally binding, and a contract one party plans to renege on will likely not result in future contracts with that party. The main reason for a contract is thus to spell out the rights and responsibilities of the parties involved: a contract is more of a conceptual device than a coercive device.

The only difference between a license in specific and a contract in general is that a license–contract is entered into by performing an action described in the license (such as copying or modifying), rather than “signing” the license. It follows, then, that the main reason for a license is not to force people to use it and punish them for disobeying, but to spell out how the licensed content can be used so that people might be convinced to voluntarily follow the terms.

We believe that the fundamental problem with copyleft — and thus free culture licenses in general — is that they place an emphasis on the minority coercive aspect of contracts, rather than the majority persuasive and informative aspects.

What the vast bulk of participants in free culture want to do is simple and inter–compatible, but you would not know it from the set of free culture licenses out there, particularly the ones in heaviest use, and especially copyleft licenses.

As suggested earlier, the formative days of free culture were simpler times. This is not to say that they were easier times — in fact, they were quite the opposite for free culture. Free culture, particularly in the realm of software, was almost nonexistent, and proprietary culture was a seemingly unstoppable behemoth. This inspired Stallman’s concept of the world as “Sharer vs. Hoarder,” leading to his solution of copyleft to oppose copyright–with–proprietary–license. Copyleft is a notion of free culture license which is plainly based on coercion–forcing future users to use copyleft. Going even beyond the text of GNU licenses, Richard Stallman is quite explicit in his GNU Manifesto (Stallman, 1985) that the world (particularly “hoarders”) must be forced into a situation where they are forced to utilize and participate in free culture.

This conception of licensing served its purpose for the first 20 years of contemporary free culture, the last ten or so actually containing most of the societal impact. Indeed, a coercive notion of licensing may have been necessary to break the momentum of the similarly coercive proprietary–licensing situation.

Copyleft was a “fire” to fight the “fire” of proprietary license, but now innocent bystanders are getting burned.

But the time has likely passed for this method of coordination as the exclusive or predominant one for free culture. For a short time, executing free culture with an emphasis on coercion rather than persuasion worked. Copyleft was a “fire” to fight the “fire” of proprietary license, but now innocent bystanders are getting burned.

Copyleft achieved its purpose so well that the world has changed far beyond and far faster than its creators imagined, making its deprecated conceptual framework all the more evident and limiting.

Today, Fortune 500 companies use and support F/OSS software not because they have to, but because they see it as furthering their own best interests (and that of their shareholders) [10]. Governments have realized that free culture and open implementations of government systems are demanded by citizens and coincide with the ideal of democratic rule, resulting in a sweeping wave of official F/OSS proclamations by nations [11]. After flirting with proprietarism for half a century, the scholarly world is beginning to realize that openness and free knowledge are not only better for achieving its short–term ends, but in fact, are the very essence of its mission [12].

The upshot is that we seem to be safely “over the hump” in the adoption of free culture. It seems to us that if we were, today, to remove all coercive elements of free culture licenses (those that force sharing), free culture would still continue to flourish and grow. In fact, as we have pointed out, the copyleft regime has actually become limiting, so there is likely growth to be lessening the restrictions of copyright, above and beyond the risk of “hoarders” creating proprietary culture.


Figure 5: When free culture licenses aren't free: licenses that cannot be modified place similar licenses off-limits to free culture innovators, or at the very least, cast a growing fog on their use.
Figure 5: When free culture licenses aren’t free:
licenses that cannot be modified place similar licenses off–limits to
free culture innovators, or at the very least, cast a growing
fog on their use.


In the next section, we discuss ways that this can be done, as well as creative, non–coercive ways to ensure and increase the prevalence of free culture in society.



Lifting the fog

Despite the seemingly rigid license–lock structure of copyleft and the sometimes fractious nature of the admittedly loose conglomeration of activists and creators we have been calling the “free culture community,” there are some promising potential solutions. We divide these solutions into two families: the technico–contractual, having to do with legalistic solutions that can be adopted in the near term, and the socio–political, dealing directly with the social milieux of free culture and solutions that will take concerted, longer–term effort to bring about.

Technico–contractual solutions

Firstly, for the sake of free culture itself, we recommend abolishing the practice of placing functional restrictions on the text of licenses themselves (i.e. copying or modifying them for new uses). There is no longer any need to have this or any other control in the domain of meta–free culture, just as it undermines the entire basis of social justice to have secret laws or ones that cannot be revised through a democratic process. In Figure 5, we provide a graphical representation of the problem of restrictively copyrighted free culture licenses, showing how it gradually limits the available options to coordinate free culture activity as the “fog” creeps in.

To begin to make life easier for collaborative projects, one idea is to improve contractual arrangements between umbrella entities of CBPP projects and their contributors, giving the umbrella entities “options” to make license determinations further on down the road. For example, the project could require contributors to agree to a contract whereby the current owner/lead maintainer of an article (or module, section, etc.) can make re–licensing determinations (as long as the subsequent licenses respect the intentions of the original authors and the guiding principles of the project). We believe some sort of methodical treatment of CBPP efforts is needed to deal with what is an essentially new kind of intellectual property and creative process.

To improve free culture licenses in general, one of our favorite ideas is to promote licenses that explicitly, voluntarily build in limited terms — as in the original, constitutional conception of copyright. In this model, licenses with fewer restrictions (such as public domain or attribution–only) would automatically become enacted after the initial limited term. For example, two years could be a reasonable term in our current landscape of rapid conceptual innovation. This kind of license would eliminate the need for relinquishing copyright, as in the case of the CC–Founder’s license.

The above could be paired with stipulations that the “copyright clock” always begins with the initial copyright, that is, the initial moment of authoring of the oldest piece of the work (or equivalently, version of it). Without this clock being reset with every adaptation of a work, copyleft licenses would have a natural sunset, and so would their flaws and limitations.

Perhaps most radically, an entire universe of new free licenses is opened up if license lock is done away with. In fact, license lock turns out to be just a “legal shortcut” relative to the goal of keeping subsequent works free. In reality, creative freedom is a matter of the what people are allowed to or prevented from doing with an intellectual work — it is not a matter of the specific wording of a particular instance of a copyleft license, which is just an attempt to achieve a set of free culture objectives. If instead of self–referentiality, functional capabilities are used to specify re–licensing, license lock would disappear.

The least controversial, “low–hanging fruit” of this kind of reform would be to change copyleft so that it “locks” works into availability by stipulating that derivative works can be licensed under any license which requires open sharing. This would create the “open licenses” category shown in Figure 3 and save the free culture movement a lot of headaches.

People still share when they are allowed to — even if they are not forced to — because they want to and because it is in their own best interest.

Of course, we would like to see more critical examination of the need to force sharing in the first place. In fact, many extant free licenses do not utilize license lock to achieve their objective. The BSD/MIT/X11 family of licenses specify that only that the copyright notice of the source work be propagated with derived works. Other than this basic requirement, adapters are free to do what they want with their derived works, including decide how they want to license them.

While the GNU mindset is that such works could not possibly become important artifacts in a vibrant free culture (due to the lack of license lock forcing derivability), we see plenty of works with BSD–style licenses out there in the wild (most notably, BSD and X11). As we argued above, people still share when they are allowed to — even if they are not forced to — because they want to and because it is in their own best interest.

Finally, we advocate a general principle of “license minimalism” to reduce fog. In our surveying of the available free culture licenses [13], nearly all of them (particularly the copyleft licenses) contain a significant number of redundant clauses. Some general categories of these clauses are superfluous disclaimers against liability and warranty, disclaimers of termination that only repeat general legal rules, express reservations of the ability to multiply–license a work, and many more.

Another problem is the adoption all or in part of pre–written licenses. While this is often done for economic reasons (and to make the task even possible without law expertise or the need to hire a lawyer), it runs the risk of carrying along the “baggage” of old clauses which are only in the original situation.

Licenses with such a high degree of insubstantial “filler” and “legacy code” simply become overwhelming for non–lawyers. Unfortunately, this characterizes the vast majority of free culture licenses out there. In cases where there is concern that extra disclaimers are truly needed, perhaps this is a sign that laws need alteration or clarification. Besides this, we suggest that a useful conceptual test to apply to a potential clause would be to imagine whether the clause would be likely to alter any standing of parties involved in court, before an actual judge.

Socio–political solutions

As discussed above, contracts are not really meant to coerce, and copyright was not really meant to apply to anything like the current usage landscape. Accordingly, we believe the free culture community needs to ask itself whether wielding copyright law in the highly coercive manner of copyleft is really the best means of achieving its objectives. While there is no question that licensing is an appropriate means for clarifying responsibilities and protecting rights, it is not at all clear that it is appropriate or advisable to force sharing by means of copyleft.

Perhaps a better strategy would be to promote institutions and attitudes which offer social and economic incentives to share content and corresponding disincentives to those who chose to be selfish. For instance, we propose that the following measures would go a long way towards curbing abuses of excessive copyright:

  • Learned societies might refuse to admit or endorse individuals and publishers who impose restrictive copyright terms on the grounds that they are inimical to the norms of science and scholarship.

  • Businesses and professional organizations might deem reliance on closed licensing terms as an unethical business practice.

  • Consumers might regard cripple–ware as an inferior, defective product to be avoided.

The history of this subject has been one of reaction. Publishers reacted to the advent of new technology by demanding stricter copyright laws to deter wouldbe pirates, and the free culture movement reacted with copyleft. Unfortunately, reactive approaches rarely address the deep causes of problems and tend to foster division and debate rather than harmony and mutual understanding. It would be better to adopt a pro–active stance.

On the public level, this means shifting the focus of attention from the judiciary to the legislature. By its very nature, the judiciary is reactive — it can only respond to events which have already happened and is circumscribed both by existing law and precedent. For instance, one could lobby for an amendment to copyright law which would officially create new rights which copyleft has attempted to impose by ad hoc constructions. One could also work towards legal recognition of CBPP and regulation which delineates the rights, responsibilities, and liabilities of those involved in this new mode of production.

In the private sphere, this means that the different parties affected by this issue need to meet to discuss their concerns, make a sincere attempt to understand each other’s viewpoints, seek common ground and frame a plan of action which allows the new technologies to be used to their full potential — in a manner which respects the legitimate interests of all and is mutually beneficial. This is a considerably more difficult route to follow — it requires an unusual degree of foresight and enlightened self–interest, and it demands leaving comfort zones to try new business models — but the results are concomitantly great. This may be the only road which will lead to a satisfactory resolution of the issues at hand and usher in a climate in which the now–intertwined “digitized” culture and computer technology can flourish to their full potential.




We close with an observation fitting for a pair of mathematicians: if the reader believes the general argument of this paper is wrong, then the paper is right. This is because if the paper is wrong, the authors must be confused about copyleft. But if this is so, then the premise of the paper is right: copyleft must indeed be causing a burdensome fog.

We do not believe that copyleft is inherently a bad or flawed concept, but that it has served its historical purpose and its time has waned. The world has now accepted code and content as free culture voluntarily; it does not need to be forced to do so through aggressive licensing. Copyleft has thus become a legacy that continues to be all too often uncritically adopted and propagated.

To remedy this, we recommend that policy pursued should be many-pronged: one of simplifying licenses to make more free culture content inter–compatible, reducing the use of licensing, eliminating control from the “meta” level of free culture, and encouraging users of licenses to take more initiative in understanding them and using them properly. Simultaneously, we should invest in changing social practices so that they fundamentally encourage the development of culture in a free, rather than proprietary state. End of article


About the authors

Aaron Krowne holds an MS in Computer Science and a BS in Mathematics from Virginia Tech, where he was a research assistant at the digital library research lab (DLRL) working under Edward A. Fox on the CITIDEL ( digital library project. He currently is Head of Digital Library Research at Emory University’s Woodruff Library, in Atlanta, Ga., where he leads the research on grant projects in the MetaScholar initiative (

In 2001 Krowne co–founded the PlanetMath collaborative digital library and Web encyclopedia ( with a fellow student at Virginia Tech and an informal, worldwide group of mathematics students and enthusiasts. Recently PlanetMath reached a milestone of 5,000 encyclopedia entries. There have been hundreds of active contributors to the site and it has about 20,000 page views a day. Now PlanetMath is a non–profit, tax–exempt public charity, with Krowne as president. The non–profit entity seeks to sustain and advance the Web community, free mathematical knowledge, and commons–based information production itself.
E–mail: akrowne [at] emory [dot] edu

Raymond Puzio holds a B.A. in Physics from Columbia University and a Ph.D. in Physics from Yale University. He has held appointments at at the Centre for Gravitational Physics and Geometry at Pennsylvania State University and the Physics Department of the University of Mississippi. Currently, he is an Adjunct Professor at the University of Memphis where he is part of the Fractional Calculus research group. In addition to his interests in mathematical physics, Dr. Puzio is interested in the development of computer and internet tools for mathematicians — he is heavily involved with PlanetMath and the Hyperreal Dictionary of Mathematics both as a contributor and in administrative capacity.
E–mail: puzio1 [at] excite [dot] com



We especially want to thank Joe Corneli for his participation in extensive discussions which led up to this article, and for giving his many useful comments on drafts.



1. As an example, can anyone tell us the legal status of Wikipedia articles (perhaps the vast majority) which have anonymous, likely completely un–identifiable contributors? This would seem to place such articles in a legal grey area, both in terms of the GNU FDL and copyright itself. Copyright licenses, and in fact contracts as a working instrument, have always been predicated upon distinct, identifiable parties.

2. And we mean all copyleft licenses; we have not found a single one which functions without license lock.

3. Assuming, of course, that the Congressional lobbying for copyright extensions will stop or fail to be effective at some point.

4., accessed 14 July 2006.

5. See, accessed 14 July 2006.

6. This would solve other problems related to obfuscated law, such as public legal codes (e.g. building codes) that are not available except at a significant monetary cost.

7. Private conversation with Richard Stallman regarding the future compatibility of the FDL with Creative Commons.

8. Why Wiktravelisn’t GFDL, accessed 14 July 2006.

9. This figure might lead one to believe that the “surplus value” created by inter–compatible free culture content is of quadratic proportion. We believe this is in fact incorrect, with the actual difference in value being of exponential order, on a group–forming–networks (GFNs) basis. We explore this quantitatively in an appendix to the full version of this paper.

10. IBM is a famous example of this, with their embrace of Linux.

11. Brazil, France, and Spain are noteworthy here. The U.S., in fact, lags in this department.

12. Ironically, it is probably the corporate world that is furthest ahead in the voluntary embrace of free culture, and the scholarly world that is furthest behind. Speculation on this is fodder for a separate, lengthy digression we will not delve into here.

13. See, for a good list,, accessed 14 July 2006.



R. Stephen Berry, 2001. “Is electronic publishing being used in the best interests of science? The scientist’s view,” In: Second ICSU–UNESCO International Conference. Electronic Publishing in Science, at, accessed 14 July 2006.

Aaron Krowne and Nathan Egge, 2001. “Planetmath: A collaborative, Web–based mathematical encyclopaedia,” at, accessed 14 July 2006.

Lawrence Lessig, 2004. Free culture: How big media uses technology and the law to lock down culture and control creativity. Mew York: Penguin Press.

Richard Stallman, 1999. “The GNU project,” In: Chris DiBona, Sam Ockman and Mark Stone (editors). Open sources: Voices from the open source revolution. Sebastopol, Calif.: O’Reilly.

Richard Stallman, 1985. “The GNU manifesto,” at, accessed 14 July 2006.

Siva Vaidhyanathan, 2001. Copyrights and copywrongs: The rise of intellectual property and how it threatens creativity. New York: New York University Press.

Jimmy Wales and Larry Sanger, 2001. “Wikipedia: The free encyclopedia,” at, accessed 14 July 2006.

Editorial history

Paper received 1 May 2006; accepted 18 May 2006.

Copyright ©2006, First Monday.

Copyright ©2006, Aaron Krowne and Raymond Puzio.

The fog of copyleft by Aaron Krowne and Raymond Puzio
First Monday, Volume 11, Number 7 — 3 July 2006 (Proceedings of the 2nd First Monday Conference, May 2006, Chicago)

A Great Cities Initiative of the University of Illinois at Chicago University Library.

© First Monday, 1995-2019. ISSN 1396-0466.