In the proprietary production world, what matters about a copyright is who owns it. In the free production world, however, who owns a copyright is relatively unimportant. What matters is what license it is offered under. There is a very simple rule of thumb about the best license to use: use a “free, copyleft license”. Such licenses provide the ideal balance of freedom versus limitations, and projects that use them are overwhelmingly more successful than ones that don’t (both those with more free “non-copyleft” licenses and those with more restrictive “non-free” licenses).

The Culture of Innovation

The heritage of open source development stems largely from academia, where intellectual freedom is as fundamental an ideal as “democracy” or “freedom” is to most people throughout the industrialized West. It is this view of the concept which leads to the ideologically-based “Free Software Movement” and its preference for emphasizing user freedoms over developer process.

While this approach is probably not so good as a method of persuasion, since it relies on cultural norms that do not apply broadly across all human societies or even across professions, it has a special importance to commons-based production: it is a core belief of the people who do the most work.

Whether you share this belief system or not, you cross it at your peril. Many people regard these ideas as moral imperatives and one of the first rules of the freedom game is learning not to offend the very people who are likely to be your most important asset in success. You cannot play the game halfheartedly, hoping to create a business advantage through appropriating publicly-created work, while holding back your own.

Intellectual Freedom

Intellectual Freedom is a fundamental principle that underlies many of the beliefs shared by knowledge workers, particularly in academia, but also in a much broader area of complex engineering and scientific disciplines. Although it is often couched in ideological terms, the real point is that secrets are wasteful. Scientists learn from very early in their training the faults of suppressing information, perhaps most iconically in the person of Galileo Galilei, who published evidence supporting the Copernican theory that the planets orbited the Sun (primarily his observations of Jupiter’s satellites), and was proscribed and forced to recant his beliefs by the Catholic Church.

Scientists learn from very early in their training the faults of suppressing information

Scientists view Galileo in heroic terms, and the Church’s resistance to the Copernican theory was ultimately futile. Without the Copernican theory, we’d have never made it to the Moon. So it is fitting that Galileo’s famous hammer and feather experiment was actually demonstrated by Cmdr. Dave Scott at the Apollo 15 landing site on the Moon.

Commander Dave Scott demonstrated Galileo’s “hammer and feather” experiment on the surface of the Moon during the Apollo 15 mission (Credit: NASA/PD)

When scientists are free to share information and regard it as a duty, they fuel the process of science, which needs to check and recheck assumptions to reach an ever more accurate understanding of the world. Engineers and inventors also share information, so as to attain ever more refined improvements to the inventions that they develop. Software developers use this freedom to find bugs and refine their software as well as to improve upon what has been written before. All of them are using it to avoid wasting time re-inventing what has gone before.

Intellectual Property

The utilitarian argument for intellectual property is fairly simple: producing information costs time and effort of those who do the work, just as much as any other kind of production. Yet, unlike other forms of production, information can be freely copied, so, in a completely free market, the monetary value per copy of an information product tends to be very nearly zero.

Intellectual property systems make it easier to recoup the development costs of information products via artificially inflating the cost of sales to cover the initial investment. This mimics the natural behavior material products, where barriers to entry such as manufacturing tooling costs give the first entry into the market a chance to recoup its development costs so as to make a profit.

Intellectual property systems make it easier to recoup the development costs of information products via artificially inflating the cost of sales to cover the initial investment

Of course, there are problems with the Intellectual Property idea. Perhaps the most obvious is that, in the limit, it’s the just like the medieval guild system that locked Europe into a dark age for nearly a thousand years!

Suppressing the flow of information damns us to repeat the same mistakes over and over again, retarding technological progress and resulting in massive wastes of human capital. Only when inventors, authors, engineers, and scientists are able to build upon each others’ works can civilization reap the renaissance rewards of a booming technological and intellectually creative society. Thus, even if and when intellectual property law is needed, it must always exist in tension against the long range benefits of preserving intellectual freedom.

An unnecessary evil?

Most serious creators of intellectual works in the United States know about the limited constitutional basis for Intellectual Property, but they still view it as a “necessary evil”: a fictive arrangement we have to adopt in order to create intellectual works within our capitalist society. The free market, they argue, demands that we respect Intellectual Property as a tradable good, so that we can profit by producing intellectual works.

At the very least, we know that a free market society can produce intellectual works without the need to resort to the restrictiveness of conventional intellectual property

The experiences of free software and free culture, however, have empirically disproved this idea. At the very least, we know that a free market society can produce intellectual works without the need to resort to the restrictiveness of conventional intellectual property. Free-licensing, which intentionally releases such works from these confines, produces more value from the free exchange of information than it loses to lost licensing sales and free rider problems.

In his essays, The Cathedral and the Bazaar, Homesteading the Noosphere, and The Magic Cauldron, Eric S. Raymond illustrates the strategies that commercial entities and amateur developers have employed to defeat the conventional wisdom that locked-down “IP” is essential to business success. I highly recommend reading these essays yourself.

Why use a copyleft?

There is one serious problem with all this freedom. If everyone is free to do what they want with the work, then one thing they can do with it in a society which has strong intellectual property laws is to claim it for themselves, appropriating all of the effort that has gone into the project.

Over 80% of the software projects hosted on the Sourceforge free software project incubator use a copyleft license of some type

The actual degree to which this happens may not be that great, and the “damage” to the contributors may not seem so large, but the fear of this kind of disparity is large enough to chill contributions to a project. Consider for example, how much smaller is the community surrounding the various “BSD Unix” distributions than the one surrounding “GNU/Linux”.

Why is this?

One need look no further than Apple Computer to see the answer. Apple’s “OS X” operating system, which is used commercially on modern Apple Macintosh computers, is built on top of “Darwin” a particular distribution of BSD Unix. OS X contains many, many improvements on Unix, and tools to make it easier to use. But of course, OS X is proprietary.

People contribute to free projects under the promise that they will stay free. Copyleft offers that promise

This can happen because BSD has no copyleft. The fear of it happening, and thus robbing contributors of the quid pro quo of commons based production, is sufficient to reduce development on BSD Unix platforms. They simply do not interest as many developers.

On the other hand, GNU/Linux, which is mostly licensed under the GNU GPL or LGPL licenses, and therefore protected by copyleft limitations from this kind of hijacking of community effort, continues to boom in popularity.

In other words, people contribute to free projects under the promise that they will stay free. Copyleft offers that promise. Therefore copyleft is very important to the morale of contributors and to the sustainable interest of the community of developers.

“Copying” and “use”

The term “use”, when applied to intellectual works, can be treacherously ambiguous. After all, “copying” a work, “distributing” it, or “deriving” from that work, are clearly ways of “using” the work in the English vernacular. Yet theorists talking about copyright or software freedom generally do not consider these uses to be included in the word “use”.

The term “use”, when applied to intellectual works, can be treacherously ambiguous

Three of the four major definitions of software freedom that exist in the community include the requirement that a work must be “free to use for any purpose”, and yet they also allow copyleft requirements to ensure that a work remains free by placing terms on “copying” and “distributing” it. These definitions include the Debian Free Software Guidelines from the Debian Project, the Open Source Definition from the Open Source Initiative, and the Definition of Free Cultural Works from the Freedom Defined wiki project).

The Free Software Definition from the Free Software Foundation and its GNU Project, is less demanding: it says only that you must have the freedom only to “run the program for any purpose”. That’s less vague, but of course, it also only makes sense for an executable program, which is why the other definitions opted for a broader expression.

The Free Software Definition introduced the idea that there are four fundamental freedoms needed for information works (traditionally numbered from 0 to 3, to reflect their origins among computer programmers!). It originally specified these in very software-oriented ways. This version of the “Four Freedoms” is from FreedomDefined.org’s definition of “Free Cultural Works”

This has always been contradictory if “copying”, “distribution”, and “derivation” are to be regarded as “use”. The true ambiguities of this definition, however, only came to light with the introduction of “digital rights management” (DRM) and “technological protection measures” (TPM). These are both euphemistic names for encryption technologies designed to interfere with users’ ability to copy and decode digital intellectual works.

Some licenses do not permit encrypted distribution whenever it would interfere with users’ legal rights under the license

It was argued by some that the right to distribute a work in such an encrypted file format, even when no key is made available to allow users to unlock the work, was a valid “use” (i.e. not an act of “copying” or “distribution”, which might be subject to copyleft restrictions). Some licenses, particularly those from the Creative Commons organization, do not permit encrypted distribution whenever it would interfere with users’ legal rights under the license. A strong lobby was formed to try to convert Creative Commons’ language over to an alternate form of protection against DRM-laden files, which relied on a requirement to provide a non-encrypted distribution of any file which was distributed in DRM format (an idea which seems logical based on the success of the GPL’s requirement of a “source code” distribution along with any binary distribution).

However, some observers, notably Greg London, noticed an exploit which showed how this form of “protection” could fail to protect users’ freedom to use and/or derive from encrypted works in a useful way. As a result, the Creative Commons licenses retain the anti-DRM language, although the issue remains somewhat controversial.

Non-commercial licenses

A concept in competition with the idea of copyleft is the “non-commercial” license, which attempts to restrict the use of a work for “commercial” purposes. This is a somewhat compelling argument for aesthetic works, since for aesthetic works it is much harder to develop the kind of “service and support” models that have worked so well for free software.

Many people (wrongly) think of free software products as being “non-commercial” because you can’t (or can’t profitably) sell individual copies of the software. However, there are many other ways of using software “commercially” (such as providing support for it, using it as a promotional, delivering advertising with it, and so on). A “non-commercial” clause forbids them all.

Ironically, the only really rational use for a “non-commercial” license is when you want to operate commercially. If you are in the business of selling your work for commercial use, you can partially protect your monopoly, while still taking advantage of the fluid distribution and marketing provided by free internet file-sharing. However, the work never really enters the “commons” of free-licensed work unless you re-release it under a “free” license later (or until the copyright runs out, which takes practically forever under today’s copyright laws).

So, while they may have other uses, for commons-based projects, “non-commercial” licenses are a dead end.

Copyleft conflicts

It is extremely difficult to write a license which insists only that the intent of the licensing on derivatives is the same. It’s much simpler and much more enforceable to require derivatives to be under the same license terms.

Even if you could interpret the copyleft as requiring only the same basic conditions, this will still invariably create obstacles. For example, the GPL insists that no “legal venue” be specified, but some free licenses, like the original Python license insisted (as do many proprietary licenses) that court cases be held in a particular jurisdiction. As a result, the Python license was “GPL incompatible”, even though it was otherwise “free”, according to the Free Software Definition.

Because copyleft licenses can conflict with each other, it’s not good to have a lot of them. Over two-thirds of the projects on Sourceforge are simply licensed under the one “best practice” free software license: the GNU General Public License. Over 92% are “GPL compatible”, meaning that derivatives based on them may be released under the GPL

As a result, copyleft licenses are subject to incompatibilities which can make it impossible to publish a fusion between two packages with different free/copyleft licenses. Since this is obviously undesirable, there is a strong pressure in the community to stick with a very few copyleft licenses—the main one for software being the GPL—thus avoiding problems with “license proliferation” as it is called.

Note however, that proliferation is a much bigger problem for copyleft licenses than it is for non-copyleft licenses. This is why there is relatively little concern over the large number of non-copyleft licenses (such as the BSD, MIT, and Apache licenses). These licenses can be combined into derivatives, and will even allow conversion to GPL, so they do not really interfere with user freedoms in the software. They still make the licensing more complex to read and understand, which is why (even for non-copyleft licensing) there are recommendations not to use so-called “vanity” licenses when one of the standard licenses will do.

Upgrade and compatibility clauses

One way to reduce problems with “license proliferation” is to provide a more flexible “upgrade” or “compatibility” clause. For example, some licenses simply have a clause explicitly allowing them to be converted to GPL licensing (overriding any otherwise conflicting terms). The Creative Commons started introducing a mechanism for forming cross-licensing agreements with its version 3.0 licenses. Upgrade clauses provide a mechanism for migrating from older versions of licenses to newer ones, so that old licensing problems can be fixed by new licenses (the GPL does this through a suggested voluntary statement in the license grant, while the Creative Commons license provide such a clause in the body of the license itself).

Critics argue that such agreements would gradually erode the copyleft, leaving the works effectively little better off than if they were released under a non-copyleft license. However, if you’re going to use another license, it’s a very good idea to assure compatibility with the GPL (for software) or the Creative Commons By-SA license (for aesthetic works).

The problem with hardware

Hardware licensing presents another special problem, since hardware manufacturing processes are generally not subject to copyright or copyright-like protection (with a few exceptions). This means that hardware designs are in the position that software source code would be if there were no copyright protection for executable binaries (the linked article explores this analogy). Thus, it’s apparent that an open hardware copyleft will probably require stronger rules in order to be effective.

Some hardware projects today use the GPL or BSD licenses, but it is likely that a strong copyleft license for hardware will emerge as an evolution of specific licenses like the

When not to use a copyleft

There are obviously some negative effects to using a copyleft. Despite “Freedom Zero”, there are a number of permitted limits to uses of copylefted software, and occasionally they get in the way. If you are interested in supporting commercial proprietary software development, or simply don’t want the hassle of license compatibility issues, then a non-copyleft license may be more desirable.

Freedom, copyleft, and the commons

With proprietary projects, what matters is who owns an intellectual work. This is because such projects operate on a permissions basis, and so what you have to know is who to ask for permission. The overhead of managing this “permissions culture” is enormous. Our society, indeed, is practically being crushed by its costs. The most visible costs (such as lawsuits) are bad enough, but the worst damage happens when people simply give up and assume they can’t get permission. It is this “chilling effect” that is most damaging to innovation in society, and the removal of that burden is the source of the success of the commons-based production movement.

With commons-based production, then, what matters is not who owns the work, but what freedoms you already have in the work. Thus, the nature of the public license granted in the work is of paramount importance. So, don’t pick a bad one and don’t write your own! At least not until you have reviewed all of the existing popular free software, free content, and open hardware licenses, and still can’t find one that works.

The best and simplest choice is to simply use the GNU General Public License. It is quite versatile, and will work for many kinds of utilitarian works such as software or logic designs. You’re in good company if you do this, as this is what about four out of five software developers will do.

If the work is not software, you may be looking at a more complicated choice. A good bet here is to stick with the Creative Commons Attribution or Creative Commons Attribution-ShareAlike license, unless there is a clear and unambiguous definition of “source code” for the work you want to copyleft.

About four times out of five, though, free software programmers pick a copyleft license. And given the success of free software, that is probably the best recommendation for commons-based production.