First Monday

There's no such thing as free software (And it's a good thing, too) by David McGowan

The following commentary is part of First Monday's Special Issue #2: Open Source.



One of the more prominent F/OSS slogans is: “Think free speech, not free beer.” That sounds nice, but I often wonder whether F/OSS advocates have any idea of what a mess free speech law really is. This is not the place to go into that, but take my word for it: The doctrine is unenlightening; the theory is worse.

So in thinking about what we have learned and where we are going, I want to continue the theme that free speech and free software have something in common. They do: Neither one is “free” in the most intuitive sense of the term. Free speech law does not construct a realm of human behavior free from state-backed ideology; it is a state-backed ideology. Free software is not free from “intellectual property” law; it rests squarely upon it. And, to borrow (steal shamelessly, actually) from Stanley Fish, it’s a good thing, too.


What Have We Learned?

Copyright is a wonderfully flexible tool for unleashing human creativity. Pretty much everyone recognizes now that F/OSS programming is not an escape from copyright but a particular way of deploying it. That is not going to change. There will be no Nirvana in which the rights fall away and all software is “free,” which in any event would only lead to commercial firms incorporating F/OSS code and then treating compiled versions as trade secrets.

That is why F/OSS advocates sometimes sound like any other author who wants to control a work. In response to criticisms of the “viral” nature of the GPL, they respond in substance: “I wrote it, this is what I want to do with it, if you don’t like it, write your own.” They are right.

Licenses are important, but it is the work that matters. Every lawyer knows that the sociology of agreements is at least as important as the law of them. [1] The social relations that form around and through agreements are the real life-blood of contracting or licensing. Agreements lay out what people can expect from each other, give them security in their expectations, and thereby give them the confidence to proceed to do things they want to do. The terms of the agreement need to be reasonably clear, and it helps to have lawyers lurking in the wings to justify people’s confidence that their expectations will be met, but the value of the agreement lies in the behavior it facilitates, not in the document itself.

So it is with the GPL. For all that is written about it, and all the crazy things that are said, it has worked well because it has captured and sustained a development sociology that works. That matters a lot more than the legal details of the document, or the details of how it might be enforced in different jurisdictions around the world. Those matter, but they matter less than the work. The GPL stands as an example of how a fairly simple but creative idea (copyleft), put in plain terms, can establish a structure where a lot of good work can get done, lawyerly nitpicking notwithstanding.

Competition is good. GNU/Linux is too often used to represent all F/OSS projects, which I think tends to distort views of F/OSS development. A random survey of projects on SourceForge would produce a different picture of the importance of F/OSS than would a study of GNU/Linux.

Having said that, I will commit the sin myself and say that GNU/Linux has shown how F/OSS development sociology can spur competition. I have in mind competition on two axes. The first is on openness, which we can see when vendors like Microsoft offer source to customers (albeit on a more restricted basis than F/OSS advocates would like, but that is the author’s option). The second is over whether it is better to fix computing knowledge in an OS or to commoditize the OS and sell human knowledge as services. We can see that form of competition when Red Hat backs GNU/Linux and charges a per-server license for support.

I don’t know how these competitions will end, and as a non-technical user, I don’t really care. Competition is good. It spurs competitors to greater efforts, and it benefits doltish consumers like me. Even those outside the F/OSS movement will respect it as the best way of producing better, cheaper products.


What Happens Next?

As either or both Niels Bohr and Yogi Berra said, predictions are hard, especially about the future, but I think it is safe to say that F/OSS development will confront a couple of important issues in the next few years.

Managing the GPL Brand. Though there are a lot of F/OSS licenses (more on that in a minute), the GPL is the flagship.[2] My guess is that programmers choose it as much for what it says as for what it does when compared to alternative licenses. That means the GPL functions sociologically as a trademark, though legally it is a license (or contract, if you take that view). That makes the FSF—which owns the rights in the GPL—a brand manager. How it manages the GPL brand will determine how robust that brand will be over time.

Two management issues seem relevant in the near term. The first, and most obvious, is managing the development of version 3 of the GPL. There are several issues here, including how far the license will try to be a strategic tool against patents, whether it will present problems for projects that have developed their own understandings of GPL 2, and whether its differences from version 2 will be so substantial that projects could fork between different versions of the same license. That might happen if some project contributors opted for version 3, others did not, and the differences between the two mattered.

The second issue is related to two of these management concerns. It has to do with proliferation of interpretations of the GPL’s language. The GPL is a form that the FSF owns and allows developers to use. The GPL is not itself GPL’d—another example of the point that there is no such thing as free software—so the FSF can keep the language itself from fragmenting too much.

The GPL’s language does not confer rights on the developer, however. The law does that by default, and developers who have rights may exercise them or not as they see fit. That means individual developers might allow practices in their own GPL’d projects that the FSF would not allow, and vice versa.[3] The management issue is to allow enough flexibility for the GPL to continue to facilitate work without diluting the brand so much that it loses the ability to express the message its users find attractive.

Dealing with commercialization of GNU/Linux. As noted earlier, GNU/Linux has exemplified F/OSS development for many people, and GNU/Linux is becoming increasingly commercial. A glance at the corporate logos associated with Open Source Development Labs suggests that GNU/Linux has been taken over by Nascar.[4]

A cynic could be forgiven for thinking that these firms are largely united by a desire to make money by selling complements to a cheap OS. (Firms at every level in a stack of complements will want each other complement to be as cheap as possible, so they can grab the largest possible fraction of the surplus generated by the whole stack.)

Programmers might be less willing to write for a project that generates profits for someone else than for a project run strictly on the basis of mutual sharing. This should not be a problem for F/OSS development generally—GNU/Linux is not the modal project—but it might mean that more and more kernel development will be performed by employees of firms who see GNU/Linux as an element in their business strategy. If that is the way things go, it will be interesting to see how the F/OSS development sociology evolves as an increasing fraction of work on the flagship project is performed by firms that are in it for the money.

Dealing With License Proliferation. When I first started researching F/OSS development, there were about 40 licenses. There are now more than 60. As the OSI’s Larry Rosen has written, the proliferation of licenses creates the risk that technologically compatible code will be legally incompatible.[5]

License proliferation exemplifies the tension between freedom to do what one wants and the need for coordination and standardization to get things done. This issue also exemplifies why there is no such thing as free software. Coordination isn’t freedom, but neither is legal incompatibility.

Dealing With Litigation. I suspect that net social welfare varies inversely with the income of litigators, but that does not mean litigation is going away. As F/OSS projects move farther into commercial space, presenting profit opportunities and threatening for-profit firms, there will be litigation. SCO v. IBM is not a crisis, it’s just the way the commercial space works.

F/OSS developers will need to develop litigation management techniques and experience, along with complementary institutions such as insurers, to normalize this fact of life. That work has begun; its success or failure will be an important factor in the future robustness of F/OSS projects that substitute for commercially significant software.


Netting It Out

No doubt you will have noticed that my lessons are fairly anodyne and the challenges are pretty mundane. There is no high philosophy here, no metaphysics of software. It is normal (social) science, not a paradigm shift. That is what I see, and it seems to me a far more promising basis for the future than the alternative.

At the end of the day, F/OSS development is about smart people using generally applicable legal tools to help them arrange their affairs as they please. Partisan rhetoric—on both sides--can tend to obscure that fact, but that is a reason to ignore the rhetoric rather than to suspect the tools. Simplicity is elegant in law as well as elsewhere, and it is the simplicity and straightforwardness of the approach that we should celebrate. Give people rights in their work, let them use those rights to realize their vision, let creativity flourish, and let the chips fall where they may. It is not really freedom, but it is really cool.

So next time someone says “free as in freedom,” or talks about speech and beer, remember: There is no such thing as free software, and it's a good thing, too.End of article



1. I'll ignore the debate over whether F/OSS licenses are permissions or contracts; it does not matter to my point here.

2. For example, Professors Josh Lerner and Jean Tirole found that 72% of open-source projects hosted on SourceForge employ the GPL. JOSH LERNER AND JEAN TIROLE, THE SCOPE OF OPEN SOURCE LICENSING (Harvard NOM, Working Paper No. 02-42, 2002).

3. For example, MySQL recently stated that it would allow its GPL’d program to be modified and redistributed in connection with programs under F/OSS licenses that are not compatible with the GPL; the GPL itself forbids such modification and redistribution.MySQL FOSS Licensing Exception, at (Mar. 12, 2004).




Copyright ©2005, First Monday

Copyright ©2005, by David McGowan

There's no such thing as free software (And it's a good thing, too)
First Monday, Special Issue #2: Open Source — 3 October 2005