September 8, 2003

The indispensable GPL

- by John O'Sullivan -
When I first became interested in Linux in 1998, Richard Stallman, the godfather of the GNU General Public License, was widely regarded as a crank -- someone whose past accomplishments were crucial, but who like Trotsky insisted on continuing the revolution long after it was over. A Linux Weekly News article of the time said, "Richard Stallman is kind of like the eccentric uncle that everybody loves, or at least respects, but whose behavior means that nobody wants to take him anywhere. It seems quite possible that he will fade into irrelevance and obscurity." To be fair, they went on to say that would be a bad thing. The consensus was that Open Source had outgrown both Stallman and the GPL.

Now we have SCO mounting the most serious challenge yet to the open source movement. And what has proved the movement's trump card in dealing with these brigands? The GPL. With the GPL, we don't need to depend on IBM's fleet of lawyers. We don't need to seek help from the FTC or the SEC. We don't need relief from lawmakers. All of these things will likely have a positive effect on the outcome of the case, but we don't need them to survive this challenge. Some columnists have asked, "What if SCO is right?" And the answer, besides reiterating that they aren't, is that it doesn't matter. SCO can be squashed with the GPL alone. It has functioned exactly as Stallman intended it to.

Our industry has been conditioned to view all relationships as a zero-sum game, in which every transaction has exactly one winner and one loser. This peculiar view reflects the take-no-prisoners attitude of successful companies like Microsoft and Oracle. However, this attitude reflects the personalities of those firms' founders more than any inherent condition of the software market. I think that Steve Ballmer and Bill Gates were sincere when they described the GPL variously as "cancer," "viral," and even "communistic." If the GPL is an alternative to their licensing model, then it must somehow be designed to hurt them. If someone else is winning, then they must be losing, because that (in their minds) is how the business works. The notion that GNU/Linux was created with no regard for them or their market share is simply an impossible idea for these very clever men to comprehend.

But the GPL is not the expression of a zero-sum agenda. It is designed to protect the inherent value of software separately from the economic value. It measures inherent value by the greatest utility rather than the greatest realizable revenue. Realistically, very few software programs are "worth" much money.

The artificial scarcity model of value creation doesn't work well in the digital universe. In the world of tangible goods, if I make TVs, then I have TVs, and if you need one, it will suit you to buy one. The relative scarcity of free TVs means that supply/demand will work to both our benefits. I have an incentive to make them, and you have a source to obtain them. If I could pick up a decent TV on any street corner for free, however, it would be a different story. But where would a large supply of free TVs come from? It costs money to make each unit. That cost may decline with greater volumes of production, but it never reaches zero. TVs are also difficult to make, I can't realistically turn them out in my basement.

With software or digitally encoded content (e.g. MP3 and DVIX), the incremental cost of producing a new unit is zero. In addition, technology like broadband and desktop PCs have given the power of creation to almost everyone. Therefore, in the long run an efficient market will crowd out middlemen that do not add value. To protect some scarcity, and therefore value, companies have relied on various means to thwart market efficiency. They rely on copy-protection, licensing agreements, copyright law, and other means to keep the effective cost of incremental units high and to keep end-users from unauthorized creation. Otherwise, there would be no scarcity and thus no value to their roles as creators or distributors.

The GPL is based on a world-view that transcends the scarcity model for economic value. It recognizes that content cannot be kept scarce. All schemes to artificially protect pricing that are based on scarcity must ultimately fail, not because people are greedy and want something for nothing, but because market efficiency will eventually prevail. This is precisely what we are now seeing in the music business. Music companies do not offer value sufficient to protect their pricing models. They are now desperately trying to prevent the collapse of those models by raising the cost of incremental reproduction -- meaning that the cost to me of reproducing their content might include a ruinous lawsuit. There is probably a Nobel Prize in Economics for the first researcher to properly explain this.

The GPL places value not on scarcity, but on ubiquity. The more useful something is, the more used it becomes and therefore the more valuable -- the so-called network effect. Economic value is derived from the content, not inherent in it. Consider the example of Bell Canada, which expanded the number of pay phones dramatically in the 1980s. Their ubiquity probably held up the penetration of cell phones by many years. It was cheaper and easier to use a pay phone than to have a cell phone. That remained true as long as cell phones were expensive and pay phones were everywhere one wanted to make a call.

This model of value is inherently more efficient for some kinds of goods than the scarcity model. It's particularly true in software where there are no barriers to entry. The GPL ensures that anyone who wants to use that software can do so with great flexibility. But most importantly, it prevents anyone from creating artificial (and inefficient) scarcity. Nobody can take my software and erect a fence around it for their own economic gain. I can't be exploited.

This does not mean money can't be made. There is widespread confusion about the GPL and economic value. That confusion is based on the word free. The GPL addresses free-as-in-speech, and it has nothing to say about free-as-in-beer. Anyone can sell GPLed software for money if they want to. Red Hat does every day. However, the requirement to publish the source code effectively removes artificial scarcity from the value chain. To successfully charge people for GPLed code, a business must add some tangible value for the customer. That can be as simple as burning some RPMs on a disk for those without broadband, or as complex as providing turnkey clusters. The GPL does not demand that anyone give their work away for free. This fundamental misunderstanding has been willingly perpetuated by those whose businesses depend on the scarcity model.

Likewise, there is an oft-stated myth that having looked at GPL code one cannot write non-GPLed code that does the same thing without making that code open as well. This is a nonsensical distinction. If you look at any code and substantially copy it for a new application, you are violating copyright. The GPL is actually more liberal in this regard than traditional licenses. It provides a mechanism to use existing code provided your new code meets some conditions. Try that with Windows source code. You can't copy or utilize it in any way whatsoever without Microsoft's express permission. And they don't often give it.

My company is going to dual-license our development libraries under a scheme similar to that of MySQL. We can give the libraries away to encourage people to try them. If they want to continue using them, they can do so at no cost. And why should most people pay? They are not going to realize great sums of money from the software they create. So the economic value of our libraries to them is nothing. But for those that will realize revenue from their creations, there are value propositions worth paying for. We can provide them service, training, printed manuals, consulting, and a licensing scheme whereby they can keep their derivative code closed. Even these customers don't have to pay us for a closed license if they don't want to. The GPL provides that one needs to make source code available for applications that are distributed. If the user creates an application that they don't distribute, they don't need to show anyone the code.

This balance of potentially competing interests is the brilliance of the GPL. It recognizes that content has no inherent value. Nobody is going to pay to look at my code because that code is so great. It is only valuable to the extent it is useful. And that value may or may not be economic. The GPL encompasses all of these variables.

It has taken me years to understand the true subtleties of the GPL. I now see it as one of the visionary ideas of the digital revolution. It's hard to tell what the future will regard as important in our age, but I am guessing the GPL will be a term in common use long after people forget who Darl McBride is or what RIAA stands for.

John O'Sullivan takes care of sales, marketing and product management
at Hotsprings Inc., a Toronto
company creating open source applications and development tools. Hotsprings
technology was acquired from Hotline Communications, developer of Hotline
Connect and a pioneer of P2P applications.

Click Here!