Open Source Software Model

by Russell Nelson

Open Source software is all the rage these days. Big companies and small companies alike are either releasing their software as Open Source or considering it. On the forefront of their minds is “Yes, it's more reliable; yes, it meets more users' needs. But how do we make money on it?”

First, we must be clear that no industry is sacrosanct in any economy. Almost all companies rely on special protection from the government, called intellectual property rights. These rights, covering both copyright and patent laws, exist because they are thought to be economically efficient—that is, they create more than they destroy. Should that prove wrong—that protecting computer software as intellectual property actually makes society poorer—then the computer industry will just have to rely solely on contract law. If society is better off with all software as Open Source, then that is how it will be, even if it makes computer programmers worse off financially.

Open Source software need not impoverish all programmers—indeed, there is good evidence that the deleterious effects of switching to all Open Source software would be only short term. Remember the Luddites? The dislocations in the weaving industry were only temporary. Forty years later, the industry was employing just as many weavers and, of course, weaving a greater amount of cloth.

If you want to see this model in action, look through this issue of Linux Journal. You'll see ad after ad, most of them for Open Source software. The way to make money in a free market is to solve someone's problem and take a cut of the benefit they receive. There are many different ways to do so, as reading the ads proves.

Second, I have been assuming that Open Source software is applicable to all types of software. This is not clearly true or false. What is definitely true is that an operating system can be Open Source and be successful, as all the readers of this magazine know. Perhaps other types of software are not as amenable to Open Source distribution. If this is the case, then the creators of such Open Source software will either not exist in the first place or will go out of business. For this type of software, the proprietary model would work best.

People have gotten rich from Open Source software, and still are. As long as you have an industry where people can get rich, you've got plenty of incentive to enter the industry.

The big question being discussed on the Free Software Business mailing list ( is: “How do you create a piece of software which requires a large up-front investment of human energy?”

Companies such as Netscape or Corel can easily afford to free a major piece of code. The software has been proprietary for a number of years and has recouped its initial investment. The Crynwr Packet Driver Collection started small and was initially useful with only one driver in it, so it doesn't fit the open model. The first released version of Linux was a real UNIX kernel with a file system and serial port drivers. To the extent that it needed an up-front investment, Linus has been paid back for it.

No, recovering the investment is a tough job without intellectual property laws. Problem is, they're needed precisely for only those tough jobs. More than any other type of intellectual property, creation of most computer software involves very little capital costs. Invoking the coercion of the state (even as far as stealing a person's thoughts, when independent creation occurs) is less and less appropriate. Software creation is cheaper now than when PCs were new. The areas where Open Software cannot succeed seem smaller and smaller.

Russell Nelson has been developing Open Source software long before the term existed. His first freed software, Freemacs, came out 15 years ago. He was the progenitor of the Clarkson Packet Driver Collection, the first freed software to be recognized by PC Magazine's Technical Excellence awards. Currently, he works for Crynwr Software, founded in 1991, an Open Source support firm. He can be reached at

Load Disqus comments