Software as Intellectual Property

In the Agrarian Age, arable land was the primary source of wealth. In the more recent Industrial Age, capital was the primary source for the production of wealth. In the Information Age, the source of wealth is information, or knowledge. It is packaged in the form of either static databases or computer-based files that are accessed by software. Increasingly, information will be packaged as computer simulations in which very complex scientific and engineering relationships can be demonstrated in an operational way (see Sidebar 12.4). There is a fundamental difference, however, in information as a source of wealth as opposed to arable land or capital, either of which are committed resources. I cannot use my 800 acres to both grow soybeans and lease to another farmer to grow corn. I cannot invest my $10,000 in both GE shares and T-notes. But if I understand some novel software-based Information Age technology, I can explain it to you, and then we will both know it and can use it as a source of wealth. Dr. Richard Stallman, a Macarthur Fellow at MIT, and his colleagues have noted this fundamental difference and argue for all software to be "open-source" like Linux rather than proprietary like Windows XP. The vehicle they have devised for this is the copyleft (as opposed to the copyright). Stallman's own software, such as GNU (recursively defined as "GNU is Not UNIX"), and his remarkable editor, Emacs, are available only under a copyleft in which the user warrants that he or she will provide the software without cost to others and inform Stallman of any bugs or errors in the code.

But software truly is intellectual property in the age of computers. It can be protected in order of relative security, as trade secrets, by copyrights, or by patenting it. Actually, the Trade Secrets Act is strong but has a trapdoor. If the software engineer who knows your secret tries to make money using it on his own, you can get a permanent injunction to prevent him from doing so. However, if he retaliates by going to a developer's conference and disclosing your secret technology, it is now no longer a secret and therefore is no longer protected by the Act. The copyright appears to be the logical protection mechanism for something like software, which is created and transferred as a written document. In fact, integrated circuit designers have found that copyrighting their masks as images provides more effective protection than patenting the chip's architecture as a mechanism. Clearly a software-based algorithm or process is a mechanism. In fact, some of the pioneers of software patenting, such as John Rooney, patent counsel of Univac in the 1960s and later chief patent counsel at Medtronic, got the first software patents by a back door process using this fact. Rooney, in his Socratic questioning style, would ask the designer of a novel software technology at Univac if it could also be implemented as microprogrammed firmware. The answer was always yes. Then he would ask if it could be implemented as firmware, couldn't it also be implemented in hardware as combinatorial logic or a programmed logic array (PLA). Again the answer was yes, but the software engineer would admit that he didn't know how to do that. So then Rooney would offer to provide a hardware designer to assist as co-inventor by designing the hardware implementation. He would then apply for a patent on the hardware as the preferred implementation, and the firmware and software as alternative implementations. Indirect, perhaps, but protected by patent law they were.

In the 1980s it became possible to patent software directly after the U.S. Patent Office was able to recruit examiners who could deal with it. Many of the early software patents were quite solid, but more recent patents and their failure to stand up under challenge in the courtroom have cast doubt on the use of patents to protect intellectual property cast as software. In a series of two articles in the IEEE Spectrum, Ben Klemens has argued that no clear boundary between software and mathematics exists and that software is better protected by copyrighting it rather than patenting it.^[15] Dr. Klemens is a guest scholar at the Brookings Institution working on the forthcoming book Math You Can't Use: Patents, Copyright, and Software, to be published by the Brookings Institution Press. His thesis is well illustrated by numerous examples of mathematical formulas and numeric methods that have been granted patents as the patent office pendulum has swung too far the wrong way. Although mathematical algorithms were never patentable as scientific principles, what happened was recent developments in mathematical methods that are totally dependent on high-performance computers for their execution. This has opened a back door to patenting them due to blurring the distinction between mathematics and machinery. To a mathematician, Dr. Klemens' argument is simple: a modern computer is equivalent to a Turing or state machine; all modern programming languages can turn any computer into a state machine; and Church's Lambda-calculus can express any mathematical procedure, and likewise any software process, as a set of purely mathematical Lambda-calculus expressions. Therefore, it is impossible to distinguish between software and mathematics.^[16] Incidentally, this was the major theme of the late Prof. Edsker Dykstra's work and is the basis of functional programming and any other denotative programming technology.

If software is formally equivalent to mathematics and mathematics cannot be patented, every software patent is a legal contradiction. In his second article (and presumably in his forthcoming book), Dr. Klemens argues that software patents are not viable and that the system will collapse under its own weight because it will soon prove to be unworkable. There is no way of reconciling a highly decentralized industry such as software, dependent on massive independent invention, with a law that makes independent invention a liability.^[17] So, what is the alternative? His answer is clearly the copyright. Although the copyright does not offer a patent's monopoly protection, it provides adequate protection for software as intellectual property and is much simpler and much less expensive.

Sidebar 12.4: A Picture Is Worth...

Some years ago, when one of the authors was director of the Minnesota Supercomputer Center, he hosted a National Science Foundation (NSF) sponsored conference on "The Role of Supercomputers in Observational Astronomy." The last speaker in the plenary session was Professor Joan Centrella of Drexel University, who showed her famous eight-minute video on the creation of the universe. The author was standing at the back of the auditorium at the University Radisson Hotel, waiting to herd the attendees onto waiting buses to go to a party at the Supercomputer Center. The session had run a half-hour over. A "Minnesota swede" janitor was leaning on his broom beside me at the rear door, waiting to clean up and rearrange the room for another event. As Professor Centrella was receiving her well-deserved standing ovation, I leaned over and asked the janitor what he thought of the presentation. He replied: "Vell, dat vas pruddy gud. I can't vait to get home and tell my vife how the universe vas created." This is the power of a visual simulation of a complex natural event (or, in this case, the Big Bang, a "supernatural" event, since it cannot be described by the laws of physics as we know them). A visual simulation communicates with everyone, whereas the relativistic or quantum equations it is modeled with can communicate with only a few hundred specialists. The future of scientific communication lies in the use of simulation models like these, even for specialists.