Software patents and shorter Marty Goetz

Shorter Marty Goetz: Software is valuable, therefore it should be patentable.

Goetz was the recipient of the first U.S. software patent ever awarded. In this ZDnet article, he defends software patents, where his basic argument is that software deserves patent protection because software development is a billion-dollar industry. For several reasons, Goetz fails to make a persuasive argument for software patents.

To begin, why do we even have a patent system? The patent system gives the inventor of a useful device a limited right to use the government to protect against unauthorized copying. The benefit to society is that patents encourage innovation: people are more likely to invent if they know a competitor can’t free-ride on their efforts. The limited term of a patent ensures that today’s protections do not prevent tomorrow’s inventions.

Going back to Goetz’s article, in some sloppy sense, yes, patents are meant to protect things that are "valuable": They should cover socially desirable inventions that are unlikely to come about without protections against free-riding.

But things are not that simple when it comes to software patents. First, it is unclear whether patent protection would encourage innovation in the software industry. Programmers might write useful software even without patent protection.

Even if patents could spur software innovation, the system may be too difficult to administer. If writing a piece of software and reducing it to practice is easier than making a physical widget, software patents will add to the patent office backlog. If figuring out whether software is novel, useful and non-obvious is more difficult because of the abstract nature of programming, software patents will add to the court dockets.

Software patents may also hinder future innovation. A great weakness of software patents is the difficulty of due diligence: There is no easy way to determine if a program is infringing on earlier patents, and the risk of litigation may deter innovation.

These are just a few issues ignored by Goetz, though any argument for software patents should address them.

The closest thing Goetz has to an argument is his point that because "software and hardware circuitry are interchangeable", it makes no sense to allow patents for one and not the other. For example, there are many functions that can be implemented in both software and hardware, such as FFTs in cell phones. Goetz argues that because the same amount of effort goes into the software and hardware implementation, patents should be available for both.

What Goetz ignores is that a patent issued on a hardware implementation does not cover the mathematics behind it. The patent protects only the physical device from copying, not the abstract ideas that make it work. One benefit of limiting patent protection to a physical device is that it makes it easier to define the bounds of the invention. The clearer the boundaries, the less litigation.

A more subtle benefit of tying patent protection to a tangible product is that it increases the likelihood that the invention is truly novel. The more abstract the patent, the more likely the device was independently invented, and vice-versa. Reducing an idea to physical form requires additional effort that narrows the field of possible inventors.

In sum, Goetz doesn’t make a strong case for software patents. Of course, there are alternative ways to protect software from free-riding: Most notably, via trade secrets and copyright. Whatever approach is chosen, let’s hope it relies on a more compelling argument than "software is valuable, therefore it should be patentable".