Patents: Where's the Invention?

Hardly a week goes by without the controversial issue of software patenting hitting the news. Particularly controversial are the business-method software patent filings that are inundating the U.S. Patent and Trademark Office. Patent attorneys are characterizing obvious and normal business methods of processing information as "machine systems" and describing the logic for automating a manual or semi-manual business method as an invention. And the patent office continues to accept such applications.

Meanwhile, the U.S. courts are trying to cope with, as well as clarify, the plethora of software patents that are in litigation. It's a real mess that seems to be getting worse. Although nearly 40 years have passed since I was granted the first software patent, the nature of software is still being debated around the world.

Many organizations are fighting for or against the patenting of all software, while others are solely for or against the patenting of business methods. As more and more business methods are computerized, it is clear to many of us in the computer field that there are no inventions involved. What constitutes an "invention" is key to this discussion.

The patent office and the Supreme Court have never indicated that a computer program is the subject matter that can be patented. Patents can be granted to processes, machines or methods when they rise to the definition of an "invention" in patent law. While the invention may be implemented in a computer program, it is not the program that is patented. Computer programs themselves are protected from unauthorized copying only by copyright law.

An invention, in lay terms, can be a novel device, material or technique that is new, inventive and useful. It has been well established that machines, including computer hardware, contain patentable subject matter. As for software, it is well recognized that whatever can be designed in hardware circuitry can be developed in a computer program to perform the same functions. The choice is a purely economic one that mainly has to do with cost, speed and flexibility. Patent applications normally show the preferred implementation, and the patent must disclose the invention adequately for one skilled in the art. But the disclosure could be in the form of circuitry for a hardware implementation, a flow chart for a software implementation or a combination of both.

In 1981, the Supreme Court ruled in Diamond v. Diehr that processes were patentable, though it left to the patent office to decide whether a process was "novel" and "nonobvious." The court's emphasis was on whether there was an invention, not on how the invention was implemented.

The patent office shortly thereafter came out with its "PTO Guidelines on Computer Inventions," stating in the opening paragraph, "The U.S. Supreme Court decision in Diamond v. Diehr significantly affects an examiner's analysis under 35 U.S.C. 101 of patent applications involving mathematical equations, mathematical algorithms and computer programs." Over time, however, the patent office started accepting business-method software that was described as a machine process solely on the basis that there was no prior art for the business method described, and not because there was an invention embedded in that machine process.

Since then, the patent office has issued patents on business-method software, such as Amazon.com's "one click" process, with little regard for whether they met the test for an invention. The surge in such patents arose from a 1998 U.S. appeals court decision agreeing with the patent office that State Street Bank had a valid patent on a method of doing business for pooling the assets of mutual funds. Unfortunately, that case never went before the Supreme Court, which has yet to decide on business-method software patents. The closest it has come was in 1975, in Dann v. Johnson, but in that case, it made a narrow decision in rejecting the patent claim, stating that the patent did not meet the test of "unobviousness."

A nonpartisan government commission, similar to the Commission on New Technological Uses of Copyrighted Works in the 1970s, needs to be formed to provide strong recommendations to the patent office. The commission should examine what other countries have done regarding business methods. The European Patent Convention, for example, has ruled that anything that consists of "schemes, rules and methods for performing mental acts, playing games or doing business, and programs for computers" is not an invention and therefore not patentable. That would not only be a good starting point; it could also be the basis for a worldwide agreement on limiting patentability to true software-related inventions.

Besides being a founder and former president of Applied Data Research, Martin Goetz received the first U.S. software patent, in 1968.

FREE Computerworld Insider Guide: IT Certification Study Tips
Join the discussion
Be the first to comment on this article. Our Commenting Policies