The recent article on ZDNet UK by Marty Goetz — the man granted the first software patent in 1965 — sparked a robust debate. Here, he expands the case for patenting true inventions in software.
There were almost 300 comments from the UK and US on ZDNet UK and on Slashdot in response to my article, Should software be patentable? That's the wrong question to ask. Rather than try to answer them individually, I want first to expand on my explanation of why software should be patentable subject matter, before I respond to the comments, many of which I agree with.
Anti-patent zealots say all software is not patentable because it consists of ideas, mental processes or mathematics. If this argument is allowed to prevail, the result would be that true inventions that contain software as the preferred implementation would not be patentable.
Fact 1. The anti-patent zealots' definition of software is factually incorrect. Software is a general term that describes computer programs stored within a computer. A computer program is made of machine instructions that, when run on a computer, produce a desired result. Nothing more, nothing less.
The stored program had its roots in the 1930s and 1940s and was the core of the invention of the computer.
Of course, a stored program may contain an idea. For example, that idea might be a new way to compress or encrypt data, or simulate a mental process such as voice recognition, or do complex mathematical computations such as weather forecasting. But those ideas do not in any way change the fact that a computer program is a discrete entity that is a part or component of a machine.
Fact 2. To obtain a patent under the US patent law the application must contain the preferred implementation of the invention in a manner sufficiently clear and complete for the invention to be carried out by a person skilled in the art. For software, the disclosure would be flowcharts, other diagrams, graphics and text.
So here's another way to debate this important issue. Should an invention in any industry that uses X as its preferred implementation be patentable in all cases, except when X equals a stored program or equals a stored program plus Y, where Y is another preferred implementation?
There are 17 major industries in Wikipedia's Outline of Industry. It is almost a certainty that all these major industries actively file for patents. And it is highly probable that many of the patents contain a stored computer program as part of the preferred implementation disclosure. The computer industry is shown as one of those 17 industries and the software industry is shown as part of it.
So should all these industries not have the benefit of filing for patents if the invention contains a stored program, all or in part? Or should inventions that only use a stored program as their preferred implementation be disqualified?
My response to anti-software patent comments
Some readers may have misunderstood the thrust of my article. I was not arguing that the current US Patent System — and how it is interpreted by the US Patent and Trademark Office (PTO) and the US courts — is effective for software-related inventions.
Early in my article I stated: "What these zealots should be arguing is that many software patents issued by the PTO, including most business-method patents, should never have been issued. With that I heartily agree. In fact the US courts are beginning to recognise that certain classes of software patents are invalid."
The intent of my article was to argue that under current US patent law a true invention should not be...