Friday, November 6, 2009

Software Patents and Innovation

by Mark Chadurjian
Senior Counsel, IBM Software Group Intellectual Property Law

We've heard a lot lately about software patents. Some folks say software should not be patentable. Others say that any business method run on a computer should be patentable. Why all the controversy?

First, no one is arguing about the importance of software in today's economy. Simply put, software is the backbone of the information age. It has enabled the extension of computing power upward (into the cloud) and downward (into handheld devices). The growth of robust operating systems and middleware has enabled the proliferation of applications that support everything from complex airline reservation systems to music downloads.

Key to that growth has been the existence of a robust patent system. Patents protect innovators who take business risks in bringing new products into the marketplace. In fact, there's a synergistic relationship between the strength of the patent system and a vibrant economy that rewards innovation and risk taking. So clearly, the U.S. economy benefits from a patent system that recognizes innovation wherever it exists, be it in hardware or in software.

So why the controversy? A strong patent system strikes an appropriate balance between private incentives to innovation, public benefit from encouraging dissemination of ideas, and restrictions the resulting patent places on the public. Right now, things are a bit off kilter in the U.S. For example, we've suffered from the issuance of overly broad software-based patents for methods of doing business. Such patents, with their attendant lack of technical content, tend to skew the scales in favor of private benefit to the detriment of associated public benefit. The question of business method patentability is now being considered by the U.S. Supreme Court in the Bilski v Kappos case, which involves a dispute over whether a method for hedging commodities trading risk is patentable subject matter. More broadly, the US patent laws are in need of updating to deal with issues such as facilitating challenges to bad patents and dispensing with archaic rules regarding who is entitled to a patent if multiple parties claim to have invented the same thing...that's why patent reform is so important.

While the courts struggle with whether business methods should be patentable, in Bilski some argue that we should throw the baby out with the bathwater, and revise the law to say that all software inventions should be unpatentable. That would be a mistake, because most software-based inventions embody compelling technical advances that strike the appropriate balance between private incentives and public restrictions. Holding all those innovations to be unpatentable would disincent software development, just when we're seeing software truly emerge as a platform for major advances in technical innovation.


  1. I have yet to see a patent for software that included the source code. They just describe what any user would see in the real world. Without the source code, I would have to use my own idea on how to reproduce the outcome. Which is the antithesis of the idea of sharing with a patent. And if I come up with my own way of coding, it shows that it is obvious to one skilled in the art.

    Software should only be copyrightable, not patentable.

  2. So go ahead and register for your copyright. Your proof for saying software is not patentable is that "if I come up with my own way of coding, it shows that it is obvious to one skilled in the art". I guess that means there you will never have to worry a "flash of genius" requirement for your inventions. Another patent anarchist. I think I'll stick with the well reasoned and educated opinions of the author of this blog and his colleagues (e.g., Mr. Kappos). I don't believe we should have complete patent reform but maybe a few tweeks would be nice.

  3. "Key to that growth has been the existence of a robust patent system."

    Given that Microsoft wrote Windows 95 and had a revenue of $5.94 billion while they still had somewhere fewer than 77 patents, and given that the whole GNU/Linux operating system has been written with zero patents, I think history shows that growth happened without the robust nuisance system, not because of it.

    ...but I think a better example is when the HTML5 committee had to give up on recommending a video format because all possibilities were closed off by patents.

    The sooner the Supreme Court rids us of the software patent pest the better!

  4. The pressure within IBM (especially, perhaps, in Research Division, where I was) is very high; our performance evaluations were based in part on how many patent applications we filed. One result of that -- throughout the industry, not just in IBM -- has been a large number of patents that should never have been issued (not novel and/or not non-obvious), and many others that are fairly pointless (not useful).

    Once in a while, one runs into a software patent that's really a good one, and which clearly supports the view that being able to get patent protection for an innovative idea that took time, effort, and resources to develop is an important mechanism for fostering innovation. Absolutely. So it's clear to me that software patents should be allowed.

    What we need is enough expertise in the US PTO to vet software patents properly, and to reject the garbage that forms the vast majority of the software patents that are issued today.

    We also need the companies that are filing these patent applications to change their ethical approach to it, to do more internal filtering, and to stop filing patent applications that they should know are not useful, not novel, and/or not non-obvious.