Prof. Randal C. Picker, University of Chicago Law School
February 25, 2015

Software patents are extraordinarily controversial.  In 2011, the White House was asked through its online petition set-up to direct the patent office to cease issuing software patents.  The White House declined, noting that Congress was in charge of the general scope of patent eligibility.  Part of the controversy is undoubtedly driven by the explosion in the number of software patents granted.  But it is worth returning to the roots of the software patent and attempting to understand why Marty Goetz pursued a patent for his sorting system in 1965.

Start with the state of software patents.  The U.S. Government Accountability Office issued a report in August 2013 on the state of patent quality in the United States, and Figure 1 in the report is eye-catching:

In 20 years – from 1991 to 2011 – software-related patents exploded and, by 2011, accounted for more than half of all new patent grants.  That undoubtedly reflected a change in underlying economic activity – more and more economic activity ran through computers and their software – but also reflected changes in legal doctrine, including new patentability of business-method patents.

Computer software has always been something of an awkward match for intellectual property law.  The core economic characteristics of software match reasonably well with the kind of things that we bring into IP law.  Making the first copy of new software can be quite expensive, but once created, it is easy to copy.  That matches the economic structure of many of the works that copyright law embraces (novels, movies, symphonies).  At the same time, though, copyright law protects expression of ideas and not ideas themselves and therefore copyright protection does not extend to “any idea, procedure, process, system, method of operation, concept, principle, or discovery.”  Computer software, as such, is a recipe to achieve an end and seems meaningfully different than the types of expressive works at the heart of copyright.  (And, lest there be any doubt about the still-uncertain status of the match between copyright and computer software, consider the recent case between Google and Oracle.)

As to patents, U.S. patent law doesn’t allow abstract ideas to be patented.  As Justice Thomas noted in 2014 in his opinion for the U.S. Supreme Court in Alice, this is a long-standing principle of U.S. patent law, but of course the trick is in the implementation of that idea.  Alice itself didn’t address software patents; the patent bar and the U.S. Patent and Trademark Office are now wrestling with how to implement this system.  (The PTO issued interim guidelines in December.)

But these are not new issues.  On May 8, 1964, the New York Times ran a story about an enterprising second-year law student at Columbia.  The student, John F. Banzhaf III, had received what was thought to be the first copyright registration of a computer program.   In the article, an unnamed lawyer rejected the strategy on the ground that trade secrets offered superior protection, since copyright registration of the work would inevitably disclose all of the ideas in the software – and that was, in the lawyer’s view, the thing to be protected.

And then there are the origins of software patents and Marty Goetz.  PBS has a terrific video short series on inventors and, in four minutes or so, you can get Goetz’s story straight from him.  But the short version of the story is that in the mid-1960s, Goetz didn’t see how he could sell software in a world in which IBM was a mainframe monopolist.  IBM was fully vertically integrated and didn’t charge separate prices for hardware, software, and services.  That meant that there was no (marginal) price for software.  It was “free” in the sense that IBM would supply all the software a firm wanted and there were no separate charges for the software.

That made life very difficult for independent software developers.  There was a market for services and tailored software but not much of a market for freestanding, for-fee software.  Goetz saw two solutions to this.  One was antitrust, and the U.S. government would eventually bring a complaint against IBM in 1969 seeking to force IBM to unbundle its business.  The government would eventually – 1982 – dismiss that case, but IBM did unbundle its sales contemporaneously with the running of the antitrust case.

But the second solution was to have an actual property right in the software invention, as that would block IBM from just copying the idea into its free software.  Goetz applied for his patent on April 9, 1965, and three years later, the first software patent, U.S. Patent No. 3,380,029 for a sorting system, was granted.  The truth is that the idea of a software patent wasn’t new.  IBM itself had considered the possibility of software patents as early as 1956, but had seemingly concluded in its internal vetting process that a software patent wasn’t appropriate.

But of course IBM didn’t need to propertize its software inventions when it knew that it could recover the benefits of those inventions through hardware charges, so IBM didn’t have a particularly good reason to try to propertize computer software during the days of IBM bundling.  Once IBM had unbundled, it changed its approach to software propertization.  Between 1964 and January 1, 1977, 1,205 computer programs were registered with the U.S. copyright office and IBM and Burroughs accounted for 971 of those (here at p. 34).  IBM actively patented and filed an amicus brief in the Alice case where it made clear its view that software was clearly patentable.

This is a story of “monopoly” substitution, though I use the word monopoly with trepidation.  The software patent emerged as a response to monopoly – or at least perceived monopoly as IBM disputed that characterization – and emerged as a tool to contest, at least partially, IBM’s bundled position on mainframe computers.  I have no idea really as to the right number of software patents, but the Goetz story suggests that the answer is not zero.