The History of Software Patents:
From Benson and Diehr to State Street and Bilski
Executive Summary:
It was not always clear that computer software was patentable in the United States. While it is clearly patentable at the present time, it is not clear that this will always be the case. The following description provides a brief history of software patents in the United States.
- 1960s-70s: No Protection if Invention Used a Calculation Made by a Computer;
- 1980s: The Supreme Court Says Some Computerized Inventions are Patentable;
- 1990s: The Federal Circuit Says Almost All Software is Patentable.
- 2000s: Courts Start Pulling Back.
For a discussion on the current state of software patents under the Supreme Court's Bilski v. Kappos decision, see the Bitlaw discussion of "Are Software and Business Methods Still Patentable after the Bilski Decision?"
1960s-1970s:
The U.S. Patent and Trademark Office historically has been reluctant to grant patents on inventions relating to computer software. In the 1960s, the P.T.O. avoided granting any patent if the invention utilized a calculation made by a computer. In 1968, the Office created formal guidelines for computer related inventions to formalize this reluctance. These guidelines stated that a computer program, whether claimed as an apparatus or as a process, was unpatentable. Under these guidelines, an invention relating to a programmed computer could be patentable only if the computer were combined with other, nonobvious elements to produce a physical result. The P.T.O. viewed computer programs and inventions containing or relating to computer programs as unpatentable mental steps, and not patentable processes or machines. As such, software related inventions were considered by the P.T.O. at that time to be non-statutory (see the BitLaw discussion on patent requirements for further information on the requirement that inventions be statutory).
The 1968 P.T.O. guidelines prohibiting most all patents on computer software were routinely challenged by patent applicants. In several decisions, the Court of Customs and Patent Appeals (the CCPA) rejected various aspects of the guidelines, at one point stating that since a new program turns a computer into a new machine, all claims that include both the computer and software should be considered statutory subject matter. However, the Supreme Court was not as willing as the CCPA to declare software to be patentable subject matter.
Gottschalk v. Benson
In the 1970s, the Supreme Court twice examined whether inventions containing computer software were patentable. Both times, the Supreme Court answered in the negative. In the 1972 case of Gottschalk v. Benson, the Supreme Court struggled with whether an algorithm to convert binary-coded decimal numbers into true binary numbers was considered patentable. The Court felt that a patent on this concept would pre-empt the entire mathematical algorithm. Since mathematics could be considered an abstract idea, and abstract ideas are not patentable, the Supreme Court held that the algorithm in question is not patentable.
Parker v. Flook
In Parker v. Flook, the Supreme Court examined whether a method for updating an alarm limit (used to signal abnormal conditions) in a catalytic conversion process was patentable. The only difference between the prior art and the invention was the algorithm that calculated the new alarm limit. The Court held that this was not patentable even though an additional step was included in the claim beyond merely the calculation step. The Court explicitly rejected the notion that "post-solution activity [alone]... can transform an unpatentable principle into a patentable process." Specifically, the court held that the invention could not be patented "not because it contains a mathematical algorithm as one component, but because once that algorithm is assumed to be within the prior art, the application, considered as a whole, contains no patentable invention."
1980s
Diamond v. Diehr
In the 1980s, the Supreme Court forced the P.T.O. to change its position. The 1981 case of Diamond v. Diehr provided the first instance in which the U.S. Supreme Court ordered the P.T.O. to grant a patent on an invention even though computer software was utilized. In that case, the invention related to a method for determining how rubber should be heated in order to be best "cured." The invention utilized a computer to calculate and control the heating times for the rubber. However, the invention (as defined by the claims) included not only the computer program, but also included steps relating to heating rubber, and removing the rubber from the heat. The Supreme Court stated that in this case, the invention was not merely a mathematical algorithm, but was a process for molding rubber, and hence was patentable. This was true even though the only new feature of this invention was the timing process controlled by the computer.
After Diamond v. Diehr
After 1981, the P.T.O. and inventors were left trying to determine when an invention was merely a mathematical algorithm, and when it was in fact a patentable invention that simply contained a mathematical algorithm. Although lower courts attempted to set forth this distinction in a clear manner, the resulting opinions were generally quite confused. This confusion originated in the inherent conflict between the Diamond v. Diehr and the Parker v. Flook Supreme Court decisions, since the majority decisions in these two cases appear to contradict each other in fundamental ways (see the Stevens dissent in Diamond v. Diehr for a discussion of this conflict). What was clear was that the patentability of a software related invention depended heavily on the claims crafted by the patent attorney.
1990s:
In the early 1990s, the Federal Circuit (the highest court for patent matters other than the Supreme Court) tried to clarify when a software related invention was patentable. The court stated that the invention as a whole should be examined. Is the invention in actuality only a mathematical algorithm, such as a computer program designed to convert binary-coded decimal numbers into binary numbers? If so, then the invention is unpatentable. However, if the invention utilizes the computer to manipulate numbers that represent concrete, real world values (such as a program that interprets electrocardiograph signals to predict arrhythmia or a program that analyzes seismic measurements), then the invention is a process relating to those real world concepts and is patentable.
In re Lowry
In 1994, the Federal Circuit considered the patentability of a "memory" containing a data structure. The claim was originally rejected as nonstatutory by the U.S. Patent and Trademark Office. The PTO Board of Patent Appeals and Interferences determined that these claims were statutory, but then used the "printed matter exception" to deny any patentable weight to the data structures when determining whether the claims were patentable under Section 102 and 103. The Federal Circuit reversed this decision, holding that the printed matter exception applies only to "printed lines or characters, useful and intelligible only to the human mind." Since data structures (and computer software) are not processed by the mind but by a machine, the exception does not apply. Consequently, a memory containing a data structure should be considered patentable subject matter and the particular data structure claimed should be considered in determining whether the invention is new and non-obvious.
State Street Bank & Trust v. Signature Financial Group
In 1998, the Federal Circuit issued its State Street Bank & Trust v. Signature Financial Group decision, which further clarified the patentability of computer software in the United States. In this case, Signature Financial had obtained a patent on a "Hub and Spoke" method of running mutual funds. In this method, several mutual funds (or "spokes") pool their investment assets into a single investment portfolio (the "hub"). Software then determines the value of each fund based upon a percentage ownership of each of the assets in the hub portfolio. This information is tracked on a daily basis, and is used to track fund share pricing and tax accountability. State Street Bank asked the court to declare this invention to be unpatentable as a mere mathematical algorithm or as a business method. The Federal Circuit rejected the arguments of State Street Bank, and instead upheld the patent by explicitly stating that business methods can form patentable subject matter. The court emphasized that software or other processes that yield a useful, concrete and tangible result should be considered patentable.
After the State Street decision, there is little doubt that computer software and data structures can be considered patentable subject matter in the United States. Certain steps are required to ensure that the software meets the examination guidelines of the USPTO, such as ensuring that the software or data structure is operated upon by a computer processor, or is stored on a computer readable media. However, these requirements are easy to meet when drafting the patent claims, and therefore are not a significant impediment to patentability (see the next BitLaw section for further information).
2000s:
In re Comiskey
In 2007, the Federal Circuit held that some of claims relating to mandatory arbitration in certain legal contracts were unpatentable subject matter under Section 101 of the Patent Act. The claims that were directed toward unpatentable subject matter related entirely to mental processes (i.e., they covered only the use of human intelligence). Other claims described the same process but further included some type of physical device (such as a "module" or a computer were considered to be statutory subject matter because the claims referred to a machine, manufacturer, or a composition of matter. One of the most interesting parts of this case is the statement that "[t]he routine addition of modern electronics to an otherwise unpatentable invention typically creates a prima facie case of obviousness."This statement echoes similar language found in Flook. If considered good law, this statement would impact the ability to patent any software in the United States.
In re Bilski
On October 30, 2008, the Federal Circuit came down with their In re Bilski decision. Many expected this decision to alter the scope of patentable subject matter in the United States, especially as it relates to business methods and computer software. Instead, the majority ruled on the patentability of processes where the process steps are not necessarily performed on a computer. The majority decision has set forth a single test for determining the patentability of processes. This test holds that a process is patentable if "(1) it is tied to a particular machine or apparatus, or (2) it transforms a particular article into a different state or thing." This machine-or-transformation test applies to process claims, but it is not clear whether the decision should alter the analysis of apparatus claims (such as computer implemented software claims).
The Supreme court rejected the Federal Circuit's holding in In re Bilski that the machine-or-transformation test is the sole test to determine whether a particular process constitutes patent-eligible subject mater. Instead the test should be viewed as "a clue" to this analysis. The majority decided that the Bilski invention was not patent-eligible subject matter because it was an attempt to preempt an abstract idea. It is not clear from this decision what test or analysis should be used going forward to analyze whether a particular process is merely an abstract idea or not, although many observers and the U.S. Patent Office (see this June 2010 memorandum) believe that the machine-or-transformation test has become one test for whether the process preempts an abstract idea. For more analysis on this decision, see the Bitlaw discussion of "Are Software and Business Methods Still Patentable after the Bilski Decision?"