Earlier I mentioned that I would explain why I sometimes put the term software patent in quotes. The reason is, quite simply, that the term is an oversimplification that is both too over-inclusive, and too under-inclusive to substantiate any productive dialog. The term is over-inclusive in that some of its users tend to conflate a patented invention implemented with software with the actual software itself. A patent protects novel products, machines, compositions and processes. A copyright protects the software itself. The term is too under-inclusive in that it arbitrarily divides otherwise equally unpatentable abstract ideas from one another based on whether or not they refer to computerized steps. (Yes, I absolutely agree that some patents describe software-implemented ideas that are too abstract to be patented, and we’ll get into that below the jump, along with a proposed test based on the Supreme Court’s guidance in Bilski v Kappos).
Simply putting a modifier “software” in front of the word “patent” might be used to indicate technology relevant to the software industry (much in the same way that you would refer to an automobile patent, a pharmaceutical patent, or a medical device patent). However, in that instance, there’s nothing intrinsically different about a “software patent” that makes it any more or less objectionable than patents on car parts, drugs or bone screws. Admittedly, other writers have objected, saying that no one gets a patent on the idea of a bone screw, but rather on a specific bone screw that is different from all the others before it. Get ready for some startling news. In the computer arts, the same thing is true. No one gets a patent on the idea of wireless email, syncing calendars, or object-relational mapping any more than one can get a patent on the idea of a bone screw. However, a patent might claim a series of steps that, out of necessity, must be performed in order to sync calendars or map objects to databases, just as a patent might claim a structural combination found in every bone screw. Proving invalidity of patents like these would only require evidence that someone performed those steps, or built the structure, prior to the claimed invention.
For some, software is simply “too abstract” to be patentable. Over at Groklaw, an essay explains why software is abstract. Dana Blankenhorn refers to his “mousetrap test” and proclaims that “[s]oftware patents, by their nature, fail” the test. Meanwhile, Rob Tiller asks the more appropriate question: “Is software too abstract to be patented?” Those who oppose patents on software implemented inventions hear the phrase “abstract idea” and proclaim, “A-ha!” Software, by its nature, is abstract and, the argument goes, shouldn’t be patentable for that reason. However, there’s a difference between an abstraction and an abstract idea. While software is an example of the former, the way operations are described may or may not constitute an abstract idea.
As I mentioned above, it is possible to describe the operation of software in terms too abstract to be patented. Let’s look at two examples, one patentable, and one likely unpatentable. An example of a patentable claim (upheld under the more rigid machine or transformation test) is found in US Patent 6,937,187:
1. A method, comprising:
estimating a plurality of states associated with a satellite signal receiver, the plurality of states including a time tag error state, the time tag error state relating a local time associated with said satellite signal receiver and an absolute time associated with signals from a plurality of satellites; and
forming a dynamic model relating the plurality of states, the dynamic model operative to compute position of the satellite signal receiver.
The claim of the ‘187 Patent is a pure method claim, and would undoubtedly be considered by many to be a “software patent.” However, this claim passes even the more rigid patentability test, despite the fact that its likely implementation is through software. The reason, according to the court of appeals in SiRF Technology v. ITC, is the recitation in the claims of “a satellite signal receiver” and the claimed result of the method, which is “to compute position of the satellite signal receiver.” The claimed method is not merely processing generic signals. Instead, the method relies on specific types of signals from a specific type of device.
In contrast, consider treatment of the following claim by the USPTO Board of Patent Appeals in this opinion, finding the following claim unpatentable:
33. A computerized method for measuring a consumer’s perception of a commercial entity’s brand equity, logo, trademark, tradename, tag line, product name and the like, comprising the steps of:
providing a present interactive advertising message to the consumer;
allowing the consumer to interact with the present interactive advertising message;
gathering data associated with the consumer’s interactions with the present interactive advertising message;
generating a statistical report from the data; and
providing the statistical report to the commercial entity.
In upholding the Examiner’s reasoning, the board agreed that this claim only presents the generic concepts of interacting with, gathering and outputting data. From the language of the claim itself, there is only recitation of “gathering data” associated interaction with a “message,” neither of which relies on any specific machine or structure. While the claim is specifically tied to the field of “brand equity” (whatever that means), there is no discussion in this claim of what kind of data to gather, or how that data should be processed into a report. The board correctly notes that this claim would “wholly pre-empt” interaction, gathering and outputting of data for measuring “brand equity”, and would amount to a patent on the concept of the same.
Of course, machine or transformation is not the only test for patentability. Instead, the Supreme Court instructed us in Bilski to look at the “guideposts” of Gottschalk v. Benson, Parker v. Flook, and Diamond v. Diehr. I plan to dive deeper into this analysis in a future post, but here’s a brief summary:
Benson and Flook both represent examples of patent claims the Court found to be non-statutory. Benson was denied a patent because the Court decided the inventor was overreaching, wanting to pre-empt the entire useful application of binary coded decimal conversion. Flook tried to limit his claim to petro-chemical processes, but was ultimately denied a patent because the claim was directed essentially to a method of using a mathematical formula. Flook‘s mention of petro-chemical processes was merely incidental.
Diehr represents the only example of the three where the claim was found to be patentable. Here, a majority of the Court was convinced because the claims were directed to the application of a mathematical formula to a specific process, not to the formula itself.
A test emerges that is consistent with these three guideposts, which looks to the end result of the claimed method to determine patentability. If the end result of the claimed method merely computes a mathematical formula, regardless of whether it is in the context of an environment, the method is unpatentable. However, if the end result of the claimed method accomplishes some result beyond just computing a formula, then method is likely patentable.