Software patents have always been controversial, in large part because the dividing line between a patentable software-based invention and one that is not has never been clearly defined. But the often hazy body of law that determines software patentability could be about to change.
On March 31, the U.S. Supreme Court will hear oral arguments in Alice Corp. Pty. Ltd. v. CLS Bank Int’l., No. 13-298, a case that could have wide consequences in the tech community and beyond. At stake is when and how a particular software-based invention—that is, an invention that incorporates the performance of a computer and software—is entitled to a patent.
At the heart of the matter is the “abstract idea.” An abstract idea on its own is not patentable, but what exactly counts as an abstract idea? The Supreme Court has never set out a specific test for what is and is not an abstract idea and the Federal Circuit, the nation’s top patent-specific court, is internally divided.
As Alice, the patent owner, frames it, the question for the Supreme Court to decide couldn’t be more fundamental:
Whether claims to computer-implemented inventions—including claims to systems and machines, processes, and items of manufacture—are directed to patent-eligible subject matter . . . ?
Forty years into the Digital Revolution (and almost 30 since I got my Commodore 64), you would think that we would have had this one nailed down tight by now. It’s not for lack of trying.
Below we look at what is clear about software-implemented inventions, the case that gave rise to Supreme Court review, and what to do while you wait for a ruling.
Some concrete limits on the definition of “an abstract idea”
To be patentable, an invention has to: cover patentable subject matter, be novel, and not be obvious. The gatekeeper is “patentable subject matter,” which has famously been defined as “anything under the sun that is made by man.”
Over time, three categorical exceptions to patentability have developed: laws of nature, physical phenomena, and abstract ideas. Never patentable no matter how new or useful they may be.
Laws of nature and physical phenomena are relatively easy to define and circumscribe — magnetism, photosynthesis, the Doppler effect, and so forth.
An “abstract idea” is less easy to define, and the “abstract ideas” bucket is where many software-based inventions go to die.
Based on precedents, here’s what’s clear about what is and isn’t patentable:
A statement of an algorithm or concept is not patentable. Take an algorithm for converting base-10 to base-2. You can do this by hand, using a formula, or you can use a computer to do it much more quickly with a series of microinstructions to shift register contents. In 1963, Gary Benson and Arthur Tabbot filed an application for that method, which was likely a novel concept at the time. Abstract, said the Court. The patent would preempt performance of the method for just about any use through existing or future machinery, with or without any apparatus. So, no pure software patents or patents on algorithms.
Including a particular machine in the claimed method might make it patentable. While the algorithm alone might not be patentable, implementing that algorithm with a specific machine might give you something patentable.
What about a system that uses the Arrhenius equation to prevent under- or over-curing of synthetic rubber? Patentable. Enter the “machine or transformation test,” which basically says that a process or algorithm may be patentable if it is tied to a particular machine or transforms an article to another state or thing. The Arrhenius equation dates back to 1889, but the patent applied the algorithm to transform the rubber to a new state.
That machine must impose meaningful limits on the scope of the patent claim, and can’t be mere “post-solution activity.” For example, printing your results at the end of a calculation won’t suffice.
But the machine or transformation test is notoriously complicated and subjective. What’s a “particular” machine, what limitations are “meaningful,” what types of transformations qualify? It’s also not the sole test for the defining the outer bounds of abstractness since, in 2010, the Supreme Court held that while this is a useful test, it is not the only test for determining patent eligibility.
The line between patent-eligible application of an idea and abstract idea remains undefined.
Subject matter is a sieve, but what kind of sieve?
The court of last resort for many patent cases is the Federal Circuit, which, for our purposes, can be roughly divided into two camps. One side views subject matter determinations as a fine mesh sieve (a chinois, if you will), ready to trap all the bits of abstract particulate and keep it them of the patent system. The other side views the determination as a coarse filter (a colander, to keep the metaphor going), configured to trap the obvious offenders, but letting the other requirements such as novelty and non-obviousness do the heavy lifting on determining patent eligibility.
After a district court proceeding (CLS won), a first Federal Circuit panel (Alice won), the Alice case went before the full Federal Circuit. Recognizing that the law of abstract ideas was unclear and even inconsistent, the full court agreed last year to re-hear the case, a routine patent infringement dispute involving computer-based risk management in financial transactions managing risk.
A panel of ten judges issued a total of six opinions, which is highly unusual. Only the invalidity of the method claims was able to garner a majority.
The Federal Circuit generally agreed that the basic tools of scientific and technological work should remain open and unpatentable, and that whatever test is employed, it should be flexible. The evaluation should depend on substance and not the “clever” drafting efforts of patent attorneys.
Where the opinions varied was in their approaches, from coarse to fine filter. Since they came from a panel of judges who are all experts in patent law, and they touch an industry as significant as software, it was almost certain the Supreme Court would review the case. And it is.
What about Alice?
The Supreme Court hears oral argument in the case on March 31 and should have a decision by June. So what do you do in the meantime?
The USPTO’s 2010 patent eligibility guidelines continue to be a valuable resource for evaluating patentability, whether you’re an inventor, an accused infringer, or a cautious developer. The guidelines follow the cases described above.
CLS Bank is the odds-on favorite to win this final round, but how it wins and what the Court’s opinion says is where the real impact will be.
In the past, the Court has resisted defining a specific test for unpatentable abstractness, but this time might be different, and many are hoping that it is. District court decisions in recent years and the splintered decision of the Federal Circuit have shown that the present framework isn’t working, to the detriment of everyone from small inventors to big business. IBM and the Electronic Frontier Foundation (among others) each weighed in with the Court and called for clear standards to determine patent validity. A little more certainty would go a long way.
Featured image by Thinkstock/olm26250/