How to Fix the Software Patent Mess: Go Back to Basics

Most people appreciate the benefits provided by computer software. Want to know how to get somewhere? Just ask Siri program on your iPhone. Reassured that newer cars are more stable and safer than just a few years ago? Thank software-controlled steering. Looking forward to self-driving cars? Of course, look to the software engineers. Software is a ubiquitous part of our lives and our economy. Most would agree that fostering software innovation benefits everyone.

But not all patents that are granted foster innovation. Indeed, after experiencing the mischief arising from undeserved privileges bestowed by the British Crown, the U.S. founders deemed the concept of patents to be a public embarrassment. See Walterscheid, Patents and the Jeffersonian Mythology, 29 J. Marshall L. Rv. 269 (1995).  Still, the authors of the U.S. Constitution thought it worthwhile to accept such embarrassment in return for a patent system that would “promote the useful arts.” The term, useful arts, as used at the time, referred to practical knowledge, in distinction to the performing or fine arts. In crafting the provision for patents found in Article I of the Constitution, they effectively directed that the government be “stingy” in granting patents, issuing such exclusive rights only when it would promote important practical knowledge.

In following this mandate, the U.S. patent system should be implemented in a way to promote software innovation. In recent years, U.S. courts have developed a series of guidelines defining boundaries for patent eligibility. To the extent that such rules block patents from being issued too freely, they should be applauded as consistent with the Constitutional mandate. However, over the past decade, the U.S. Supreme Court’s decisions have presented a “moving target” of when software and other computer-implemented inventions are eligible for patent protection. This lack of clarity is reducing business incentives to develop software.

Most recently, in Alice Corp. v. CLS Bank Int’l, 717 F. 3d 1269 (Fed. Cir. 2013), the Supreme Court endorsed a two-prong test for patent eligibility:

Determine whether the claims at issue are directed to one of the patent-ineligible concepts: Laws of nature, natural phenomena, and abstract ideas.

If so, determine whether any additional claim elements transform the nature of the claim into a patent-eligible application. An element or combination of elements that is “sufficient to ensure that the patent in practice amounts to significantly more than a patent upon the [ineligible concept] itself.”

The two-prong test has been soundly ridiculed by patent lawyers striving to counsel their clients. In the context of a patent claims, rules based on judgments of what is, or is not, “abstract” or “significantly more” are simply unhelpful.

Judge Wu of the U.S. District Court for the Central District of California has openly criticized Alice for evoking an “I know it when I see it” test. Judge Pfaelzer, a colleague of Judge Wu, writes that the Supreme Court’s patent eligibility cases “often confuse more than they clarify [and] appear to contradict each other on important issues.” Former Chief Judge of the Federal Circuit Paul Michel advises that Alice “create[d] a standard that is too vague, too subjective, too unpredictable and impossible to administer in a coherent consistent way in the patent office or in the district courts or even in the Federal Circuit.”

The uncertainty engendered by the Alice decision has serious real world consequences. To many business managers, “uncertainty” is another word for “risk.” And business managers uniformly seek to reduce risk for their investments.

Controlling risk is a basic tenet of business. When considering potential investments that promise the same return, a rational business owner will invest his/her money in the project with the lower risk. If one bond presents more uncertainty or risk than another bond, the riskier bond must promise a higher return before the market will invest in it. Increase the risk of a project not yielding the expected benefit, and business owners are less prone to invest in it.

The Alice decision has increased the risk associated with software development. It has made less clear that any expensive, new software will be eligible for patent protection. It has also accordingly increased the risk that such software will be copied by its competitors upon introducing it to the marketplace.

In light of the greater risk now associated with software development, businesses are less likely to invest in it. And for those software investments already made, a rational business owner will now have a greater tendency to keep the software a trade secret (and rely on robust code obfuscation) rather than paying for the preparation, prosecution and maintenance of patent rights of uncertain usefulness.

Judge Newman, concurring in part and dissenting in part from the CAFC’s majority opinion in Alice, implicitly calls for an end to amorphous semantics promulgated by the majority:

[T]he statutory purpose of section 101 [is] to provide an inclusive listing of the useful arts.

Then, upon crossing this threshold into the patent system, examination of the particular subject matter on the substantive criteria of patentability will eliminate claims that are ‘abstract’ or ‘preemptive,’ on application of the laws of novelty, utility, prior art, obviousness, description, enablement, and specificity.

There is no need for an all-purpose definition of ‘abstractness’ or ‘preemption,’ as heroically attempted today.

I propose that the court return to the statute, and hold that when the subject matter is within the statutory classes in section 101, eligibility is established.

To summarize Judge Newman: A process (including the process inherent in any functional software) meets the statutory standard for patent eligibility. An applicant may obtain patent protection for a process (including software) if it meets the other requirements for patentability.

Some argue that this straightforward approach runs contrary to the Founders’ expectation that the Patent Office should be stingy with its patent grants. They assert that removing the “non-abstract” or “abstract plus significantly more” tests of Alice would lead to the issuance of patents with claims that are either too broad or mere computer implementations of obvious methods.

But this doesn’t have to be the case. To avoid it, the Patent Office (and courts) should only enforce the written description and non-obviousness requirements of the Patent Statute. Consistent with the written description requirement, the Patent Office should require an applicant to narrow its claims if they exceed the bounds of what the inventor was in possession of upon filing the application.

As to the notion that the two-prong Alice test is required to stop the Patent Office from granting patents for old or obvious methods implemented on a computer, we need only refer to the often-used Winslow Tableau (described in In re Winslow, 365 F.2d 1017 (C.C.P.A. 1966)):

[P]icture the inventor as working in his shop with the prior art references . . . hanging on the walls around him. . . . Winslow would have said to himself, “Now what can I do to hold them more securely?” Looking around the walls, he would see Hellman’s envelopes with holes in their flaps hung on a rod. He would then say to himself, “Ha! I can punch holes in my bags and put a little rod (pin) through the holes. That will hold them!

For computer-implemented inventions, we can assume that a description of computers — and a statement that computers are known for quick and efficient computations — are included among the prior art references pinned to the walls of the inventor’s cubical. This is akin to the conception of a “notorious computer” implemented by the European Patent Office. We just assume that inventors know about existing computers. See EPO Guidelines for Examination. This approach simply acknowledges that computers, and how to use them, are well known.

In response to an application for a patent on a software invention, the USPTO has said:

Inventors understand that computers are good at processing, and understand how to use them. So, if you develop a method that is obvious, implementing it on a computer adds nothing to the argument that the claimed method or device is non-obvious.

Develop a new computer or a way to make a computer run better: great, you may well have a patentable invention. Take a known or obvious method and put it on a computer: no patent, in light of the “notorious computer.

A Winslow-inventor wall (stocked with a description of the notorious computer), combined with already-established tools for determining non-obviousness, provide a more reliable and predictable way to block patents from being issued for insignificant developments. If the Courts or USPTO believe that a method should not be patented, they need not probe into the murky waters of what is or is not “abstract” or “significantly more.” Then only need to apply rigorously the obviousness test they have applied for the past half-century.

This approach would make business less uncertain as to whether not their proposed investments in software could receive patent protection. And reducing this risk would promote the future useful art of software development.