Are Software Patents Dead?—Alice’s Implications for Life Sciences
Not too long ago, getting patents on software and business methods was all the rage. And concern about their effects was profound. In fact, in 2003 I spoke at a Federal Reserve Bank conference devoted to the question of whether such patents were an existential threat to the financial industry. Now, after a series of Supreme Court cases that brought about a dramatic shift in the approach taken by the lower courts and the Patent Office, the question is whether those patents are still alive. The answer is that they are, but barely, and their prognosis is bad.
Do these developments matter to people in the life sciences? The answer is a resounding yes. If we then ask why software patentability matters, the answer is that life sciences are increasingly focused on software-dependent data analysis.
These points were brought home to me when I spoke at another, more recent conference—the Bio-IT World Conference in Boston this past April. This event is a huge gathering of bioinformatics people.
These are the people who use mind-bogglingly complex statistical techniques to try to draw clinically useful inferences from genetic and other health-related raw data. Over the past 10 or so years, rapid advances in sequencing technology have increased the flow of such data from a trickle to a flood. The capacity to analyze it hasn’t kept pace. As bioinformatics tries to catch up, software obviously plays an indispensable role. Equally obviously, whether that software can be patented—directly or indirectly—is a question of enormous economic significance to the bioinformatics industry. Whether software patents look like a good or bad thing will depend on where you are positioned in the industry—that is, are you primarily a creator of analytical tools or a user of others’ creations? Finally, while most bioinformatics software might not seem to fit into the business methods category, the two varieties share some attributes that are relevant to patentability. I will turn now to the legal developments.
Stricter Patentable Subject Matter Rules
The threshold requirement for obtaining a patent is that the invention must claim patentable (sometimes called statutory) subject matter as defined by § 101 of the Patent Act. There are then further, usually more demanding requirements, including that the invention must be novel and nonobvious, and be described with great specificity in the application. Section 101 defines eligible subject matter as including “any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof.” Historically, the courts interpreted this language very broadly, to include, in one classic phrasing, “anything under the sun made by man.” The harder work of excluding bad patents was left to sections 102 (novelty) and 103 (obviousness).
However, three recent Supreme Court cases have made it more difficult to satisfy the statutory subject matter requirement. They have done so by invoking the three categories of subject matter that courts have traditionally excluded from eligibility: laws of nature, products of nature, and abstract ideas. The key cases are Mayo Collaborative Services v. Prometheus Laboratories, Inc. (2012), which rejected—for claiming a law of nature—a patent on a method of measuring the level of a drug in the body and then adjusting the dose; Association for Molecular Pathology v. Myriad Genetics, Inc. (2013), the highly publicized case that invalidated—for claiming a product of nature—patents on DNA that had merely been isolated from the body; and Alice Corp. v. CLS Bank Int’l (2014),which struck down—for claiming an abstract idea—a patent on a method of using a third-party intermediary in a financial transaction. Despite the differences in the inventions, the three cases take a similar approach to raising the subject matter bar. Rather than changing the definitions of the forbidden categories, they require inventors to show more than they formerly had to in order to avoid ineligibility.
Business Methods and Software-based Inventions before 2010
Inventions that implicate the abstract idea exclusion often involve business methods or similar processes that are implemented on a computer. In State Street Bank and Trust Company v. Signature Financial Group, Inc. (1998), the Federal Circuit upheld the eligible subject matter status of a computer-implemented method for collectively managing a portfolio of mutual funds. The court held that a process satisfied § 101 if it yielded a useful, concrete, and tangible result—even if the process involved nothing more than the manipulation of numbers, and the result was itself a number. The case also did away with the so-called business methods exception, under which some earlier cases had treated business methods as inherently unpatentable subject matter.
Immediately after State Street, there was a surge in business method and other software-based patents. Inventions that amounted to little more than using software to implement conventional process could be successfully claimed as (1) a method for doing something, using a computer programmed to carry it out; (2) a computer programmed to carry out the method (this was treated as a conventional claim to a machine); or sometimes even (3) a computer storage medium carrying the code to implement the method (treated as a conventional claim to a physical apparatus).
A related development was the emergence of the machine or transformation test, which has its roots in a series of Supreme Court cases dating back to the 1970s. A process or method that involved an abstract idea could become eligible for patenting if the claim tied the process to a particular machine or transformed something in the real world. The idea was typically a mathematical formula (which is always treated as an abstract idea) and the machine was typically a general-purpose computer that was programmed to calculate the formula. Alternatively, the claimed invention could use the abstract idea to transform a particular article into a different state or thing. In either case, the only point of novelty was computerizing the calculation—the formula itself, the computer, and the pre-computer version of the process were all well-known. For instance, in Diamond v. Diehr (1981), the Supreme Court upheld a claim to a rubber-curing process that used a computer to calculate a long-known equation for determining when the curing was complete. The underlying process was the same as it had been in pre-computer days; what the computer added was the ability to make the calculations far more frequently, which made the determination of when the curing was complete more precise. As a result of these developments, during the pre-2010 period, patenting a business or other method implemented by computer software was relatively easy for skilled patent lawyers.
Software-based Inventions Post-2010
Patenting business methods and software inventions changed when the Supreme Court decided Bilski v. Kappos (2010), which involved a method of entering into contracts to hedge risk in commodity prices. The Court characterized hedging as a well-known financial strategy—an abstract idea. The Federal Circuit had upheld the Patent Office’s rejection of the method as unpatentable subject matter, using the machine-or-transformation test. The claimed method neither tied the abstract idea to any particular machine nor transformed a physical article into a different state or thing. The Supreme Court agreed that the method was unpatentable, but its reasoning was unclear, as multiple opinions were divided on the rationale. The Court rejected the machine-or-transformation test as the sole standard but failed to produce a substitute. It also refused to hold that business methods are always patent-ineligible without providing specific guidance on when they are eligible.
Most recently, in Alice (2014), the Court rejected claims to methods and systems for using a computer as a third-party intermediary to ensure that both parties to a financial transaction meet their obligations. Purportedly following Bilski, the Court imposed a two-part test for method claims that involve abstract ideas (in this case, the well-known strategy of using a neutral third-party intermediary). First, ask whether the claim is “directed to” an abstract idea. The Court determined that the risk-mitigation strategy involved in Alice was an abstract idea. Second, if the claim is directed to an abstract idea, ask whether “the additional elements” supply an “inventive concept” in the physical realm of things, and ensure that the patent is on something “significantly more than” the abstract idea itself. Applying this standard to the facts of Alice, the Court found that using a computer as the third party to ensure that all parties have met their financial obligations before completing a transaction is not a sufficient inventive concept. The Court gave no clear guidance as to what might comprise a sufficient inventive concept, though it observed that the process in Diehr sufficed because it solved an existing technological problem in curing rubber. The Court did make clear, however, that simply implementing the idea on a computer would not be enough.
Since Alice, the lower courts and the Patent Office have displayed a new-found hostility to patents on conventional processes implemented on computers, regardless of how they are drafted. For example, in buySAFE, Inc. v. Google, Inc. (2014), the Federal Circuit rejected Alice-style claims to “methods and machine-readable media encoded to perform steps for guaranteeing a party’s performance of its online transaction.” And in Ultramercial v. Hulu (2014), the same court rejected claims to a method for distributing copyrighted media products over the Internet where the consumer receives a copyrighted media product at no cost in exchange for viewing an advertisement. In both of these cases, the court held that the claims were directed to abstract ideas and that the added technical elements were merely conventional and thus did not contribute a sufficiently inventive concept.
Although most of business method and software patents litigated since Alice have been struck down, processes implemented on computers are not automatically denied protection. For instance, in the Federal Circuit case of DDR Holdings, LLC v. Hotels.com, L.P. (2014), the patent applicant addressed the problem of a “host” website losing Internet shoppers if they were diverted by clicking on the ad for another merchant’s website. The claimed method involved software that generates a composite web page that displays product information from the third-party merchant while giving the shopper the impression that that she is still on the host website. The software essentially carries out the mathematical instructions necessary to produce the composite web page. The patent survived the Alice test, however, because the court could not identify a particular abstract idea known from the pre-Internet world, nor could it analogize this method to previously excluded claims that simply implemented existing ideas on a computer. The court distinguished Ultramercial, where the idea of exchanging media content for viewing ads was a well-known, pre-existing advertising strategy. In DDR, the court found that the claim offered an inventive concept, a new solution to a new problem that had arisen from a new technology.
Application of the New Approach to Life Sciences
The new approach to analyzing the patentability methods has had a clear effect in the life sciences area. A vivid, very recent example of this effect (albeit in a somewhat different factual context) is Ariosa Diagnostics, Inc. v. Sequenom, Inc., decided by the Federal Circuit on June 12, 2015. After previous district court and court of appeals decisions addressing other issues, the Federal Circuit this time addressed the subject matter status of a patent claiming methods of detecting paternity-identifying DNA in a maternal serum or plasma sample—an innovative and highly sophisticated paternity test. A unanimous three-judge panel of the Federal Circuit acknowledged the value of the patent owners’ discovery, but held that Mayo required a finding that the invention was not within the scope of patentable subject matter. The court said, in effect, “we’re sorry, but Mayo made us do it.”
The specific problem, as that court saw it, is that the method “begins and ends with a natural phenomenon”: cell-free fetal DNA (cffDNA), which the method is designed to detect in the maternal sample. Under Mayo and the other Supreme Court cases, if the method starts with a natural phenomenon, it must then add an “inventive concept” to achieve patentable subject matter status. This method does not, since it “amounts to a general instruction to doctors to apply routine, conventional techniques when seeking to detect cffDNA”—exactly how the Supreme Court characterized the method in Mayo.
Judge Linn, who concurred in the judgment of unpatentability but wrote a separate opinion, thought that the Sequenom method might be distinguishable from the one in Mayo. Before this invention, “no one was amplifying and detecting paternally-inherited cffDNA using the plasma or serum of pregnant mothers,” whereas in Mayo the method claimed “the very steps doctors were already doing.” But even he felt compelled to concede that “the sweeping language” of Mayo required rejection of the Sequenom method.
The larger point of the case is that the Federal Circuit has surrendered in the subject matter war. For several years, a number of Federal Circuit judges tried to resist the Supreme Court, arguing for the longstanding approach of letting almost everything in under section 101’s hurdle and leaving the real scrutiny to sections 102 (novelty) and 103 (obviousness). But the Supreme Court has now said unanimously in three different contexts—medical methods (Mayo), genes (Myriad), and computerized business methods (Alice)—that section 101 has to be taken seriously as an initial and significant criterion. The Federal Circuit seems finally to have abandoned the fight.
Practical Implications for Life Sciences
- The business method cases make the critical legal point that computerized versions of established processes should not be patentable. Simply making a calculation or other process faster and more efficient is not enough. Instead, the computerized process must be solving a technological or other tangible problem.
- This poses a particular barrier to patenting analytical methods in bioinformatics. From the law’s perspective, a statistical analysis—however complex or innovative—is an abstract idea or law, and probably also involves laws of nature. Thus, to be patentable subject method the new method must do more than simply automate or accelerate the analysis.
- To reinforce the previous point: algorithms by themselves—again, however complex or innovative—are always treated as laws of nature. This means that the creation of a new bioinformatics algorithm cannot, in and of itself, lead to a patentable invention.
- It is very difficult to envision a new method of data analysis that will clear the new and significantly higher subject matter bar.
- Ariosa v. Sequenom makes the broader point that the bar is also higher for other methods of diagnosis or analysis of medical data. Judge Linn’s effort to distinguish that case from Mayo seemed persuasive, but in the end he could not even persuade himself.
These developments could be bad news or good, depending on who you are. If your business model depends on having exclusive rights to an analytical or medical method, the news is bad. Such method patents will be extremely hard to get, and existing patents may be invalidated if you sue someone for infringement. (An infringement defendant can defend itself by arguing that it is not actually infringing the patent, or by proving that the plaintiff’s patent was wrongly issued and is invalid.) But for those whose business or research may be threatened by these kinds of patents held by others, the news is all good. Patentees will be less likely to sue for infringement, and those who do get sued will have a clear avenue of defense.