In the wake of last year’s Supreme Court decision in Alice Corp. Pty. Ltd. v. CLS Bank Int’l, 134 S. Ct. 2347 (2014), dozens of courts have declared scores of patents to be invalid as not satisfying the requirements of §101 of the patent statute. The Federal Circuit recently issued a decision that provides the same result but does not rely on the Alicestandard and may have a significant and lasting impact on patents for software-related inventions. Allvoice Devs. US, LLC v. Microsoft Corp., No. 2014-1258, 2015 U.S. App. LEXIS 8476 (Fed. Cir. May 22, 2015). In Allvoice, the Federal Circuit declared a patent’s claims to be invalid because it simply found them not to be directed to one of the four statutory categories of inventions identified in §101.

The Alice case and its progeny have all relied on certain judicially created exceptions to §101, stating that inventions are not of a patentable type if, for instance, they encompass an abstract idea. Various tests evolved over the past few years, culminating in Alice, to determine whether the claims of a patent fall into one of these judicially created exceptions. In Allvoice, the Federal Circuit took a very different approach.

Judge O’Malley, writing an opinion for a unanimous panel including Judges Prost and Dyk, took a strictly literal reading of §101 and said that if a patent claim does not recite an invention that matches one of the four categories recited therein (i.e., a “process, machine, manufacture, or composition of matter”), it is invalid.

For context, the patent at issue is U.S. Patent No. 5,799,273, which carries the somewhat awkward title, “Automated proofreading using interface linking recognized words to their audio data while text is being changed.” The patent has 78 claims. Independent claims 1, 52, 69, and 75 are directed to “data processing apparatus.” Independent claims 15 and 54 are directed to “a data processing arrangement.” Independent claims 28, 56, 71, and 77 are styled as “a data processing method,” while independent claims 40 and 58 begin, “a method of processing data.” Independent claims 51 and 73 are directed to “a computer usable medium having computer readable instructions.” Independent claims 60 and 64 recite “a universal speech-recognition interface.”

It is the last of these that the Federal Circuit considered to be outside the ambit of §101. The court began its analysis by stating that, “except for process claims, the eligible subject matter must exist in some physical or tangible form.” (quoting Digitech Image Techs., LLC v. Elecs. for Imaging, Inc., 758 F.3d 1344, 1348 (Fed. Cir. 2014)). Diving deeper, the court quoted other passages from Digitech, with citations back to the mid-1800s, for the notion that manufactures and compositions of matter must be tangible.

In contrast, the Allvoice court held that claims 60 and 64 (and their dependents) “do not recite a process or tangible or physical object and, thus, do not fall within any of the categories of eligible subject matter.” The court noted that Allvoice had, at the district court level, argued that the claims were limited to software, but before the Federal Circuit, argued that the claimed interfaces are software instructions. The Federal Circuit held that unless a process is claimed, “the subject matter must exist in tangible form. Here, the disputed claims merely claim software instructions without any hardware limitations.”

Closer review of the claims calls this conclusion into question. The claims recite interfaces that comprise three elements: “input means for receiving speech-recognition data,” “output means for outputting the recognized words,” and “means… for determining positions of the recognised words.” The Federal Circuit relegated to a footnote any recognition that this claim form might benefit from the functional claim provisions of §112(6) of the patent statute and essentially ignored that issue based on an assertion that, “the means-plus-function limitations, as construed by Allvoice, do not correspond to tangible structure.”

Thus, the court appears to have simply looked to what it saw as a flawed argument by Allvoice, rather than independently reviewing the specification to determine proper means-plus-function determination of these claim elements. Such an approach seems particularly harsh, placing undue emphasis on lawyering skills as opposed to the court’s independent review of the patent. In this instance, figure 2 of the patent “illustrates a schematic overview of the internal architecture of the computer” and shows the speech recognition interface as being part of a processor of that computer. See ’273 patent at column 5, lines 37 – 44. More generally, the patent includes several system block diagrams and corresponding flow charts describing at several levels the tangible structures that correspond to the elements of the claimed interface. For reasons that are not explained in the opinion, though one might surmise conservation of judicial resources, the Federal Circuit chose not to independently undertake analysis of the patent but instead to allow these claims to fail due to the way that the patent owner’s counsel characterized them in litigation.

Notably, the Allvoice opinion is marked “NOTE: This disposition is nonprecedential.” Under old Federal Circuit rules, nonprecedential opinions could not be cited by parties to later cases in briefs and the like, but this has not been the case since 2007. Current Federal Circuit Rule 32.1 says that a panel of the court can designate an opinion as nonprecedential, but that no longer prevents parties from citing to it or prevents the court from looking to it “for guidance or persuasive reasoning.” In addition, not just parties but “any person” is afforded 60 days after issuance of a nonprecedential opinion or order to request that it be reissued as precedential.

What is particularly concerning in this instance is that other courts may adopt the approach in Allvoice as a convenient proxy for invalidating patents—one that may be even easier to apply than the Alice test. It would be incorrect to look at a claim that begins “an interface comprising” and invalidate such a claim merely because an interface is not necessarily a process, machine, manufacture, or composition of matter.

Practitioners would be wise to take note of this decision in framing claims as well. Constructs that have been commonplace for years may now be suspect if the Allvoice approach begins to be used by courts in other cases. Particularly in the case of software-implemented inventions, but also in other circumstances, patent lawyers may want to clarify both the nature of the interfaces, engines, systems and the like in their claims both to clearly show the tangible aspect of these and to tie them as directly as possible to the specific statutory categories.

One important aspect of §101 that the Allvoice court did not appear to consider is that §101 is not, by its own terms, limited to the four categories recited above. The actual text of the statute reads, “process, machine, manufacture or composition of matter, or any new and useful improvement thereof.” It does not follow that an improvement to a machine must itself be a machine, for instance. For example, a machine can be improved by adjusting it in a particular way. More specifically, a computing machine may be improved by programming it in a particular way. That Congress intended §101 to be broadly rather than narrowly interpreted is settled law, so ignoring or strictly limiting the “improvement” language in §101 would be difficult to square with dozens of Federal Circuit and Supreme Court decisions.

It may be that courts and the Patent and Trademark Office itself are enjoying the convenience of invalidating patents and rejecting patent applications based on simplistic analysis of what qualifies as patentable subject matter. It may be that such approaches provide a quick and inexpensive way to dispose of poorly written claims and overly aggressive patent plaintiffs. However, there is great danger in such an approach as well, since such shortcuts can lead to invalidation of patent claims that do not deserve such a fate. Patent law is not easy, and we must be careful not to jump to anything that seems like too simple a solution. As always, we will see how the situation develops and report further.