Posted on July 11, 2023

In this article, we explore two 2023 patentable subject matter cases at the Federal Circuit ("Fed. Cir.") that are instructive as to how to approach claim wording and how to address challenges to the claims after issuance, particularly for computer-related inventions.

Red Hat, Inc. v. Sequoia Technology, LLC (Fed. Cir. April 12, 2023) is an interesting case; one that any practitioner with computer (server) storage company client(s) should be familiar with. Red Hat is a consolidation of several cases in which Sequoia sued other companies employing virtual disk drives, also known as logical volumes. The patent involved, U.S. Patent No. 6,718,436 (the '436 patent) describes the problem sought to be solved as "… managing a logical volume, the huge size of metadata delays system booting time and uses too much memory." The solution is given in the patent as "…using a disk partition as a volume construction unit…" for logical volumes in order to minimize metadata.

In the '436 patent, a "logical volume" is made up of multiple disk partitions, a "disk partition" is a minimum unit of the logical volume, and "extents" are a minimum unit of storage space that make up disk partitions. The disk partitions can be resized into disk partition units. Claim 1 is a method claim and claim 8 is a computer-readable medium ("CRM") version of method claim 1.

Claim construction was referred by the district court to a magistrate judge. The parties disputed the construction of "computer-readable recording media" ("CRM" with "recording"), "disk partition," "logical volume" and "used or not used." The magistrate judge adopted Red Hat's construction of "computer-readable recording medium," which construction includes transitory media (e.g., signals or waves). The specification provides a list of examples for CRM's, none of which were transitory media. However, the specification included no express definition for CRMs. The magistrate judge also noted that the specification did not exclude transitory media and relied on Red Hat's expert that one of ordinary skill in the art would understand "CRM" to include transitory media based on definitions in 34 other patents, none of which were owned by Sequoia. In addition, the magistrate judge's claim construction required that a logical volume is made up of whole partitions, not portion(s) of partitions. The district court adopted the magistrate judge's claim construction.

Based on the claim construction of "CRM," the CRM claims were found to be ineligible subject matter under 35 U.S.C. §101. On appeal to the Fed. Cir., Sequoia challenged the construction of all four terms. The Fed.Cir. started by recognizing that the '436 patent claims do not use the term "computer-readable medium" or the term "computer-readable recording medium." Rather, the claims actually recited "computer-readable recording medium storing instructions," which the Fed.Cir. referred to as "narrowing" compared to the two other terms. When construing claim terms, the Fed. Cir. emphasized that instrinsic evidence (e.g., claim language and the specification) is first and foremost relied on. Due to the nature of transitory media, i.e., being ephemeral or short-lived, the Fed. Cir. agreed with Sequoia that one of ordinary skill in the art would not logically understand the claim term, in the context of memory systems, to include transitory media. According to the Fed. Cir., the claim term itself, as well as the specification (every example was hardware-based), supported this conclusion. Oddly, in this case, not having an express definition in the specification actually benefitted Sequoia. It is generally better to have a definition for CRM in the specification, though to be fair, once a term is defined, you will be stuck with that definition, so craft any definitions with great care.

The Fed. Cir. made a point of emphasizing that it was solely relying on intrinsic evidence (a stretch, as you will see), whereas Red Hat and the district court relied on extrinsic evidence (e.g., the expert opinion). Red Hat also argued that the '436 specification states that CRM "include[es]" non-transitory media, so it is open ended and could include transitory media. That might seem logical, but it does not mean that Sequoia's version could include transitory media; rather, it could include other types of hardware (physical storage), as nothing in the claims or specification even hinted at including transitory media.

Red Hat's expert analyzed 34 patents/patent applications unrelated to Sequoia's to determine what one of ordinary skill in the art's understanding of "computer-readable recording medium," notably lacking "storing instructions" that appears in the claims. Not surprisingly, the Fed. Cir. was not impressed and held that the district court clearly erred in considering Red Hat's expert testimony, which was inconsistent with the intrinsic evidence. For example, an object of the invention is to provide "a computer-readable recording medium storing a program or data structure," which the Fed. Cir. found incompatible with a transitory signal. Indeed, to store a program or data structure for any length of time requires persistent storage, not ephemeral or temporary storage.

Likewise, the Fed. Cir. was not impressed with Red Hat's reliance on Mentor Graphics Corp. v. EVE-USA (Fed. Cir. 2017), which affirmed a district court's construction of "computer readable medium" as including transitory signals. However, the '436 patent's CRM term is considerably more narrow and the Mentor Graphics holding was based on an express definition of the term in the specification that included transitory signals, which is of course not the case in Sequoia.

Turning to other disputed terms, "disk partition" and "logical volume," the issue was whether the claimed invention can allocate a portion (versus all) of a disk partition to a logical volume. Despite Sequoia's arguments that a portion can be used, it could not escape the words of the patent and statements made during prosecution (greetings, extrinsic evidence) to the contrary.

However, that extrinsic evidence pales in comparison to the following. For the final disputed claim term, "used or not used," the Federal Circuit rubber stamped the district court's reliance on a paper written by the inventors about an earlier version of the claimed invention, showing even more reliance on extrinsic evidence.

We now turn to the second case, People.ai, Inc. v. Clari Inc. et al. (Fed. Cir. 2023). People.ai is an appeal from the Northern District of California' grant of judgement on the pleadings in favor of the defendants. Involved at the district court were seven patents, all claims of which the district court held invalid under 35 U.S.C. §101. People.ai appealed the decision for three of the seven patents, all of which are directed broadly to business-analytics software to optimize customer relationship management (C.R.M.) and other systems.

More specifically, the three patents are directed to the way data is added to "systems of record," which can be C.R.M. systems, ERP (enterprise resource planning) systems, document management systems and other types of systems. People.ai's description foreshadows the outcome, stating that the claims are directed to "…the use of an objective rules-based approach for using tailored filtering policies to intelligently derive useful business information from emails, meetings and phone calls[.]" The patents identify the problem as manual updating of systems of record, which the patents are said to overcome by linking electronic activities, e.g., email, phone calls and calendar events, to record objects included in systems of record. In other words, the patents are automating a previously manual process.

Due to People.ai's failure to brief arguments regarding dependent claims and not objecting to the independent claim used, the Fed. Cir. limited the analysis to one independent claim of one of the patents. The claim, paraphrased here for emphasis, is directed to a system configured to "access" electronic activities and record objects of systems of record, "identify" electronic activities to match record objects, "determine" a data source provider, "identify" systems of record corresponding to the data source provider, "select" candidate record object(s) in two sets of candidate record objects and "store" an association between the candidate record object(s) and an electronic activity. The claim of each of the other two patents likewise used similar language as the quoted actions, all of which were applied to a generic computer. Such terms as those quoted above and other common ones, such as "calculate," are trigger terms for the Fed. Cir. finding patent-ineligible subject matter. While some use of such terms may at times be unavoidable, it is best practice to minimize such uses.

To refresh, the "Alice" test for subject matter eligibility involves determining whether the claims are directed to a patent-ineligible concept, such as an abstract idea, and, if a patent-ineligible concept is found, examine the claim for transformational inventive aspects. Alise Corp. Pty. Ltd. v. CLS Bank Int'l (Sup. Ct. 2014). Applying the Alice test, the district court held that the claims are directed to an abstract idea and do nothing more than recite common business practices performed by people. For example, for an independent claim of one of the patents, the district court found it to be "directed to the abstract idea of data processing by restricting certain data from further analysis based on various sets of generic rules" and lacked any transformational inventive concepts.

The Fed. Cir. agreed with the district court's analysis under Alice. The Fed. Cir. observed that its relevant precedent has consistently held that automation of a previously manual process, by itself, does not pass the Alice test. The benefits of People.ai's automation include improved accuracy, speed and efficiency, the death knell for computer automation claims.

The Fed. Cir. opinion reminds that automation of a manual process may not be an abstract idea if it differs from the manual process and provides "a specific means or method that improves the relevant technology," citing McRO, Inc. v. Bandai Namco Games America Inc. (Fed. Cir. 2016) and Enfish, LLC v. Microsoft Corp. (Fed. Cir. 2016). For example, in McRO, the invention was held not to be an abstract idea because it used unconventional rules and only automated part of preexisting 3D animation method to synchronize realistic lip and facial expressions in animated characters. In addition, unlike McRO and Finjan, Inc. v. Blue Coat Systems, Inc. (Fed. Cir. 2018), the Fed. Cir. found the claims at issue not to improve computer functionality.

Despite People.ai's argument to the contrary, the Fed. Cir. found the steps of the claimed process to be the same as those in the corresponding manual process long-practiced in business.

People.ai did make what seem at first to be reasonable points. The first point is that manually practicing the claimed steps would require opening and reading all incoming and outgoing communications, as well as monitoring all phone calls and meetings at a given company. The second point is that manually performing the claimed steps would require reading the content of all those communications in order to identify relevant business information and apply the selected filtering policies. Unfortunately, the actual claim language came back to haunt People.ai. For example, one of the claims only required an analysis of two communications to identify a first and a second electronic activity, not all communications. As another example, the same claim required parsing a first electronic activity (not all) to identify electronic account(s). It may be advisable in such situations to add these types of things in dependent claims.

The Fed. Cir. then weeded out from People.ai's arguments those that go beyond what is actually in the claims. This left three claim aspects, but none were seen by the Fed. Cir. to be different than "the steps employed by the prototypical salesperson in the manual process." Arguments regarding other claims met with the same fate. For example, People.ai argued, in part, that mailrooms do not maintain a node graph. However, none of the claims actually required a node graph. A hard lesson, but arguments both in prosecution and litigation need to use the exact language in the claims, i.e., no paraphrasing or embellishing of the actual wording. In prosecution of a patent application, it can be particularly maddening to receive the Examiner argument that (quoting your argument in the prior response) "is not in the claims." It feels to the practitioner like a cop-out, and to some extent, it is any easy out for the Examiner, so do not give the Examiner the ammunition to take that easy out.

Moving to step two of the Alice inquiry, the Fed. Cir. found that all the steps of the claim noted above with the trigger words all to be necessary parts to the abstract idea of "data processing by restricting certain data from further analysis based on various sets of generic rules." The remainder of the claim language was viewed as simply generic computer features.

As to the representative claim, the Fed. Cir. ended with a comment that People.ai made no specific argument as to any depending claims, so their work was done. This is another practice point to keep in mind; that is, separately argue any dependent claim with any arguable aspect(s) not already argued with respect to the corresponding independent claim. If claimed properly, the dependent claims should be substantive, not just along for the ride. Indeed, the purpose of dependent claims is something to fall back on if the independent claim fails. This is true both during prosecution and after the patent issues when it is challenged or enforced.

Wayne F. Reinke