A computer does just three things: receives information in the form of bits, transforms this information, and provides output based on the information as transformed. The receiving may take place by way of various types of input modalities, such as keyboards, touchscreens, mice, audio microphones, video cameras, network interfaces, sensors, and so on. The transforming may involve a literally infinite number of possibilities in the form of algorithms. The output may also take on various modalities, including images displayed on a screen, audio, storage to volatile or non-volatile memory, and transmission by way of network interfaces, just to name a few.
Given the fundamental nature of information processing in computing technology, it would be natural to assume that such endeavors would be protected by patent law -- after all, the stated constitutional goal of such law is to "promote the progress of science and useful arts." But due to how the Federal Circuit has interpreted recent Supreme Court decisions, a ground-breaking information processing invention can be unpatentable even if it is novel and non-obvious. More than a philosophical or legal concern, the Federal Circuit's narrow (and many would agree, inconsistent) view on what subject matter can be patented causes business and economic uncertainty across the many industries in which computers have become ubiquitous.
Differentiating between which types of information processing are eligible and which are not is fraught with danger, and even the most experienced patent attorneys (not to mention judges) can find this area of the law perplexing. As a consequence, many entities are struggling to obtain patents on inventions that involve information processing.
This article dips into these waters by exploring the origins of this current bias against information processing, and attempts to stake out a few areas in which claims to information processing may find more or less success. Such an understanding, in turn, provides practical considerations for patenting decisions as well as prosecution and litigation strategies. Nonetheless, more clarity is ultimately needed from the courts.
At the outset, it is undisputed that information itself is not patentable. Information per se, in bits, writing, or some other form, is not a process, machine, article of manufacture, or composition of matter, the four statutory categories set forth in 35 U.S.C. § 101. See Digitech Image Techs. v. Electronics for Imaging (Fed. Cir 2014). Further, attempts to make the mere storage of data on a tangible medium statutory will fail due to printed matter doctrine. See C R Bard Inc. v. AngioDynamics, Inc. (Fed. Cir. 2020). Thus, trying to obtain a patent on a CDROM containing a performance of Beethoven's 5th Symphony will come to naught no matter how you claim the representation of the information encoding the performance.
Information processing is different. It involves steps taken by a computer (or other devices or maybe even humans) to obtain, transform, and provide information. Information processing inventions are not directed to static information -- they are drawn to using and changing information.
Nonetheless, ever since Alice Corp. v. CLS Bank Int'l, it has become increasingly difficult to patent information processing claims, no matter how innovative or useful. In that case, the Supreme Court brought the ill-defined notion of abstract ideas to the fore, which has been used extensively by the Federal Circuit, district courts, and the USPTO to find information processing claims ineligible, even if those claims are otherwise statutory.
A case that typifies the current difficulties information processing inventions can face is Electric Power Group, LLC v. Alstom S.A. (Fed. Cir. 2016). The claims therein recited, albeit in a rather broad fashion, (i) gathering streams of data related to sensor measurements of a power grid, (ii) from this data, determining events occurring in the power grid, and (iii) displaying visualizations of the events. Here, the events could be faults in the power grid that could cause the power grid to fail. Despite this sort of invention seemingly being of the "science and useful arts", the Federal Circuit knocked down the claims under § 101. The Court wrote "[t]he focus of the asserted claims . . . is on collecting information, analyzing it, and displaying certain results of the collection and analysis." This, according to the Court, directly leads to the conclusion that the claims are focused on an abstract idea.
The justification for this conclusion was broken down to mirror the three functional parts of the claim. Regarding the gathering steps, the Court cited to precedent in Digitech, as well as the pre-Alice case CyberSource Corp. v. Retail Decisions, Inc. (Fed. Cir. 2011), for the notion that gathering information is an abstract idea. Particularly, this gathering was deemed to be an abstract mental process even when carried out by computer. CyberSource cites to a line of cases reaching back to Gottschalk v. Benson for the notion that gathering data for mathematical calculations has little -- if any -- patentable weight under § 101. Regarding the data analysis steps, the Court once again viewed these as abstract mental processes, and cited to support for such a conclusion going back to Benson. Finally, regarding the displaying steps, the Court cited to Ultramercial, Inc. v. Hulu, LLC (Fed. Cir. 2014), where displaying an advertisement was considered to be an abstract idea. The Court wrote: "we have recognized that merely presenting the results of abstract processes of collecting and analyzing information, without more . . . is abstract as an ancillary part of such collection and analysis." What you are hearing is an echo of the so-called post-solution activity from Parker v. Flook.
The Court also took issue with the claims' lack of specificity, noting that "rather than claiming some specific way of enabling a computer to monitor data from multiple sources across an electric power grid, [the claims] purport to monopolize every potential solution to the problem—any way of effectively monitoring multiple sources on a power grid." In other words, the claims recited an outcome rather than specific steps to achieve that outcome.
So the issue seems to be not only the information processing nature of the claims, but also that the processing allegedly could be carried out mentally, included mathematical calculations, and lacked specificity. We also know from related cases that the type of information can matter. For example, processing of financial, business, or social information can be viewed as abstract. See, e.g., Intellectual Ventures I LLC v. Capital One Bank (Fed. Cir. 2015); Versata Dev. Group, Inc. v. SAP Am., Inc. (Fed. Cir. 2015); Voter Verified, Inc. v. Election Systems & Software, LLC (Fed. Cir. 2018).
But information processing inventions are not doomed to fall under the § 101 inquiry. Notably, the claims of McRO, Inc. v. Bandai Namco Games America Inc. (Fed. Cir. 2016), involved obtaining rules and phoneme timing, generating morph weights based on the rules and phoneme timing, and then producing lip synchronization and facial expressions in animated characters from the morph weights. Like Electric Power Group, these claims involve gathering, transforming, and displaying information. Nonetheless, the Federal Circuit found the claims to not be abstract.
Why the different outcome? Several reasons, including the claims being more specific and not directed to a mental, financial, or business process. The Court looked to the specification for the understanding that a goal of this invention was to eliminate the subjectivity associated with human-rendered animations. Thus, the invention was more closely tied to a computer than that of Electric Power Group.
The dichotomy of these two cases represents how patent eligibility can be twisted like the proverbial nose of wax. In Electric Power Group, the Court viewed the claims in a deconstructionist manner -- looking at the invention as a series of discrete pieces and finding each of them abstract. In McRO, the focus was on what the claim as a whole provided -- the problem that it solved. The reasoning of Electric Power Group certainly could have been used to justify the opposite outcome in McRO. But the latter's claims were just different enough to be found eligible.
Sometimes information processing inventions can be made patent-eligible by having their output control another device or physical process. The classic case of this is Diamond v. Diehr, the claims of which, at their core, involve a computer receiving temperature readings, the computer performing calculations based on these readings, and the computer causing a rubber press to change state when the calculations take on a certain value. But not all technologies lend themselves to being claimed similarly to Diehr.
Indeed, there are many computerized inventions that are similar to McRO, in that they solve a specific problem in a specific manner without affecting another device -- i.e., the processing is all "in the box". And yet, they are still vulnerable to subject matter eligibility challenges in the USPTO and the courts. As one example, a virus scanning application may involve no more than gathering data (network traffic and/or files), analyzing that data (determining if it matches a virus signature), and providing output (an indication of a detected virus or the lack thereof). As another example, a machine learning application may involve no more than gathering data (training sets), analyzing that data (training the model), and providing output (the trained model). As a further example, a computer-aided design application may involve no more than gathering data (parameters of the design), analyzing that data (formulating the design), and providing output (the final design).
In each of these cases, it may not be viable to claim the invention in a way that is directly analogous to Diehr, because doing so would lead to a problem with divided infringement. For instance, the entity that makes the computer-aided design application is unlikely to also use the designs to create physical products.
For at least these reasons, there is a need for the Federal Circuit to elucidate and define what factors are more or less likely to make an information processing inventions patent-eligible.[1] The current take on § 101 puts too much flexibility in the hands of judges and examiners; particularly, at what level they consider the claims, how much emphasis they put on the specification, and the role that prior art plays. Patentees, especially small- and medium-sized companies with limited budgets, need to make smart decisions with respect to their investments in patents. Without clarity, the risks and rewards associated with such investments will remain difficult to quantify.
[1] One approach that we have outlined before is to, when possible, draft claims that are specific, solve a technical problem in a technical fashion, and recite features not in the prior art.
