Recentive Analytics, Inc. v. Fox Corp. (Fed. Cir. 2025)

By Michael Borella --

For the last several years, patentees and patent practitioners have been waiting for the Federal Circuit to weigh in on the patent eligibility of machine learning models.  There was an expectation that, like any other technology, the patentability under 35 U.S.C. § 101 of inventions that incorporate machine learning would need to be evaluated on a case by case basis.  But there was also some hope that the Court would provide the public with at least some hint of what aspects of machine learning might signal eligibility.  In Recentive Analytics, Inc. v. Fox Corp., these signals are weak at best.

Patent eligibility was historically not a major concern until the Supreme Court's judicially-imposed patent eligibility test[1] of Alice Corp. Pty. Ltd. v. CLS Bank Int'l that came down in 2014.  A valid criticism of Alice is that it behaves like the proverbial nose of wax -- the test exhibits a large degree of inherent subjectivity.  Depending on who is applying the test, what case law they are considering, and perhaps the phase of the moon, virtually any claim under review can be viewed as both eligible and ineligible.

To that point, the Alice opinion made it crystal clear that the computer-controlled rubber mold of Diamond v. Diehr was still patent-eligible.  Even though Diehr was not overturned and remains good law, the Alice test as currently applied by the Federal Circuit and USPTO can easily be used to establish a strong position that the claims of Diehr were in fact ineligible.[2]

As a consequence, the last decade has seen patent assets take on the nature of Schrodinger's Cat.  Any issued patent is in a quantum superposition -- both valid and invalid -- until its eligibility is "observed" by a judicial body.  This uncertainty has real-world anti-competitive consequences, making patent acquisition more expensive yet less effective for small businesses while allowing incumbent technology companies to develop into monopolies or cartels of two-three major players who never assert their patents against one another.  Why does the United States -- ostensibly a proud capitalist society that welcomes competition -- have such a weak patent law?  As a stated in the film version of All the President's Men, "Follow the money."

In any event, Recentive sued Fox for infringement of U.S. Patent Nos. 10,911,811, 10,958,957, 11,386,367, and 11,537,960 in the United States District Court for the Eastern District of Pennsylvania.  According to the Federal Circuit:  "The patents purport to solve problems confronting the entertainment industry and television broadcasters: how to optimize the scheduling of live events and how to optimize network maps, which determine the programs or content displayed by a broadcaster's channels within certain geographic markets at particular times."

The Court divided the four Recentive patents into two categories:  machine learning training and network maps.  The former are directed to (in the words of the Court):

(i) a collecting step (receiving event parameters and target features); (ii) an iterative training step for the machine learning model (identifying relationships within the data); (iii) an output step (generating an optimized schedule); and (iv) an updating step (detecting changes to the data inputs and iteratively generating new, further optimized schedules).

The latter are directed to (again in the words of the Court):

(i) a collecting step (receiving current broadcasting schedules); (ii) an analyzing step (creating a network map); (iii) an updating step (incorporating real-time changes to the data inputs); and (iv) a using step (determining program broadcasts using the optimized network map).

Despite the Court's summarization of the claim language at a high level, the actual claims are quite lengthy and narrow.  They are reproduced in the appendices below.

Fox moved to dismiss for failure to state a claim, alleging that the patents were invalid under § 101.  The District Court agreed, finding that the claims were "directed to the abstract ideas of producing network maps and event schedules, respectively, using known generic mathematical techniques" and further lacked an inventive concept "because the machine learning limitations were no more than broad, functionally described, well-known techniques and claimed only generic and conventional computing devices."

On appeal, the Federal Circuit affirmed.  In doing so, the Court relied heavily on admissions from Recentive that its iterative training was a known technique, fundamental to development of a machine learning model.  But the Court when further, dismissing Recentive's argument that the invention comprised a technical improvement by noting that "neither the claims nor the specifications describe how such an improvement was accomplished."  The Court also quickly disposed of Recentive's argument that the claims were eligible because they apply machine learning to specific acts of event planning.  Indeed, the Court illustrated its reductionist, myopic view of patent eligibility by stating "patents may be directed to abstract ideas where they disclose the use of an already available technology, with its already available basic functions, to use as a tool in executing the claimed process."

The Court also found no inventive concept in the claims.  Recentive argued for "using machine learning to dynamically generate optimized maps and schedules based on real-time data and update them based on changing conditions."  Here, the Court provided a rather substance-free and conclusory analysis, stating that:

[T]his is no more than claiming the abstract idea itself.  Such a position plainly fails to identify anything in the claims that would 'transform' the claimed abstract idea into a patent-eligible application . . . .  [W]e perceive nothing in the claims, whether considered individually or in their ordered combination, that would transform the Machine Learning Training and Network Map patents into something significantly more than the abstract idea of generating event schedules and network maps through the application of machine learning.

Accordingly, all four Recentive patents were found to be ineligible under § 101.

This decision, like § 101 jurisprudence as a whole, leaves itself open to criticism on numerous fronts.  Let's discuss just a few.

First, the claim language of the machine learning patents involved updating the models based on a "real-time change to the one or more user-specific event parameters."  Once trained, models are typically used for quite a while before being retrained.  Thus, it could have been argued that this feature is unconventional.  The opinion never specifically addressed the feature, and did not appear to appreciate this nuance.

Second, continuing a disturbing trend, the Court appears to ignore its older holdings (notably, BASCOM Global Internet Services v. AT&T Mobility) finding that computer-implemented inventions can be eligible even when involving only a combination of generic components and claimed at a high level.

Finally, the Court fails to articulate exactly what is required for us to consider claims as an "ordered combination."  Here, as in many § 101 decisions over the years, the Court's reasoning amounts to no more than "Because we said so."

It is a cliché that bad facts make bad law.  In an effort to stave off the worst types of overreach that might follow because of this decision, let's make it clear what the Court did not hold.

• That all machine learning inventions are ineligible (the Court hinted that more detail in the claim and the specification regarding the technical improvement provided may be sufficient to establish eligibility);

• That any claim reciting a machine learning process at a high level is ineligible (such a claim could be eligible based on its other elements);

• That any claim applying a machine learning model to a new field of data is ineligible (the Court leaves open the possibility that the model or its training process can be the technical improvement if explained in enough detail).

Nonetheless, § 101 is still a mess.  This case does nothing to change today's irrational, arbitrary, and mine-ridden patent eligibility landscape.

Appendix A: Claim 1 of the '367 Patent

1.  A computer-implemented method of dynamically generating an event schedule, the method comprising:
    receiving one or more event parameters for series of live events, wherein the one or more event parameters comprise at least one of venue availability, venue locations, proposed ticket prices, performer fees, venue fees, scheduled performances by one or more performers, or any combination thereof;
    receiving one or more event target features associated with the series of live events, wherein the one or more event target features comprise at least one of event attendance, event profit, event revenue, event expenses, or any combination thereof;
    providing the one or more event parameters and the one or more event target features to a machine learning (ML) model, wherein the ML model is at least one of a neural network ML model and a support vector ML model;
    iteratively training the ML model to identify relationships between different event parameters and the one or more event target features using historical data corresponding to one or more previous series of live events, wherein such iterative training improves the accuracy of the ML model;
    receiving, from a user, one or more user-specific event parameters for a future series of live events to be held in a plurality of geographic regions;
    receiving, from the user, one or more user-specific event weights representing one or more prioritized event target features associated with the future series of live events;
    providing the one or more user-specific event parameters and the one or more user-specific event weights to the trained ML model;
    generating, via the trained ML model, a schedule for the future series of live events that is optimized relative to the one or more prioritized event target features;
    detecting a real-time change to the one or more user-specific event parameters;
    providing the real-time change to the trained ML model to improve the accuracy of the trained ML model; and
    updating, via the trained ML model, the schedule for the future series of live events such that the schedule remains optimized relative to the one or more prioritized event target features in view of the real-time change to the one or more user-specific event parameters.

Appendix B: Claim 1 of the '811 Patent

1.  A computer-implemented method for dynamically generating a network map, the method comprising:
    receiving a schedule for a first plurality of live events scheduled to start at a first time and a second plurality of live events scheduled to start at a second time;
    generating, based on the schedule, a network map mapping the first plurality of live events and the second plurality of live events to a plurality of television stations for a plurality of cities,
    wherein each station from the plurality of stations corresponds to a respective city from the plurality of cities,
    wherein the network map identifies for each station (i) a first live event from the first plurality of live events that will be displayed at the first time and (ii) a second live event from the second plurality of live events that will be displayed at the second time, and
    wherein generating the network map comprises using a machine learning technique to optimize an overall television rating across the first plurality of live events and the second plurality of live events;
    automatically updating the network map on demand and in real time based on a change to at least one of (i) the schedule and (ii) underlying criteria,
    wherein updating the network map comprises updating the mapping of the first plurality of live events and the second plurality of live events to the plurality of television stations; and
    using the network map to determine for each station (i) the first live event from the first plurality of live events that will be displayed at the first time and (ii) the second live event from the second plurality of live events that will be displayed at the second time.

[1] For sake of completeness, the test was explained by the justices as "First, we determine whether the claims at issue are directed to one of those patent-ineligible concepts [(i.e., laws of nature, natural phenomena, and abstract ideas)]. If so, we then ask, '[w]hat else is there in the claims before us?' To answer that question, we consider the elements of each claim both individually and 'as an ordered combination' to determine whether the additional elements 'transform the nature of the claim' into a patent-eligible application. We have described step two of this analysis as a search for an 'inventive concept'—i.e., 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.'"

[2] Michael Borella, Could Alice Be Used to Invalidate Diehr? Of Course It Could, Patent Docs, Apr. 20, 2021.

Recentive Analytics, Inc. v. Fox Corp. (Fed. Cir. 2025)
Panel:  Circuit Judges Dyk and Prost and Chief District Judge Goldberg
Opinion by Circuit Judge Dyk

Comments

[Benjamin A Rellinger]

Agreed that this case is basically "business as usual" with respect to 101.

If we can take any more concrete practical conclusion from the overall timbre of this case, it's that "technical improvement" is the only avenue for 101 eligibility that is at all predictable. You can try to argue eligibility on other grounds, but the courts (and their precedents) allow a wider variety of prejudices and subjective "as a whole" analyses to substitute for proper "show your work" logical reasoning in rebutting those arguments.

In contrast, the court seems to respect the "technical improvement" argument. The worst that they will do is say that a purported improvement is non-technical; however, I've never seen that happen in an example where there actually *was* a technical improvement. Of course, that's partially down to the court's definitely if "technical improvement" being very strictly and narrowly constrained.