By Jason Rantanen and Nikola Datzov. Professor Datzov is an Assistant Professor at the University of North Dakota School of Law.
The Mayo/Alice framework used to determine patent eligibility has been a lightning rod for criticism since the Supreme Court’s decisions a decade ago. Some have argued that the two-step framework is inconsistent with earlier patent eligibility precedent, while others have focused their objections on its purported negative effects on innovation. But arguably the most popular narrative is the asserted fatal flaw that the framework lacks administrability and cannot be applied predictably.
Too many critics to count—including academics, practitioners, legislators, and judges—have lambasted the patent eligibility framework as an unpredictable morass of confusion. Even some judges on the Federal Circuit have labeled the eligibility framework as an “incoherent doctrine”[1] that might tempt district courts into “an effective coin toss,”[2] while others have openly confessed that “the nation’s lone patent court … [is] at a loss as to how to uniformly apply § 101.”[3] The latest legislative attempt to reframe patent eligibility is similarly premised on “extensive confusion and lack of consistency [in applying the 101 exceptions] throughout the judicial branch of the Federal Government and Federal agencies.”[4] These concerns for unpredictability are undoubtedly echoed by countless practitioners who have been in the trenches of litigating this polarizing issue. Given the particular emphasis on bringing predictability to patent law in creating the Federal Circuit, these criticisms raise a grave concern regarding one of the most important areas in patent law.
Yet, empirical analysis suggests that those claims of unpredictability may stand on shaky grounds. In an attempt to better understand whether judges have been able to predictably apply the doctrine, we analyzed the Federal Circuit’s entire body of 368 cases on § 101 from 2012-2022 at a more granular level than any prior study. To evaluate the level of predictability within § 101 jurisprudence, we used a multi-dimensional approach that considered: (1) whether lower tribunals are reaching the legally correct outcome (i.e., reversal rates); (2) whether lower tribunals are correctly applying existing law in each case (i.e., error rates); and (3) whether appellate judges demonstrate disagreement in applying the law (i.e., dissent rates).
What we found shocked us. It turns out that patent eligible subject matter jurisprudence looks remarkably like other patent law issues at the Federal Circuit and lacks the kinds of empirical hallmarks that we would expect given the rhetoric for unpredictability. In fact, under one of the most well-established metrics for measuring the predictability in the law, § 101 proved to be more predictable than other areas of patent law over the past decade.
Importantly, our goal was not to examine or argue where the line should be drawn for determining what is eligible for a patent. Instead, we just sought to evaluate whether judges can tell where the line has been drawn by the Supreme Court in Mayo. In other words, whether the Mayo/Alice framework has proven workable and predictable through ten years of litigation. As to that question, our analysis suggests that the popular narrative that § 101 and the Mayo/Alice framework cannot be predictably applied, particularly by judges might be more of a misconception than an accurate narrative.
Below are some of our key findings regarding predictability from the research study. If you’d like to jump ahead to the draft paper itself, here’s a link: http://ssrn.com/abstract=4380434.
A Historically High Affirmance Rate
Our examination of the Federal Circuit’s body of case law on § 101 revealed that from the Federal Circuit’s perspective, the district courts and the PTO are getting the right result nearly every time, boasting an overall 87.2% affirmance rate.
Figure 1
Figure 1 shows that the Federal Circuit believes district courts and the PTO are getting the right result in a very high percentage of cases. This is especially notable given that 98.2% of the district court decisions reviewed by the Federal Circuit arose in the context of a Rule 12 motion, summary judgment, or JMOL—procedural postures in which the standard of review on appeal owes no deference to the district court.
Thinking about these numbers in context, the high affirmance rate on patent eligibility is not only a far cry from the Federal Circuit’s one-time 50% affirmance rate on claim construction, it’s higher than the Federal Circuit’s track record on obviousness. In fact, this may be the highest affirmance rate of any significant patent law issue tracked over a significant period of time.
District Courts Very Rarely Err in Their Analysis
To take a deeper look, we also examined the Federal Circuit’s analysis when it did affirm to see whether maybe the lower tribunal got the right result but for the wrong reason. Although an analysis of affirmance rates has been an established and important marker in measuring the predictability of the law, it provides a somewhat incomplete picture of judges’ ability to apply the law predictably because it focuses only on the outcomes and not the process of making the decision. It’s possible that a judge can err in the legal analysis (or incorrectly apply a legal standard) and still reach the correct overall result—in other words, get the right result for the wrong reasons. Thus, looking beyond mere outcomes to determine how often a judge applies the correct analysis is an important perspective in determining whether a law can be predictably applied.
We found that district court and PTAB judges not only rarely get the outcome wrong, they also make very few errors in applying the law. When district courts reached the right outcome (i.e., complete affirmance on § 101), the Federal Circuit noted a mistake in the district court’s § 101 analysis a mere 4.5% of the time—and 0% of the time for PTAB judges. There were a mere 7 errors in 153 affirming opinions (excluding Rule 36 affirmances). If looking only to precedential opinions (those written for the bar and interested persons other than the parties), there were 4 errors in 67 opinions, resulting in a comparable 6.0% error rate. Overall, taking into account reversals and vacated decisions, more than 80% of the time for the district court—and 95.5% of the time for the PTAB—the judge’s Mayo/Alice analysis was error free.
This type of granular examination of appellate outcomes has been largely absent from earlier empirical studies, so it’s difficult to put the § 101 error rate in historical context. Still, the low rate of errors in district court and PTO § 101 decisions appears to be remarkably low for an area of law identified to be in crises. Indeed, it appears to be another strong indicator that district courts and the PTO understand how to apply the law, overall.
Federal Circuit Judges Rarely Disagree Regarding § 101 Outcomes
Athena Diagnostics v. Mayo, 927 F.3d 1333 (Fed. Cir. 2019) and American Axle v. Neapco, 966 F.3d 1347 (Fed. Cir. 2020) are § 101 decisions frequently cited as exemplars of what some—including several judges on the Federal Circuit—have argued to be a complete breakdown among the Federal Circuit on how to apply § 101 law. Surprisingly, despite the attention § 101 has received, there have been almost no empirical studies to examine this question on a deeper level.
Yet, in what may be the most surprising finding from our study, in all but a few cases, Federal Circuit judges have shown remarkable agreement (93.5%) in deciding § 101 issues over the past decade. In fact, under this measure of predictability, § 101 proved to be more predictable than the other areas of patent law. In the 368 § 101 cases decided by the Federal Circuit from 2012 to 2022, there were just 24 dissenting opinions relating to § 101. As shown below, the number of cases in which there was a dissenting opinion on § 101 has remained consistently low and peaked in 2019-2020:
Figure 2
Putting the § 101 dissent rates over the past decade in historical and subject matter context further indicates that § 101 law has not been the subject of more disagreement than other areas of patent law.
Figure 3
The summary tables above show that the dissent rate in Federal Circuit decisions involving § 101 over the period 2012-2022 is identical to the rate among all other Federal Circuit decisions, and was lower than in non-101 patent decisions. And while the rate of dissents in § 101 opinions is somewhat higher than in all other opinions that don’t involve § 101, it’s still lower than the dissent rate in non-101 patent opinions generally and nearly identical for patent opinions arising from the district courts—likely because a substantial number of § 101 appeals are summarily affirmed. With that in mind, it’s remarkable that the dissent rate for § 101 decisions (including Rule 36 affirmances) arising from the district courts is actually lower than the court’s dissent rate in appeals from the district courts that don’t involve § 101.
***
More details on the methodology and analysis—as well as additional findings on the types of appeals, procedural posture of decisions, breakdowns by exception type, and invalidity outcomes—can be found in the working draft paper on SSRN: http://ssrn.com/abstract=4380434. In addition to our core findings on predictability, we also provide updated data on § 101 issues studied by previous scholars. Comments are welcome, and can be communicated by email to Jason Rantanen.
[1] Interval Licensing LLC v. AOL, Inc., 896 F.3d 1335, 1348 (Fed. Cir. 2018) (Plager, J., concurring-in-part and dissenting-in-part).
[2] Realtime Data LLC v. Reduxio Sys., Inc., 831 F. App’x 492, 493 (Fed. Cir. 2020) (emphasis added).
[3] Am. Axle & Mfg., Inc. v. Neapco Holdings LLC, 977 F.3d 1379, 1382 (Fed. Cir. 2020) (Moore, J., concurring).
[4] Patent Eligibility Restoration Act of 2023, S. 2140, 118th Cong. § 2(3) (2023) https://www.congress.gov/118/bills/s2140/BILLS-118s2140is.pdf.