Over the last year, several Federal Circuit judges have filed opinions lamenting the state of the case law that interprets the abstract idea exception to patent eligibility under 35 U.S.C. § 101. For example, Judge Linn wrote in Smart Systems that the “problem” with the Supreme Court’s test “is that it is indeterminate.” Similarly, Judge Plager wrote in Interval Licensing that the “incoherent body of doctrine” applying Alice makes it “near impossible to know with any certainty whether the invention is or is not patent eligible.” Many practitioners agree that § 101 decisions are difficult to predict and the state of the case law exacerbates this difficulty. This article considers affirmance data to explore whether it supports the premise that § 101 decisions are not just difficult but fundamentally “indeterminate” or “near impossible” to resolve.
One set of data shows the Federal Circuit’s affirmance rate in appeals involving § 101, as compared to patent appeals in general. Much of this data comes from a recent law review article by Paul R. Gugliuzza and Mark A. Lemley. (Judge Plager cited a footnote from the article in his Interval Licensing opinion but did not address the article’s empirical data.) Another set of data shows affirmance rates in ex parte patent appeals before the Patent Trial and Appeal Board (“PTAB”). This data comes from Juristat’s online patent analytics tool.
The likelihood of affirmance on appeal is a narrow but useful measure of predictability. Judge Plager’s Interval Licensing opinion seems to recognize the importance of this metric. For example, his opinion argues that the abstract-idea doctrine “falls short in the sense of providing a trial judge with confidence that the judgment will be understood by the judges who come after”—i.e., the appellate judges who “have the final say in the matter.”
If § 101 decisions were not just difficult but truly indeterminate, one might expect such decisions to be reversed more frequently than decisions involving other legal issues. Indeed, appellate judges should have no trouble reversing § 101 decisions that they regard as incorrect, because such decisions are mostly reviewed de novo. But the data indicates that § 101 decisions are affirmed more frequently than other types of decisions. Thus, the data does not seem to support the premise that § 101 decisions are indeterminate.
Those who are skeptical of § 101 decisions might argue that their high affirmance rate is a result of their indeterminate nature, not their predictability. After all, if a question has no objectively correct answer that can be articulated as a rationale for reversal, an appellate judge may be more likely to let the lower court decision stand. In practice, however, this argument is implausible. Even Judge Linn and Judge Plager, in their opinions criticizing § 101, stated that they were bound by precedent to concur as to the invalidity of at least some claims. These judges’ conclusions that precedent dictated specific outcomes—outcomes that both judges wished to avoid, in fact—undermines the notion that the § 101 analysis cannot yield definitive answers.
Federal Circuit Appeals
The Federal Circuit’s affirmance rate in § 101 cases is markedly higher than its overall affirmance rate for all patent cases. Specifically, for the three-year post-Alice time period covered by Gugliuzza and Lemley, the Federal Circuit’s affirmance rate in § 101 cases was about 88%. In contrast, the Federal Circuit’s overall affirmance rate in patent cases last year was about 75%.
The overall affirmance rate cited above comes from an article that provides data for 2017 only. Data for prior years is available in two other studies, but these studies use a different counting methodology that makes them less useful as a basis for comparison. Specifically, the studies count precedential opinions only. They omit non-precedential opinions and summary affirmances under Rule 36, so the affirmance rates they present are understated. Still, the studies provide further evidence that the Federal Circuit’s affirmance rates for § 101 cases are at least as high as the court’s overall affirmance rates. The first study—covering precedential cases from 2015 and 2016— shows a § 101 affirmance rate of 57%, compared to an overall affirmance rate of 56%. More dramatically, the second study—covering precedential cases from 2016 and 2017—shows a § 101 affirmance rate of 69%, compared to an overall affirmance rate of 59%.
The data regarding Federal Circuit affirmance rates is summarized in the following table. As the table shows, the affirmance rates in § 101 cases consistently exceed the overall affirmance rates for all types of patent cases.
Additional data on the predictability of § 101 decisions is available in the context of PTAB appeals. These appeals arise from examiners’ rejections of pending patent applications. Juristat’s patent analytics platform tracks the outcomes of such appeals.
Among many other useful tools, Juristat can aggregate appeal data for individual art units, USPC/CPC classes, and examiners. The data presented below focuses on USPC 705, which covers the art units with the most Alice rejections.
Juristat also tracks the types of rejections issued in each application. Thus, it is possible to compare the PTAB’s affirmance rate in applications with Alice rejections to the overall affirmance rate for a class (here, USPC 705). One such comparison is shown below. Note, however, that even when an application has faced an Alice rejection, that rejection may not have been at issue on appeal. Thus, the data presented below provides only a rough approximation of the PTAB’s affirmance rate for Alice rejections.
The comparison above is limited to patent applications with a disposition date in 2018, which means that the applications were allowed or abandoned this year. Juristat provides data for previous years also, though such data is omitted from this comparison to avoid capturing any appeals arising from pre-Alice rejections.
The Juristat data suggests that, in PTAB appeals as in Federal Circuit appeals, the affirmance rate for Alice rejections exceeds the overall affirmance rate.
The empirical data presented above belies the notion that it is impossibly difficult to predict the ultimate outcome of the Alice analysis. Such predictions may be difficult in many cases and uncertain in some, but the same is true for many issues in patent law, such as obviousness and claim construction. Whatever flaws the Alice analysis may suffer from, the data above regarding likelihood of affirmance does not show a clear basis for the widely held sense that Alice outcomes are especially unpredictable.