Federal Circuit: Alice May Be Hard to Apply, but District Courts Must Show Their Work

October 29, 2020

The Supreme Court’s Alice decision is widely considered to be difficult for courts to apply consistently, a notion regularly acknowledged by judges working through that case’s two-step eligibility test. But as the Federal Circuit recently held, neither the complexity of Alice nor the fact that Section 101 decisions are reviewed de novo on appeal provide an excuse for district courts to skimp on their written analyses. On October 23, the appeals court reversed and remanded an Alice decision by Delaware District Judge Colm Connolly that invalidated five data compression patents asserted by Realtime Data LLC, holding that “this case presents one of those rare circumstances in which a district court’s treatment of a complex and close legal issue is too cursory to allow for meaningful appellate review”.

The decision just overturned was handed down in four separate cases, against Aryaka Networks (1:18-cv-02062), Fortinet (1:17-cv-01635), Reduxio Systems (1:17-cv-01676), and Panzura (1:18-cv-01200)—just a few of the numerous lawsuits filed by Realtime Data in a sprawling campaign that began in 2008 and has since hit more than 130 defendants. Those four companies each filed Alice motions at various points throughout 2018 and 2019, collectively challenging the five patents at issue across their cases (7,415,530; 8,717,203; 9,054,728; 9,116,908; 9,667,751), with Aryaka, Fortinet, and Panzura joining motions already filed in other lawsuits. Reduxio, for its part, alleged in its own motion that the claims of the five patents fail under Alice because “(a) they are directed to the results or effect of abstract ideas involving content dependent/independent compression/decompression untethered to any specific process or device, and (b) can be performed by a mental process, an algorithm, and/or a pen and paper”, all while reciting “only conventional computer components and processes”.

The court scheduled a joint hearing on those motions in July 2019, and asked each party, around a month before that date, to identify in a letter to the court “which Supreme Court or Federal Circuit case that party contends is most similar to the challenged claim(s)”. The defendants chose the Federal Circuit’s April 2017 decision in RecogniCorp v. Nintendo (2016-1499), while Realtime Data selected the August 2017 opinion in Visual Memory v. NVIDIA (2016-2254). The court then held its hearing as scheduled, with that proceeding lasting—as the Federal Circuit later emphasized—just over two hours, including a recess, and for which the “complete oral analysis of the patent eligibility of the 159 claims in the asserted patents fills only five pages of transcript” (per the appellate ruling). At the end of the hearing, Judge Connolly announced that he was “prepared to rule on the pending motions”, informed the parties that he would “not be issuing written opinions”, and stated that he had “‘followed a thorough process before making the decision,’ including considering the briefing and engaging in oral argument”.

It was this lack of explanation, in part, to which the Federal Circuit objected in its October 23 decision on appeal, which was authored by Circuit Judge Kathleen M. O’Malley. While acknowledging that as a result of Alice’s complexity and the specter of de novo review, “district courts might be tempted to opt for an effective coin toss rather than a reasoned analysis when faced with a challenge under § 101”, the Federal Circuit argued that “the system is not supposed to work that way. The parties are entitled to more and the Court of Appeals needs more”. The reason, explained the appeals court, is that a “district court opinion ‘must contain sufficient findings and reasoning to permit meaningful appellate scrutiny’”. “De novo appellate review certainly does not justify resolving a complex legal issue without an opinion or reasoned analysis”, the Federal Circuit further emphasized—finding that, “[u]nfortunately, that is exactly the type of improper justification the district court proffered in this case”.

On that front, the Federal Circuit found that Judge Connolly’s decision—limited solely to the hearing transcript by reference—lacked the requisite findings and legal analysis. What the record did show through the transcript led the Federal Circuit to conclude that Judge Connolly’s reasoning was either unclear, or possibly even problematic.

The Federal Circuit identified four key issues presented by this limited record. First, it found that “the colloquy between the district court and Realtime leaves us unclear as to the true basis for the district court’s decision”—noting that rather than focusing on whether the claims were “directed to” an abstract idea, the court appeared to focus more on whether certain claim limitations were “novel”. Additionally, the Federal Circuit thrice highlighted Judge Connolly’s observation “that the claimed invention of the ’751 patent was ‘obvious’—but not in the patent sense, ‘like it’s common sense’”. While noting that novelty and nonobviousness may come into play at certain points of the Section 101 inquiry, it held that “they are rarely issues appropriate for resolution on the pleadings”—particularly “where, as here, Realtime repeatedly contested as factually incorrect propositions posited by the district court”. If those issues were ultimately relevant, continued the appeals court, it would be at step two of the Alice test (which focuses on whether a claim contains an inventive concept) and not step one (whether a claim is “directed to” an “abstract idea”). In the absence of a more fulsome record, ruled the Federal Circuit, “it is difficult to discern what part of the court’s concerns with the claims were directed to which step of the Alice analysis”.

Second, the Federal Circuit indicated that “to the extent the court purported to answer the Alice step 1 ‘directed to’ question, it is unclear that it did so correctly”—focusing on the fact that Judge Connolly appears to have not identified a representative claim, in which case it was “incorrect to consider whether a patent as a whole is abstract”. To the extent that Judge Connolly’s “analysis simply generalized the claims”—which the Federal Circuit posited that it did, as it apparently omitted “key aspects of the claims”—the appeals court held that this would have been a legal error.

Third, the Federal Circuit faulted Judge Connolly for not “address[ing] or even acknowledg[ing]” a pair of recommendations issued by Magistrate Judge John D. Love of the Eastern District of Texas that “considered the patent eligibility of the ‘728, ‘530, and ’908 patents, as well as the eligibility of members of the ’203 and ’751 patents’ families, in two separate cases” filed by Realtime Data against Carbonite and Actian, recommending that the patents be deemed eligible. In both cases, remarked the Federal Circuit, the presiding judge adopted Judge Love’s recommendations: “Judge Robert W. Schroeder III of the Eastern District of Texas and, due to an intervening transfer, Judge William G. Young of the District of Massachusetts—each with significant experience in patent cases”. Although the Federal Circuit noted that Judge Connolly was under no obligation to follow those conclusions, it held that he “should have, at a minimum, provided a considered explanation as to why those judges were wrong”. Rather than expressly or implicitly addressing Judge Love’s reasoning, “the court recited a series of legal conclusions and § 101 cases, without analysis. That simply was not enough”, underscored the Federal Circuit.

Fourth, the Federal Circuit took issue with the fact that Judge Connolly did not “address or distinguish” the Visual Memory case cited by Realtime Data as exemplary at the court’s request, nor any related case law. Moreover, while Judge Connolly did acknowledge the defendant’s citation of RecogniCorp, the Federal Circuit flagged his short summary of that case’s holding—and his conclusion that “[t]hat’s what we have here”, with no further explanation—as problematic in their brevity. Although it stated that it was not taking a position on the merits of either parties’ chosen case analogies, the Federal Circuit “merely note[d]” that on this issue, “the district court did not do enough”.

Circuit Judge Alan D. Lourie concurred, arguing that “[t]he foundation of a proper determination of the eligibility of claimed subject matter under 35 U.S.C. § 101 and the framework of [Alice] . . . is an accurate identification of the focus of the claimed advance at Alice’s Step 1, to be followed (if necessary) by an accurate identification of all specifics of the claims at Alice’s Step 2”—without overgeneralizing the claims, as the court did here. “In the present cases”, continued Judge Lourie, “the district court erred at the foundational stage”, as its “truncated characterization of claim 1 of the ’728 patent, and of some or all of the other claims at issue, created an incorrect starting point for the required analysis”, with similar errors apparent in the “colloquies between the district court and counsel”. By “disregard[ing] limitations, in at least some of the patent claims at issue, that are part of the focus of the asserted advances”, Judge Connolly “overgeneralized, or oversimplified, the claims in [a] fundamental way” and thus “failed to conduct the inquiries required under the branch of § 101 doctrine relevant here”—which, to Judge Lourie, justified the “unusual step” of reversing the court’s ruling without first addressing it de novo.

While the Federal Circuit’s decision here is nonprecedential, it may nonetheless give pause to judges tempted to issue shorter rulings on Alice—especially those inclined to rule on such eligibility disputes from the bench. More to the point, the appeals court’s opinion may be of particular relevance for the District of Delaware, which has on occasion seen its judges engage in a practice commonly known as a “Section 101 Day”—wherein on a given day, a district judge will hear multiple Alice motions from unrelated cases and issue oral rulings from the bench, one after another. Orders issuing from such motions have tended to limit their analysis to language incorporated from the hearing transcript.

For details on one such omnibus hearing, see “Third ‘Section 101 Day’ in Delaware—This One Not Held by Judge Stark—Leads to Disparate Results” (October 2019).