The statute describing the Director’s authorization of institution of an inter partes review (IPR) has been interpreted broadly to allow the Patent Trial and Appeals Board (PTAB) to deny institution of an otherwise meritorious IPR. See 35 U.S.C. 314(a) (“may” institute). The USPTO has used that reasoning to fit an elephant through the keyhole; indeed, nearly 20% of all institutional decisions in 2021 were denied for procedural reasons. See PTAB Procedural Decisions Report: Non-Merits Based Decisions Rise, Denials Fall (Feb. 6, 2022) (https://www.unifiedpatents.com/insights/ptab-procedural-decisions-report). Unsurprisingly, section 314(a) “discretion,” as interpreted by the agency in Fintiv, has been used more frequently than any other statutory provision in issuing non-merits denials. 

Highlights:

• In 2021, more than 45% of all IPR institution decisions (420 of 918) spent considerable time and resources dealing with the discretionary considerations of the PTO’s NHK-Fintiv rule.

•  In these, the Board on average used more than 8 pages to analyze the NHK-Fintiv factors when instituting—representing more than 20% of their page count—and more than 11 pages when denying based on Fintiv– representing more than 50% of their page count. This represents 8.37 pages of all institution decision pages where NHK-Fintiv comes up.

• Overall, the percentage of Board time spent addressing the NHK-Fintiv factors in deciding to institute an IPR is 22.63%.

But the impact goes far beyond just the denials; of all cases where institution was decided in 2021, more than 45% (420) of all of the Board’s 918 institution decisions in 2021 spent considerable time and resources dealing with the newly posed question of Fintiv, at length.  When looking at institution decisions that moved to institute, the percentage go up; of the decisions moving to institute (coincidentally, also 42), more than three-fourths—77% (324)—dealt with Fintiv at length, as detailed below.

On average, the Board analyzed Fintiv for more than 8 pages in institutions (and more than 11 in denial), representing on average more than 50% of the Board’s workload for denials, and more than a fifth of its workload for institutions. It should be noted that this does not include any settlements, withdrawals, requests for adverse judgment, or any other outcome besides Instituted, Not Instituted on the Merits, Not Instituted Procedurally.

Indeed, in 2020, Sections 314(a) and 325(d) formed the basis of about 95% of the PTAB’s procedural denials—and 73% of all procedural denials were based on 314(a) alone. Board denials under Section 314(a) have been increasingly reliant on the precedential decisions of NHK Spring Co., Ltd. v. Intri-plex Technologies, Inc. in 2018 and Apple Inc. v. Fintiv, Inc. in 2020. Initially, NHK denied the institution of an IPR based on Section 314(a) in view of the advanced stage of a parallel district court litigation.  Following NHK, Fintiv outlined a six-factor test (now called the Fintiv Factors) when considering a non-merits denial under Section 314(a).  

Figure 2 below shows the significant rise in procedural denials over the last few years. Including, a majority of those procedural denials were on the basis of Section 314(a), with a significant rise since the NHK and Fintiv decisions in 2018 and 2020, respectively.

The six Fintiv Factors are weighed against each other, and are:

1. Whether the court granted a stay or evidence exists that one may be granted if a proceeding is instituted

2. Proximity of the court’s trial date to the Board’s projected statutory deadline for a Final Written Decision

3. Investment in the parallel proceeding by the court and the parties

4. Overlap between issues raised in the petition and in the parallel proceeding

5. Whether the petitioner and the defendant in the parallel proceeding are the same party

6. Other circumstances that impact the Board’s exercise of discretion, including the merits.

For Factor 1, as stated by the Fintiv panel itself, “[a] district court stay of the litigation pending resolution of the PTAB trial allays concerns about inefficiency and duplication of efforts.” Thus, a stay of the district court litigation “has strongly weighed against exercising the authority to deny institution.” Regarding Factor 2, “if the court’s trial date is earlier than the projected statutory deadline” for a Final Written Decision, Factor 2 has often weighed in favor of exercising authority to deny institution. However, “if the court’s trial date is at or around the same time as the projected statutory deadline…or even significantly after the projected statutory deadline, the decision whether to institute will likely implicate other factors.” For Factor 3, the panel analyzes “the amount and type of work already completed in the parallel litigation,” with a particular emphasis on whether the district court has already entered any substantive orders. As for Factor 4, when an IPR petition contains “the same or substantially the same claims, grounds, arguments, and evidence” as the district court litigation, there are “concerns of inefficiency and the possibility of conflicting decisions.” Regarding Factor 5, when the defendant in the parallel district court proceeding is the same as the petitioner in the IPR, this factor has generally weighed in favor of discretionary denial. Lastly, Factor 6 acts as a catch-all provision for any case-specific arguments that may weigh for or against the Board’s exercise of discretion.  

The raw data for this analysis was scraped from published opinions from the PTAB website. All PTAB institution decisions from 2021 that cited the Fintiv factors in the opinion were scraped. The Unified Patents team focused on 2021 out of convenience. After downloading the qualifying opinions, we noted the following data points for each qualifying opinion: a) the total pages of the opinion; b) the pages spent focusing on analyzing the Fintiv Factors; c) the outcome of each factor such as favorable to denial, unfavorable to denial, or neutral; and d) the overall outcome of whether the petition was denied or instituted. It should be noted that Unified solely focused on the first 5 Fintiv factors, with the catch-all provision being ignored since it could not be classified easily. 

When looking at the length of institution decisions, a natural pattern was evident. On average, granted institution decisions were 20-25 pages longer than any other outcome. In procedural denials under Fintiv, the Fintiv sections typically ran under 12 pages and encompassed, on average, more than half of the decision itself. While the number of pages does not conclusively denote the outcome, it can generally provide an indication of where the board places priority in the decisions. 

Looking at the overall context in 2021, 77% (324 out of 420) of decisions Instituted considered Fintiv. Turning to Not Instituted on the Merits, unsurprisingly, only about 4% (10 out of 310) of those analyzed Fintiv at all, and as illustrated above; when they did, these typically involved a short one-page analysis suggesting further analysis wasn’t necessary, as the Board was deciding on the merits. Looking at Not Instituted Procedurally, 48% (86 out of 178) of these cases have decided Fintiv. It is clear, looking at the page counts and the Fintiv decisions in the individual outcomes, that Fintiv is being addressed at length in nearly half of all Board decisions.

Turning to the individual factors, we broadly categorized them as Favorable, Neutral, or Unfavorable. Favorable in this context means the Board found the factor favored institution; Neutral means the Board found it had no impact on the institution decision; Unfavorable means that the Board found the factor disfavored institution. Each outcome (Instituted, Not Instituted - Merits, and Not Instituted - Procedural) were given a number of 1, 2, and 3 respectively.  In sum, 420 cases in 2021 were found to have mentioned Fintiv or use the factors. Of those, 324 cases were instituted, 10 were nonetheless not instituted (merits), and 86 cases were not instituted (procedural). 

Unified then used a feature selection method, called information gain, to understand how the five Fintiv factors were leading to various outcomes. Information gain calculates the reduction in entropy or surprise from transforming a dataset in some way. It is commonly used in the construction of decision trees from a training dataset, by evaluating the information gain for each variable, and selecting the variable that maximizes the information gain, which in turn minimizes the entropy and best splits the dataset into groups for effective classification. Information gain can also be used for feature selection, by evaluating the gain of each variable in the context of the target variable. In this slightly different usage, the calculation is referred to as mutual information between the two random variables. Unified selected the latter to understand how each factor or variable influenced the outcome. 

In a simpler way, think of the factors as different branches in a tree that lead to various outcomes or the leafs if you will. Depending on the factor influence, Favors, Neutral, and Unfavorable, this changes the probability of getting to a certain outcome. The driving factor was first determined by using information gain and then a sub factor was found that influenced the decision.
Using this approach, Factor 2 - Proximity of the court’s trial date to the Board’s projected statutory deadline for a Final Written Decision, played the greatest influence on the decision. In fact, if Factor 2 were found to be unfavorable, there is a 20.89% chance of a procedural denial based on Fintiv. Interestingly enough, even when the factor weighed in favor of institutions there was still a higher chance of procedural denial based on Fintiv at 28.18%.

Going down the branch further, the same test to find the root factor was then applied with the remaining factors. Surprisingly across all factors, Factor 1 then played the biggest role in determining the outcome of the institution. Factor 1 looks to see whether the court granted a stay or evidence exists that one may be granted if a proceeding is instituted.

What seems counterintuitive is that, even when Factor 1 is found to be in favor of the institution, and Factor 2 is found to be in favor, there is an almost 37% chance of procedural denial. Even neutrality on Factor 2 still yields an almost 27% chance of being denied procedurally. What is not surprising is that there is nearly a 29% chance of being denied procedurally when Factors 1 and 2 are both found to be unfavorable, and Factor 1, at least by the numbers, tends to be a nonfactor or largely unfavorable, as district courts will not normally have stayed the case and the Board normally refuses to speculate. This begs the question of how effective the Fintiv framework is, and if the framework is even objective. Looking purely on a mathematical basis, these counterintuitive probabilities of procedural denials resulting from consideration of Fintiv Factors 1 and 2 raise more questions than actual answers.

© 2022 Unified Patents, LLC. All Rights Reserved.