Clarity Through Data: A System-Wide Reconstruction of Recent “Serial” IPR Studies

A recurring debate since the passage of the America Invents Act has been about the issue of “serial” or “duplicative” petitions for inter partes review (IPR), with some stakeholders arguing that the filing of multiple IPRs against the same patents and claims is unfair to patent owners and others arguing that the problem has been overstated.

This debate has returned to the forefront over the past few months, as two recent studies have attempted to quantify the extent of the issue. In September 2018, Robins Kaplan attorneys Steven C. Carlson and Ryan M. Schultz published an article¹ that examined the filing behavior of frequent IPR petitioners, observing that a sizable portion of those companies’ IPR petitions are “serial” or “duplicative” in that they overlap by at least one challenged claim. Following references to the Carlson-Schultz study made during a hearing by the Senate Judiciary Committee’s Subcommittee on Intellectual Property, the Computer & Communications Industry Association (CCIA) sent a letter² to the Subcommittee in March that in part contained a reframing of the Carlson-Schultz piece. The CCIA argued that Carlson and Schultz used a series of assumptions that significantly overstated the number of actual “serial” petitions and provided a revised study that it argued better accounts for the “practical realities present in IPR practice”.³

While both studies provided useful discussion points, they are limited by their focus on just the top five petitioners. To provide a more complete picture, RPX has leveraged its hand-curated Patent Trial and Appeal Board (PTAB) data to provide a system-wide reconstruction of both the Carlson-Schultz study and the revised CCIA study, applying the same methodologies to a dataset including all IPR petitions filed by all petitioners against utility patents. As Congress, the USPTO, and PTAB practitioners and litigants continue to consider the “serial” IPR issue, including in Subcommittee hearings, RPX hopes that this reconstruction will provide useful context for policymakers and stakeholders alike.

The Carlson-Schultz Study

The Carlson-Schultz study discussed the issue of “serial” or “duplicative” petitions against the backdrop of 35 USC Section 315(e)(1), an estoppel statute establishing that neither a petitioner whose IPR of a given patent claim reaches a final written decision or its real party in interest or privy may “request or maintain a proceeding before the [USPTO] with respect to that claim on any ground that the petitioner raised or reasonably could have raised during that inter partes review”. Carlson and Schultz interpreted this provision as meaning that “a party is entitled to a single written decision as to a challenged patent claim” and that “once there is a final written decision as to a patent claim, that petitioner may not ‘request or maintain’ another proceeding as to that claim”. Under this interpretation, they then question “whether the PTAB should ever issue multiple written decisions against a claim that has been repeatedly challenged by a petitioner; and if not, whether a single petitioner should be allowed to take to trial [or even institute] multiple petitions against a single claim”.

Carlson and Schultz then defined a “duplicative” petition as one that “challenge[s] at least one claim that is the subject of attack by another . . . petition” from the same petitioner or named real party in interest. Applying this definition, they analyzed the petitioning practice of five companies that they identified as the “top five filers of IPRs”: Apple, Samsung, Google, Microsoft, and LG Electronics. Examining a dataset consisting of petitions filed from the earliest availability of IPR through June 23, 2018, they asserted that “it is clear that serial, overlapping petitions are commonplace at the PTAB”. This analysis, by Carlson and Schultz, claimed that more than 56% of Apple’s petitions are duplicative; with 39% for Samsung, 39% for Google, 60% for Microsoft, and 35% for LG Electronics.⁴

The CCIA Reframing

The CCIA’s opposing analysis argued that Carlson and Schultz’s definition of a “duplicative” petition was overly broad, identifying four factors that it argued had resulted in overcounting and providing its own analysis accounting for those factors.

• First, the CCIA noted that Carlson and Schultz had counted as “duplicates” all petitions directed to the same patent, including the original petition in a series—whereas in the CCIA’s view “the first petition against a patent is inherently not duplicative”. “Eliminating this overcounting reduces the study’s duplicative petitions count by 42%” of the 1,139 petitions considered, the CCIA argued.
• Second, the CCIA disagreed with Carlson and Schultz’s decision to count petitions as duplicative where the challenged claims “overlap only slightly”, noting that “strict word limits” often require petitioners to file multiple petitions—particularly for “large patents with many claims or elements”. The CCIA thus excluded petitions that overlapped on only a single claim, explaining that a petitioner “may effectively be required to include an overlapping claim when different dependent claim sets are best addressed by different prior art” (meaning that the “inclusion of the common independent claim” may be required).
• Third, the CCIA asserted that Carlson and Schultz should not have counted multiple petitions filed “simultaneously and near-simultaneously (within several days of the first-filed petition)” as duplicative, noting that petitioners may also legitimately file multiple petitions at once or in close succession when dealing with large claim sets and word limits. Its revised analysis thus excluded such petitions.
• Fourth, and finally, the CCIA argued that Carlson and Schultz should have excluded follow-on petitions, which are “carbon copies” of earlier-filed petitions and are accompanied by motions for joinder to those proceedings, and did so for its revised analysis—explaining that these petitions are “explicitly contemplated by the statute” and place no additional burden on the patent owner or the USPTO.

Overall, the CCIA found that a much smaller share of the petitions brought by the top five filers were “duplicative” under this revised methodology, with the percentage of such petitions dropping as follows:

RPX’s Reconstruction of Both Analyses

RPX recreated both the Carlson-Schultz and CCIA studies using its own PTAB database, which includes data from the PTAB’s End to End system and is systematically and regularly supplemented with human-reviewed RPX metadata. This dataset includes 9,368 petitions, challenging 5,704 distinct utility patents, filed as of May 31, 2019; measured by petitions filed by unique petitioners, the total is 12,284; and measured by petitions filed by unique petitioners or real parties in interest—as was the approach used in the Carlson-Schultz study—the total is 18,469. In other words, the unit of measure of the Carlson-Schultz and CCIA studies, as well as the RPX reconstruction, is the count by case and petitioner or real party in interest.

As part of reconstructing the two analyses, RPX made certain assumptions as a result of methodological ambiguities present in the prior studies as summarized by their respective authors. For example, the CCIA study excluded petitions filed simultaneously and near-simultaneously, but for the latter the precise timeframe was uncertain, having been specified only as “several days”. RPX considered petitions to be filed near-simultaneously if no more than two calendar days had passed between petitions filed within a given series. Finally, RPX considered all joined petitions to be follow-on petitions when excluding that category.⁶

After confirming that the results of these “reconstructed” analyses substantially matched their original versions,⁷ RPX then applied the same methodologies to a dataset that included all patents subjected to IPR since its inception through May 31, 2019. (The Carlson-Schultz study included IPRs filed through June 23, 2018, so the RPX reconstruction also includes an additional 11 months of petitions.)

RPX’s analysis shows that of the roughly 18.5K IPR petitions filed by unique petitioners or real parties in interest as of May 31, just under 6.5K—or 34.8%—of those petitions would be considered “serial” under the methodology applied by Carlson and Schultz.

RPX then excluded petitions that were considered to be overcounted under the CCIA’s methodology, including 2.8K first-instance petitions⁸ (i.e., original petitions in a series, which the CCIA called “inherently not duplicative”) and 137 petitions that overlapped others by a single claim. RPX also excluded about 2K petitions filed simultaneously or near-simultaneously as others in a series (as defined above) as well as slightly over 600 follow-on petitions.

This left roughly 1K petitions out of the 18.5K total—about 5.4%—that would be considered “duplicative” under the CCIA’s corrected version of the Carlson-Schultz study, compared to the 34.8% that would be duplicative under the Carlson-Schultz methodology.

The waterfall graphic below provides a breakdown of RPX’s two analyses of the full 18.5K-petition dataset, with the analysis applying the Carlson-Schultz criteria on the left, the analysis applying the CCIA’s rubric on the right, and the exclusions made under the CCIA rubric each highlighted separately in between.

A comparison of RPX’s results for the top five petitioners to those for the entire set of petitions shows the following:

• Under the Carlson-Schultz rubric, 46.3%⁹ of petitions filed by the top five filers are considered “duplicative”, while 34.8% are duplicative under the full dataset. This suggests that the Carlson-Schultz methodology overestimates duplicative petitions for the entire set of petitioners and real parties in interest when applied to the top filers alone.
• In contrast, under the CCIA’s methodology, 4.8%¹⁰ of petitions brought by the top five filers are duplicative, versus 5.4% for the full set of petitions. Thus, the CCIA’s criteria, when applied to the top five filers, produce a result that is modestly lower than the share of duplicative petitions for the entire dataset.

In all, RPX’s reconstructed analyses highlight the impact that basic definitional questions can have on the analysis of filing patterns and other quantitative issues and underscore the comparative value of applying a given methodology to as large of a dataset of possible. To that end, RPX will continue to leverage its data to answer a variety of other questions of interest to industry stakeholders in the months ahead.

For questions on this analysis, please contact reports@rpxcorp.com.

Footnotes

1. Carlson & Schultz, “Tallying Repetitive Inter Partes Review Challenges”, Law360 (Sept. 14, 2018).
2. Letter from Joshua Landau to Sens. Thom Tillis and Chris Coons (Mar. 20, 2019).
3. ld. at 4.
4. The Carlson-Schultz study appears to have listed the “duplicative” percentages with the decimal truncated; as such, in some cases, there is a 1% variance between those numbers and the ones presented here, which come from the same absolute values but were rounded.
5. The CCIA rounded their decimals with the exception of the Carlson-Schultz percentage for Microsoft, for which the decimal appears to have been truncated; as a result, the percentage the CCIA listed for Microsoft is 1% lower than the rounded number presented here.
6. RPX considered petitions to be follow-on petitions whenever a motion to join was granted; counting motions, regardless of outcome, would result in a higher number of follow-on petitions.
7. The data used in the Carlson-Schultz study, which the CCIA also cited, incorrectly identified the petitioners for a small number of IPRs, among other issues noted. Where RPX found discrepancies, we confirmed and corrected them via manual review. This resulted in minor differences in petition counts for both studies when recreated using RPX data.
8. The order in which duplicates are evaluated is significant. For example, if a first IPR challenged claim 1, a second IPR challenged claim 2, and a third IPR challenged claim 1 and 2; evaluated in that order, there is one duplicate; however, if the third IPR came before the second or first, there would be two duplicates. RPX ordered the IPRs for analysis by their case number in ascending order to ensure that they were counted in the order they were filed.
9. This represents RPX’s independently calculated results for the top five filers, which include certain corrections and methodological assumptions as documented earlier.
10. ld.