RPX Data Disputes Bias Allegations

As I explained a few months back,  a due process challenge is pending at the CAFC that challenges the PTAB’s fee model.  New Vision Gaming v. SG Gaming Inc.  The argument, in a nutshell, is that because the fees for AIA trial proceedings include a separate component for the trial, and that collection of this fee makes up about 40% of the PTAB’s yearly budget, that this amounts to “a structural bias unlike any other in the federal executive branch.” (My take on that theory is found in the earlier post).

An amicus filing made in New Vision further alleges that the PTAB demonstrates a bias only in October?

The amicus (here) argues that there is an “October Effect” whereby APJs are supposedly more likely to institute less meritorious petitions in October than in September—the first and last months of the USPTO’s fiscal year, respectively, which align with the conclusion of the APJs evaluation periods.  The amicus suggests that APJs are “stuff[ing] the pipeline” in the beginning of each fiscal year to “guarantee … future work”. Specifically, it claims to find with “statistical clarity” a significantly higher percentage of “questionable institution” decisions made in October versus September as support for its thesis.

Before even reaching the new RPX study, this is plainly a faulty premise.  First, it’s not as if all PTAB trials conclude in September and new ones have to start in October to refresh an empty APJ docket.  There is simply no need to “fill the pipeline ” in October over any other month of the year.  Trials end and new ones are instituted all year round. Moreover the PTAB was actively hiring APJs in virtually every year of this study and has never had a problem with workload, or a single reduction in force.  There would be no reason for an individual APJ to baselessly institute trials for job security let alone hundreds of APJs.  In fact many APJs also handle ex parte appeals that have (over the same time span) had up to a two year backlog of inventory. There has never been a shortage of work at the PTAB.

As to the amicus statistics, finding one claim patentable in a group of challenged claims is argued as somehow demonstrating a questionable institution decision. The brief explains:

[A]fter final outcomes are known, we can go back to test the quality of institution decisions. If institutions during some time periods tend to result in proportionally more final decisions favorable to the patentee, and institutions during other time periods result in proportionally more final decisions favorable to petitioners, and those time periods correlate with APJs’ annual salary reviews, a reasonable person would question whether the compensation cycle influences APJ likelihood to institute.

This makes no sense whatsoever.

The standard for trial is based on one claim, not EVERY claim.  The fact that a single claim of a group of progressively narrow claims survives is evidence of nothing but the PTAB’s thoroughness and evenhandedness.

Likewise, this data compares apples to oranges.  Every PTAB trial is not equal in issues, includes equally skilled counsel, or even applies equal effort to every patent of a portfolio – some are just more important than others.  The argument also assumes that if an APJ institutes and doesn’t cancel EVERY claim, that they are somehow proving their earlier work to be questionable.  An odd message from an inventor advocacy group. This is the way the system is supposed to work, institution is preliminary, not final.  Finally, the brief seems to assume that denials of institution are somehow worth less to an APJ on a productivity scale than a decision to institute —that’s simply incorrect.

Even accepting the premise that work product in September is slightly more reliable (based on the definition above) than October demonstrates nothing more than typical docket management practices.  That is, APJs trying to meet the end of the productivity period in September will invariably pull low hanging fruit off the docket first.  More difficult, issue laden cases will be pushed to October (deadlines permitting), this is hardly evidence of any form of bias.

As to the numbers themselves, RPX has done a fine job of breaking this down and explaining a more fulsome view of PTAB statistics.  Well worth the read.

You can find the RPX study (here).

RPX concludes that:

Between 2013 and 2018, the PTAB received 9,533 petitions, instituted 4,432 trials, and issued 2,424 FWDs. In contrast to [the amicus] surprising conclusion about the existence of an “October Effect”, RPX’s analysis of the institution decisions made by the PTAB from 2013 through 2018 demonstrates a remarkable consistency in the PTAB’s institution decisions as measured by outcomes when FWDs are reached, trial institution rates, and institution decision pendency.

Political polls and statistics, good times.