Shocked, Shocked To Learn The Patent Office Is Structurally Designed To Approve Shit Patents

The book Innovation and Its Discontents, by Adam Jaffe and Josh Lerner, was first published in 2004. We've cited the book frequently around here, as it did a bang up job describing structural problems with our patent system (and the judicial review of patents). There were a few big points that it made about why our patent system was so fucked up, and a big one was the incentive structure that heavily incentivized approving patents rather than rejecting them.

Specifically, there were two big ideas mentioned in the book about the US Patent & Trademark Office: (1) that because Congress forced the USPTO to fund itself from fees, it had the direct financial incentive to encourage more patent applications, and a good way to do that is to approve a lot more patents and (2) individual examiners were rated and reviewed based on productivity scores on how many patent applications they completed -- and it is much faster and less time consuming to approve a patent, rather than reject one. That's because once you approve a patent it's completed and gone from your desk (and into the productivity metrics as "completed"). But, if you "reject" a patent, it's not done. Even though the USPTO issues what it calls "Final Rejections" there's nothing final about it. The patent applicant can keep going back to the well over and over again, making minor tweaks on the application, requiring the examiner to go through it again. And each time they do, that hurts their productivity ratings. As an additional "bonus" -- the USPTO actually makes significantly more money when it grants a patent, because in addition to application fees, there are also issuance fees and renewal fees.

In the years after that book came out, the USPTO actually seemed to pay attention. It changed how it measured examiners' work and, magically, fewer patents were approved. For a bit. When President Obama appointed David Kappos to head the Patent Office, he decided that the number one problem that the Patent Office had was its huge backlog of patent applications. And, there's no denying that was a problem -- but it was a problem the USPTO created itself by spending the previous dozen years or so agreeing to issue patents on all sorts of crazy things, leading to more applications and more filers hoping to get their own golden patent trolling lottery ticket. So, it was little surprise when soon after Kappos took over, the USPTO started approving patents much more quickly, and a study from 2013 found that (surprise, surprise) it did so by drastically lowering the standards for approving patents.

Now there's a new study with even more empirical evidence showing how the Patent Office's entire structure is designed to incentivize the approval of crap patents (first highlighted by Tim Lee over at Ars Technica). The paper is by professors Michael Frakes and Melissa Wasserman, and they used FOIA (yay!) to get data on millions of patent applications between 1983 and 2010. The key point with that date range is that Congress only switch the USPTO over to funding itself off of fees in 1991 -- so the researchers could look at before and after data. It also allowed them to look at different cross sections within the data.

So, for example, the researchers looked at whether or not there was evidence that the USPTO approved more patent applications when there was a big backlog. The answer: hell yes!

Specifically, we compared theAgency's patent grant rate across different groups of applicantsbased on the tendency of their associated technologies to filerepeat applications; importantly, we performed this across technologycomparison for two groups-defined by their averagetendency to file repeat applications-before and after periods ofbudgetary shortfall and increases in application backlog. Ourfindings suggested that when the Patent Office begins to facemounting backlogs, it appears to act on its incentive to grantpatents at higher rates for technologies that are associated withhigher rates of repeat application.11 In figure 1, we replicate afigure from Frakes and Wasserman (2015), demonstrating thatthe Patent Office indeed began to grant at differentially higherrates for high repeat-filing technologies during the mid-1990s,a moment in time when the Patent Office's application backlogbegan to increase considerably year-by-year. Again, thisanalysis is alarming because it suggests that factors other thanthe underlying quality of applications are affecting the PatentOffice's decision to allow patents.

Then there's a separate question of whether or not the USPTO has a higher approval rate for "profit-maximizing" patents. That is: not all patent fees are the same. Smaller entities get to pay reduced fees. Big companies pay full freight. If the USPTO is being incentivized by fees... then it's likely to approve big company patents faster. And... that's what happened. The study also looked at whether or not the USPTO more readily approved patents in categories where there were higher renewal rates -- meaning a much higher likelihood of generating future fees from renewals. Take a wild guess what they found in both of those studies?

As theory predicts, the Patent Office does indeed grant patentsat notably higher rates to large entities and applicants fromhigh renewal rate technologies when it finds itself in a positionof insufficient fee revenue.9 More broadly, the parameters ofits fee schedule appear to affect the way in which the PatentOffice applies the legal patentability requirements. This isconcerning, given that the granting decision should be basedsolely on whether the application meets the legal patentabilitystandards. If the fee structure were to encourage more patentgrants overall (or more grants during times of budgetaryshortfalls), the result could be the issuance of patents lackinglegal validity, potentially leading to substantial social harms.

Of course, patent system (and patent troll) supporters will cry foul at this for all the usual reasons -- insisting that this is all a conspiracy to take away their patents. But that's ridiculous. Everyone should agree that we're all better off if the Patent Office is not approving low quality patents. I guess everyone... except those with low quality patents. Basically if people are complaining about the results of this study, there's a half decent chance that it's because they hold crap patents.

The report suggests some ways to try to fix these incentives -- or at least to minimize the social cost of them. One is to stop making the PTO so reliant on fees. You can understand why it was appealing for Congress to make the PTO funded by fees in the first place: not having to allocate taxpayer money to an agency seems like a good thing. But it's important to pay attention to what kinds of incentives that can create -- and how it could create serious problems, as it has here. The report suggests raising patent filing fees -- such that they cover the costs of the actual examination, thus making it less likely for there to be budgetary shortfalls that create these warped incentives.

Because the actual fees paid to the Patent Office for theexamination of a patent application are a fraction of the overallcost of securing a patent (which includes attorney fees), thereis reason to believe that even a two-fold or three-fold increasein examination fees will not substantially impede access tothe U.S. patent system. As a bonus, increasing examinationfees will likely also result in raising the quality of patentapplications filed with the Patent Office, as applicants becomemore judicious in selecting those inventions for which theychoose to pursue patent rights.

While that increase in filing fee may feel unfair to some, they also propose doing away with the issuance fee -- so there won't be additional expense for actually getting the patent upfront (there are still renewal fees, but that's much further down the road). As for renewal fees:

Instead of eliminating renewal fees, we recommend thatCongress decouple the renewal fee income from the revenuestream that the Patent Office can immediately access forfunding. While this decoupling goal could be achieved invarious ways, we propose the most straightforward approach:Congress would abolish the requirement that the Agency'saggregate fee income not exceed its operational costs. Renewalfees would then be allocated to a separate fund, similar to thePatent and Trademark Fee Reserve Fund, and earmarked forPatent Office use only. This fund would then be used to providerebates to small and micro-entities. As a replacement for theguaranteed fee discount for any given small- or micro-entityapplication, the Agency's excess renewal fee income would beused to subsidize the small- and micro-entity examination fee.

Perhaps an even more important modification -- and one that's been obvious for many, many years -- is to actually have final rejections of bad patent applications, so people and companies can't just keep refiling over and over again without end. As the paper notes, it's not at all clear what societal benefits allowing repeat filings creates. It overburdens the system, creates incentives for bad patents to get approved, and provides no benefit to anyone other than the entity applying for bad patents. The specific proposal in the paper is that filers can get just one chance to refile the application.

It's good to see more research being done on all of this -- and to see more empirical evidence. What's troubling is that almost all of this was clearly shown more than a dozen years ago, and Congress still hasn't done a damn thing to fix any of it.