Inspector General’s Hyperbolic Report Distracts From Improving Patent Quality

The Office of the Inspector General for the Department of Commerce recently released a report, “Analysis of Patent Examiners’ Time and Attendance,” alleging significant time and attendance abuse by patent examiners. This report seems to be intended to generate headlines rather than identifying real problems. In fact, the OIG missed a real problem that was right under their noses, specifically an extremely poor network infrastructure. According to the OIG’s own data, network outages caused more than triple the time loss of the “abuse” the OIG identified.

More importantly, the OIG’s report misses the point. Nearly everyone agrees that the quality of patents being issued is a problem, but the OIG didn’t consider patent quality at all. The OIG’s proposed solution of forcing examiners to produce more quickly is hardly a recipe for quality, particularly when examiners report that they don’t currently have enough time to do a thorough job.

In contrast, the Government Accountability Office recently issued two reports on improving patent quality. The GAO looked at internal procedures and surveyed examiners in order to identify some of the real causes of poor patent quality. The GAO had a number of important recommendations for improving patent quality; the distraction of this “fraud” may mean that we waste a real opportunity to make some positive changes.

If there is something good to come from the OIG’s report, it’s that there is a need for better direction and supervision of examiners. But the focus of that supervision should be on consistency of examination and producing high quality patents. The abuse that the OIG identified is fairly small-scale, but poor patent quality is a major drag on the economy. That’s where we should be putting our efforts.

The OIG’s Report Exaggerates the Extent of Abuse

Here’s the OIG’s summary:

For the 9-month period, the OIG reviewed specific work activities of approximately 8,100 patent examiners and identified 137,622 unsupported hours. This equates to a one-year average of nearly 180,000 unsupported hours. For the 15-month period, the OIG analyzed work activities for roughly 8,400 examiners and identified 288,479 unsupported hours.

Those numbers sound big. But they omit something important: how many hours were actually claimed? That is, what portion of claimed hours are unsupported? In order to find part of that answer, you’ll have to look at footnote 21 on page 9 of the report:

This number represents 1.6% of the total computer-related examining time. The total number of hours associated with computer-related activity over the 9-month period is 8,673,490.

Over 9 months, examiners are accused of fraudulently claiming less than 2% of hours worked. It’s disturbing that the OIG chose to bury that context in a footnote; if the OIG has no agenda, why not be up front about the scale of the problem? (For the 15 month period, the OIG notes that the percentage was about 2% as well.)

There’s probably even less time loss than the OIG claims. The review did not track voluntary overtime and it failed to account for people shifting time from one day to another.

“Voluntary overtime” is time examiners work but don’t claim. This happens fairly frequently, particularly with examiners who are just starting out or who are struggling to meet their production goals. Production is based on the number of “counts” generated per examining hour. Some number of counts are awarded for issuing office actions, allowing a patent, a request for continued examination being filed, an application being abandoned, and so on. Each examiner has a certain number of “balanced disposals” (BDs) per bi-week she must get in order to reach 100% of production; the total number of BDs required per bi-week varies by the examiner’s pay grade and technology area.

One major implication of this system is that, for a set number of BDs, working more hours means less production. Suppose, hypothetically, an examiner had to get 4.5 BDs per bi-week; that’s the 100% production level. Each bi-week has about 70 hours of examining time, but let’s say that this hypothetical examiner only has 3.5 BDs after examining for 70 hours. To hit 100%, the examiner is going to have to work overtime; we’ll assume it will take another 20 hours of work to get that last BD.

If the examiner works paid overtime, that creates a problem. The examiner would claim 90 hours of work to produce 4.5 BDs; but in 90 hours, the examiner is expected to produce more than 4.5 BDs. She’s expected to produce 5.79 BDs in 90 hours. That means that instead of being at 100% of production, which is fully successful, she’d be at around 77% production, which is unacceptable.

On the other hand, if she simply works extra hours without claiming them, she produces 4.5 BDs in 70 hours of claimed time, even though she actually worked 90 hours total. That gets her a fully successful rating, even though it costs her 20 hours of voluntary (i.e., unpaid) overtime.

For those examiners who are just fully successful (95-102% of production goal), marginal (88-94% of production), or unacceptable (below 88% production), there is a strong disincentive to claim additional hours. Rather, there’s a strong incentive to work voluntary overtime in order to get production up.

How much is voluntary overtime used? According to a survey by the GAO, nearly 70% of examiners worked voluntary overtime just to keep up with their production requirements. That’s significant. It’s a shame the OIG ignored voluntary overtime completely, because it’s time that the USPTO gets for free.

There are also examiners who end up working more than 12 hours on a given day. They may have extra responsibilities that take more time, like primary examiners. Or they may need to make deadlines for docket management, or may have personal reasons for needing free time on another day.

These examiners have two options: donate that extra time to the office or claim it on another day when they have fewer hours. While it might be noble to donate the time, it isn’t unreasonable to want to claim the time somehow.

The OIG’s methodology, however, completely misses this. Suppose an examiner works 14 hours on Thursday and 10 hours on Friday and claims 12 hours for both days (because you can’t claim more than 12 hours for any one day). The examiner has actually worked 24 hours. But the OIG’s review did not credit more than 12 hours on any one day. The OIG would only credit 12 hours for Thursday and 10 hours for Friday, flagging 2 hours as “unsupported,” even though every single claimed hour was actually worked. Even though a rule may have been broken, that’s not fraud in the sense of claiming time that wasn’t worked.

How common is this? We have no idea, because the OIG didn’t account for it.

The OIG did identify a group of 415 examiners who accounted for 43% of the unsupported hours over the 15 month period. That group is only 5% of the examiners reviewed. There’s no excuse for real abuse, but this hardly an across-the-board problem.

The OIG could have said that 95% of examiners averaged about 38 minutes of unsupported time every bi-week, which isn’t bad when you consider that the average American worker doesn’t work 34 minutes out of each workday according to the National Bureau of Economic Research. But that doesn’t generate headlines.

Instead, the OIG used large numbers out of context to make it look like it had uncovered some massive amount of waste.

The OIG Missed the Problem of Poor Network Infrastructure

Moreover, there is a real problem with lost time that the OIG failed to identify. Specifically, the poor network infrastructure at the USPTO costs a significant amount of lost time, far more than the OIG identified. The OIG actually identified system-wide network outages in Appendix E of the report. Over 15 months, the total outage time listed by the OIG comes to 106 hours and 45 minutes.

The network outages affect all examiners, not just a few. If we multiply the 8,399 examiners in the OIG review times the total time lost, we get 869,593 hours lost for all examiners. That’s more than 3 times the alleged unsupported hours the OIG found. Using the OIG’s cost figures, that’s $55.2 million in lost work. (This doesn’t include individual computer or network problems.)

How is this loss of time not the headline? If 5% of examiners allegedly claiming unworked time amounting to 288,479 hours is worth a Washington Post story, surely network outages affecting all examiners and causing more than 3 times as much lost time is worth some attention?

Of course, network infrastructure is not as sexy as alleged fraud. Still, if we’re going to work on real problems, it seems like we’ll get more bang for the buck from fixing the USPTO’s network than from punishing all examiners for the actions of a few. Abuse doesn’t have to be tolerated, but it does little good to overreact and ignore real work that needs to be done.

The GAO Made Important Recommendations on Improving Patent Quality

The Government Accountability Office recently released two reports covering patent quality vs incentives and research capabilities that show why the USPTO issues poor quality patents in the first place. The big problem is that the USPTO has been focusing on quantity, not quality, for many years. Under Director Michelle Lee, there has been a new drive to improve patent quality, but it’s been hampered by procedures and systems put in place long before she arrived.

The GAO found there is no consistent definition of patent quality at the USPTO. Without such a definition, it’s nearly impossible to develop standard practices to produce better patents. As a result, examiners have no clear performance measures with respect to quality — just very specific production goals to meet. No wonder they focus on production goals.

It turns out part of the reason for the lack of a “patent quality” performance measure is the USPTO’s outdated information technology infrastructure. I’ve met with USPTO officials, and they expressed frustration that they simply could not gather the data they need. It’s just not in the system and there’s no easy way to capture it. To the USPTO’s credit, it is developing a new system, and enlisted the Department of Commerce for expertise as well. Still, that new system is a way off.

Furthermore, it’s not easy to do a complete job when you’re under a lot of time pressure. The GAO found that 70% of patent examiners say they don’t have enough time allocated to do their jobs thoroughly. The GAO found a number of factors contribute to this time pressure and make it more difficult to improve patent quality.

First, patent clarity is a big problem. The GAO estimates 90% of examiners always or often encounter broadly worded patent applications, and two-thirds of examiners say this makes it difficult to complete a thorough examination. But even though patent clarity is a real problem, patent examiners, particularly in software, are discouraged from forcing applicants to fix unclear claims. To see what I mean about clarity, here’s an example from the GAO quality report (p. 9):

[O]ne claim for a cardboard coffee cup insulator begins by referring to “a recyclable, insulating beverage container holder, comprising a corrugated tubular member comprising cellulosic material and at least a first opening therein for receiving and retaining a beverage container.”

That kind of language is not exactly easy for laypeople – or even patent lawyers — to read. Patent examiners have the ability to reject claims as unclear or poorly specified using what are called Section 112 rejections. But the GAO found that there is actually substantial pressure to avoid using Section 112 rejections. About 42% of examiners surveyed said that they felt at least a little pressure to avoid 112 rejections, and over a third of those examiners felt moderate to a lot of pressure.

Another issue for examiners is that there’s no limit to the size of a patent application. A patent claim is the text that describes what the inventor can block others from doing. It’s critical that each patent claim be clear, new, and not just an obvious variant of something else. But many patent applications have dozens of claims to examine, and a patent examiner gets the same amount of time to examine an application with 100 claims as an application with just 1 claim. This is true even though the applicant pays extra to have more than 20 claims in the patent application.

To make things worse, the USPTO cannot get rid of patent applications except by issuing them as patents. An applicant can choose to abandon an application, but it doesn’t have to. If it wants, an applicant can keep the patent application going forever by just paying more fees and filing requests for continued examination. Examiners get a fixed amount of time per application, no matter how many requests for continued examination they have to deal with.

Eventually, the sheer volume of applications on an examiner’s docket will create pressure to allow applications just to get some breathing room. Indeed, the GAO found that around 45% of examiners felt at least some degree of pressure to allow more patents.

There’s more in the reports, which are well worth reading.

The GAO reports shine light on some areas where the USPTO can make a big difference in patent quality, and the GAO’s recommendations should be adopted in full. The current administration is showing a strong desire to improve quality. It will take time and a lot of work, but it’s crucial to innovators and the U.S. economy that this effort succeeds. We won’t get there if we get distracted by the OIG’s search for a scandal.