Tuesday, January 6, 2026 | 11:00a.m. – 12:00p.m. (PT) | B-CLE
Presented with Baker Botts
1.25 General CLE Credit Offered
B-CLE Recording (CLE: $115) | Youtube Recording | Resources | Speaker Bios & Contact Info | Download Transcript
Program Summary
This webcast, hosted by Wayne Stacy of the Berkeley Center for Law & Technology, featured Matthew Avery, a partner at Baker Botts in San Francisco, presenting empirical research he co-authored examining the relationship between patent prosecution rigor and litigation outcomes. The study was motivated by criticism that the USPTO routinely issues low-quality “junk patents,” a characterization Avery found professionally troubling and wanted to investigate objectively. His team analyzed roughly 89,000 patents litigated in district court and at the PTAB over a 21-year period, using the number of office actions a patent received during prosecution as a proxy for examination thoroughness. The central and counterintuitive finding was that patents receiving more office actions were actually slightly more likely to be found invalid in litigation — the opposite of what the research team had initially hypothesized. The panelists interpreted this largely as a signal that a higher number of office actions correlates with a more crowded prior art landscape, rather than indicating that the USPTO is doing a poor or careless job.
The panel’s discussion explored multiple explanations for these surprising findings. Daisy Yau, senior patent counsel at Oracle, suggested that patents in crowded fields attract more office actions because examiners must grapple with abundant prior art, yet examiners may ultimately allow those patents despite the volume of art — which then gets surfaced again at trial. Keith Jurek, senior patent counsel at GRAIL, pointed to 112 (written description) and 101 (subject matter eligibility) issues as potential contributors, noting that an examiner pressed for time may miss claim-support problems that a motivated litigant will later exploit. The data also revealed meaningful variation across technology centers, with certain software-adjacent fields (TC 2100 and TC 3700) showing an inverse relationship — more office actions correlating with lower invalidity rates — suggesting that 101 pressures in those areas may dominate the dynamic.
The webcast concluded with practical takeaways for both prosecutors and litigators. Megan White, a Baker Botts partner in the Dallas office, observed that patents with a lower number of office actions tend to have broader claims and are better candidates for assertion, while defendants facing patents with extensive prosecution histories might actually have stronger invalidity arguments than previously assumed. Keith Jurek added that patents examined by “red” (toughest) examiners showed a notably consistent and relatively lower invalidity rate, suggesting that surviving a rigorous examiner may yield a more litigation-durable patent. Wayne Stacy closed by raising the question of how jury behavior — specifically the tendency of juries to link infringement and validity determinations — may be confounding the data, and encouraged Avery to pursue further research to disentangle those effects.
Moderator
Matthew Avery (Partner, Baker Botts)
Speakers
Megan White (Partner, Baker Botts)
Keith Jurek (Sr. Patent Counsel, GRAIL, Inc.)
Daisy Yau (Sr. Patent Counsel, Oracle)
About B-CLE
Program Transcript
Keywords: Patent litigation, invalidity rate, infringement, unenforceability, Office actions, examiner toughness, USPTO policies, patent prosecution, claim construction, prior art, 101 rejections, 112 issues, portfolio management, litigation outcomes, tech centers.
This program, The Impact of Prosecution Length on Patent Litigation Outcomes, was hosted by Berkeley Center for Law & Technology, UC Berkeley School of Law. The program was moderated Matthew Avery (Partner, Baker Botts) with panelists Megan White (Partner, Baker Botts), Keith Jurek (Sr. Patent Counsel, GRAIL, Inc.), Daisy Yau (Sr. Patent Counsel, Oracle). The program occurred on January 6, 2026.
Wayne Stacy 00:30
Welcome everyone to the Berkeley Center for Law and Technology’s Expert Series webcast. We have a really interesting program today from a former partner of mine at Baker Botts, Matt Avery. In addition to his day job, he is leading the patent prosecution team here in California for Baker Botts. He apparently has an enormous amount of time to dig through large data sets. So that is an impressive use of your spare time, Matt, but a really, really unique data set that can help people address what’s happening in patent prosecution, try to make some sense out of the United States Patent Trademark Office. So I’ve seen this before, and I’m glad to be sharing this with a with a larger audience. So Matt, I wanted to turn it over to you and let you introduce your co-panelists. Thank you for making this happen.
Matthew Avery 01:30
Great. Thanks, Wayne. Appreciate the opportunity to present to the BCLT audience here. So again, I’m Matt Avery, partner at Baker Botts here in San Francisco. Just want to briefly introduce my co-panelists here. You’ll get to hear a lot more from them coming up as we go through the data and hear some of their insights. So we’ve got Megan White, my current partner here at Baker Botts in our Dallas office. We’ve got Keith Jurek, senior patent counsel at GRAIL, and Daisy Yau, senior patent counsel at Oracle. So we have a lot to cover today. So we are going to jump right in and start talking about the research here, and get to the data quickly and start hearing what my colleagues here have to say about it. So just to frame the research for the audience here, we published this earlier in the year. If you want to go read the full paper, there’s a link below. You can search for it on Harvard Tech Law Journal’s website. So the research was motivated by this question that bothered me, this criticism of the patent office for allegedly issuing what someone characterized as junk patents. And my primary job is being a patent prosecutor and getting patents for my clients. And so I took this a little personally, and thought, am I getting junk patents for my clients? That doesn’t sound right. I don’t think that’s right. So I wanted to investigate this. And the things that we wanted to look at in this research was, first, this criticism of patent office, of just allegedly rubber stamping patent applications and giving out junk patents. So first, you know, what are junk patents? How do we even figure that out? And then second, how do we measure whether or not the patent office is doing its job and doing a thorough examination? And is there any correlation between these two things, between the thoroughness of the examination and the patents that get issued by the patent office. Listing some of our hypotheses going into this research below. I’ll touch on these a little more in a moment. First, just to frame some background issues for the audience here. So maybe some of you haven’t thought that much about how the incentives work at the patent office, those of you doing a lot of prosecution. So just a very brief primer here. But as you’re probably aware, the patent office, they’re working effectively on a quota system. The examiners are and they are basically given a certain amount of time to do the tasks that they have to do in terms of examining an application and issuing office actions. And they have bi weekly quotas and quarterly quotas that they have to hit and so what we decided to do in our research here is use the number of office actions a patent gets prior to allowance as a measure of the thoroughness of examination, and we believe that this is a valid measure, because we know that examiners are allocated a certain number of hours to review an application and issue an office action and then to do subsequent office actions. So we can basically use the number of office actions as a proxy for the number of hours spent on examining an application, and thus the thoroughness of that examination. One key point is that the number of hours that they get as examination goes on for each subsequent action decreases over time. And so … the thoroughness gets less and less over time. We’ll talk a little bit more about this later in the presentation here. I wanted to talk about some of the main hypotheses going in that I think are going to guide a lot of our thoughts going forward. So first — that this got to be the main assumption — and this probably resonates with all of you patent prosecutors out there, that more office actions is going to lead to longer, narrower claims. So there’s been a lot of research on point here. The stat that we’re citing in the slide here is from juristat. But I’ve also seen similar numbers from other researchers. The juristat data shows that 28 words are added to the independent claim for each office action, right? So you know, if you took two office actions to get to allowance, that would be 56 words added, etc. So in view of this assumption that more office actions is going to lead to longer, narrower claims, we made a few hypotheses about litigation outcomes that we were looking at, and so we looked, we had data about three types of litigation outcomes for patents litigated over the timeframe we were looking at, which is the past 20 years or so. We had information about validity, infringement and unenforceability. So the first hypotheses about invalidity, or about validity, is that more office actions is going to need lead to narrower claims, which we think should lower the likelihood of invalidity, make it harder to invalidate a narrower claim. And this was kind of our gut thinking here that, well, a narrower claim, it should be harder to find prior art that reads on that claim. It should, because it’s longer, more likely to recite technical subject matter that’s going to get it over 101 issues. So that’s why we went in with that hypothesis. As for infringements, similarly, more office actions leads to narrower claims that should also lead to a lower likelihood of infringements, right? This probably really resonates with everyone that narrower claims are harder to get your competitors product or process to actually read on that claim. Then the final outcome that we looked at was unenforceability, which effectively translates into inequitable conduct. And we weren’t really sure how to think about this, maybe more office actions gives applicants more time to mess up during during prosecution, more time to make misstatements to the patent office, or maybe it gives them more time to fix their mistakes. Not, not really sure here, but with our hypothesis, we went in and thought, Well, maybe it’s more time to make affirmative misstatements to the patent office, more time to miss prior art that you should have been citing, and that’s maybe a higher likelihood of unenforceability. So those were our hypotheses going in. A quick note about the data. So we looked at 89,000 patents litigated in district court and PTAB over a 21 year timeframe. So March 2000 – April 2021. The reason this timeframe was chosen, why do we go back to March 2000, that’s as far back as our data set went. So we basically used all data we had available to us. We didn’t make any arbitrary date cut at March 2000 for any particular reason. That’s just where our data set ended. And below, I listed all the attributes of the patents in the litigation, their associated litigations we were able to extract out. The critical things that we’re looking at here are the number of office actions and the litigation, the outcomes of the case, validity, infringement, enforceability. And then there are a few other things that we looked at. We looked at how the art unit or tech center impacts these litigation outcomes. We also looked at how examiner toughness, which is noted here as this examiner time allocation — I’ll explain that in a second, how that impacts the litigation outcomes — and then we also looked at venue in one instance.
Matthew Avery 10:01
So examiner toughness. We use this metric examiner time allocation, which is something that patent advisor, one of the databases that we extracted that our raw data from, uses as a measure of how tough examiners are. They don’t explain exactly how they measure this, but in short, they use this cute three color coding metric, where green is an easier examiner, yellow is a medium difficulty examiner, and red is a difficult examiner. So that’s going to roughly track which allowance rate where red examiner is going to have the lowest allowance rate and green examiners are going to have the highest allowance rates. And then final note before we get into the data is that we then went through our 89,000 cases and filtered those out for just the cases where we actually had litigation outcomes. So those of you that litigate out there are probably well aware that most cases settle, right? So out of this, 89,000, there was only 10,000 litigation outcomes to analyze, right? So basically, almost 90% of cases are settling, but that still gives us a lot of data to look at, with 10,000 litigation outcomes split up between the PTAB and district court here. So now let’s get into what our results were, which are pretty exciting. So first I’m just going to frame what the data here is showing and then I’m going to have our panelists kind of give us their thoughts on what the meaning behind this is. So what we’re seeing in this slide is a plot of the number of office actions on the x axis, and on the y axis, we’ve got our percent of litigation outcome. So the green line is validity and the red line is invalidity. So these two lines are mirror images of each other and they sum up to 100%, right? Because we’re looking at an outcome here that only has two possible options, either it’s valid or invalid. And in what you can see in this trend line, let’s just focus on the red line, invalidity here, is that as we go from zero to one to two to three, up to five office actions, there is a slight upward trend in the likelihood of invalidity, right? So in other words, more office actions is going to lead to a slightly higher likelihood of invalidity. You may remember, just from a few minutes ago, I said our hypothesis was the opposite of this, that we expected that more office actions would lead to narrower claims that would be less likely to be found invalid. So this is contradicting our hypothesis, contradicting my gut thinking about what should be going on here, and certainly raised a lot of questions about trying to interpret what this data means. One thing I do want to note is that the trend line is a pretty slight upward increase. And so some some people look at this and be like, Well, hey, that’s pretty slight. It’s only going up from like 0.35 to 0.4, so 5% increase in the likelihood of invalidity. Does that even mean anything? You know, our feeling is that this is significant, even though it is slight because it’s going in the wrong direction. So anyway, that’s framing the data. I want to ask our panelists give us their thoughts on this. And Daisy, I wanted to start with you. So maybe you could just talk about what you think the correlations that we’re seeing here signal about, you know, prior art when challenging patents, and what that signals? How that correlates with a higher number of office actions?
Daisy Yau 14:06
Sure, thank you, Matt. First, I want to thank you for inviting me to the panel today. Your research definitely revealed some very interesting findings, and it would be relevant to my handling of prosecution portfolio management infringement. So thank you again. So regarding the chart here, yeah, it’s very interesting how the finding contradicts our initial expectations based on our experience of how prosecution is like. It appears that the higher number of OAS does not mean that a patent — originally we thought maybe the number of OAS is an indicator of the rigor of examination, and so if you examine it more, it would have a lower invalidity rate. But that doesn’t seem to be the case. Rather, the higher number of OAS seems to correlate with a more crowded field of art. It seems that what is happening is that in a more crowded field, the more OAS you have, the examiner was able to address — needed more OAS to address the amount of art that was there. But ultimately, at some point, the examiner still gave in and allowed the patent application, and then ultimately, at trial, that prior art was still exposed, and ultimately the patent was invalidated. Conversely, it also means that the lower number of OAS doesn’t mean that the USPTO is doing a bad job. So going back to Matt’s point about the junk patents, that doesn’t seem to be the case. There are two possible few areas of research that we could still do that would further confirm these interpretations of the data here. One is to do a similar study, but using the number or diversity of cited references through prosecution. If our assumption is right, then the number of that, the number of OAS, correlates with the crowdedness of a space, then there should be an even higher correlation with the number or diversity of cited references through prosecution, with the invalidity rates. And then the other thing that Matt earlier mentioned was the declining point system, the patent count system, where you have declining number of points or hours for the OAS that are issued. Perhaps this is contributing to the invalidity rate, if indeed what is happening here is that the higher number of OAS indicates more prior art. So the pad point system was revised in 2010 to front load the credit giving examiners more points for a first office action on merits. So if we can see whether there is a difference between the pre 2010 data and the post 2010 data, and we can see whether the change in the point system is incentivizing examiners to allow patents despite a crowded field in the art.
Matthew Avery 17:26
That is new data to me about the point system being changed in 2010 That’s great to know. I didn’t realize that, yeah, the the idea of trying to split this up between a pre 2010 and post 2010 data set would be pretty interesting to see if that changes any of these correlations here. So aside from prior art, you know, Keith, I wanted to ask you, what other things do you think might be causing the correlation and invalidity that we’re seeing here? Maybe 112 or 101 issues, or other things like the count system that Daisy was talking about.
Keith Jurek 18:06
Thanks, Matt, and I want to first echo Daisy’s comments. Thank you for inviting me to the panel and about the quality of the paper here — I’ll put in a plug. It’s a pretty interesting read, and it’s not too long, which is nice for a journal article. So it’s a good topic. As far as the 101 and 112 issues, I think that there’s a couple things that could be happening here. So to Daisy’s point, about the count system and about the decreasing amount of time and amount of benefit to the examiner for continuing the examination and providing a really rigorous examination as it goes on, one of the major areas where 112 can be invoked in order to invalidate a patent has to do with inserting new matter into the claims, meaning material that isn’t explicitly recited in the specification, and relying on that for support for your claims. So one could imagine that as an examiner has less and less time to sort of check the work of the applicants as they’re inserting new material, they may miss more, and it may just be stuff that is missed by an examiner who’s overworked doesn’t have the time, but for a motivated litigant, they can find it, and they can find the errors and point them out. Similarly, it may be that applicants may be taking advantage of that situation and trying to push the limit a little bit too far, and getting a little bit too greedy in terms of the material that they think that they can appropriately add, as far as 112. I think there are some other issues, but those are the main ones that came to mind. For 101, I think that so, because the prior art has to do with extrinsic material, and the 112 and 101 issues have to do with intrinsic material. You know, you look at the spec, you look at the claims, it sort of, to me, indicates that there’s some kind of fundamental flaw. So if I’m imagining that for 101, if I’m being challenged on the subject matter eligibility of my claims at the third, fourth, fifth office action, that, to me, signals that there could potentially be some fundamental flaw with the invention itself, or at least the framing of the invention. And I know that the availability of the granularity of the specific types of rejections was not available in the data set, so drawing a little bit of speculation here, but I think it’s reasonable to assume right, that in a case that is still getting rejected three, four or five office actions on 101 matters, there’s some kind of fundamental flaw that needs to be overcome, which means it was a borderline case to begin with, which means, again, that a motivated litigant with the extra time would have more capability to provide the arguments and to challenge the patents effectively.
Matthew Avery 20:40
Yeah, I wonder about that. Does more office actions really just mean that, like, if I still don’t have an allowance by the fourth or fifth action, me as a prosecutor, maybe means I’m the one screwing up here, right?
Keith Jurek 20:54
Yeah, I think that it again, it comes down to the extrinsic versus the intrinsic components of it, right? If you’re still facing prior art rejections to Daisy’s point, that’s a con — that’s a very thick field of patent of prior art that you’re dealing with. But if you’re being rejected on intrinsic matters, you can blame the examiner as much as you want, but I think you also have to look internally and see what errors are we introducing, what rakes are we stepping on?
Matthew Avery 21:19
Yeah, that’s great point. And also to your one thing you noted about unfortunately, you don’t have the breakdown of the underlying types of rejections, at least in this study, that is an area where we’re going to try and look at for future research. How do particular types of rejections, 101, you know, 101 or 2,3, or 12, impact these trends. So maybe a future presentation for all the all.
Keith Jurek 21:45
I think one sign of a solid line of research is when you can inject so many different ideas for new topics to look into. You know, Wayne mentioned, this is a very exciting data set, and you can kind of see that in all of us. And I’m sure the interactions that you had talking about this data that everyone says, Oh, what about this? What about that? So it’s not to fault the data. This is an initial cut, and I’m sure that there are plenty more areas of research you’re looking at.
Matthew Avery 22:06
Yep, only limited by my time to actually analyze the data. So Megan, I wanted to ask you, you know, looking at the trend here and putting on your litigator hat for a minute, what do you think the correlations here signal about the weakness, or the inherent weakness and invalidity arguments for patents with longer file wrappers. And I’m particularly thinking about issues like the longer file wrapper just more miscarr–you know–or more characterizations that the applicant might have made that you can use against them when trying to invalidate the patent, things like that.
Megan White 22:44
And I want to echo my panelists and say, Thank you, Matt. I remember when this paper came out and you shared it, and this invalidity result was the most surprising to me. I think as lawyers, we always want to go with our experience and our gut, and it’s so interesting to have data really test that and to Keith’s point, you know, I feel like good data just brings up more questions, and, unfortunately, more research. So by sharing this, Matt, you’re just putting more work on your plate, and we’re coming up with all sorts of new research for you. So thank you for that. But to answer your question, one thing I thought about is, as you’re going through prosecution, right? Oftentimes you’re narrowing what you’re emphasizing is novel. So if the examiner has found pretty good art on a number of elements, and you’re either adding an element later in prosecution or emphasizing an argument that can be used against you in litigation, where you’re like, hey, this is the really novel, this element c is the really novel thing. And so that allows a defendant or someone challenging the validity to focus on that and really emphasize like hey, they were saying this element is novel. I have this slam dunk prior art to show it’s not at all. So that might have these repercussions of the more you narrow the claim, the more you limit what you’re emphasizing is novel, and that limits your ability to make those arguments in litigation, and benefits the person trying to invalidate your patents. So again, lots of questions raised in this surprising data set, but I think Daisy and Keith and and I went through a couple hypotheticals and possibilities of why this is happening.
Matthew Avery 24:18
Yeah, I really like that idea of, the longer my file wrapper getsm maybe I’m just painting a bull’s eye for anyone that wants to try and take this patent down. It’s like, hey, you know, I made all these incremental amendments, and finally, made one last incremental amendment at the end here that pushed it over the allowance line for the examiner, but now you as a litigator, it’s like, great. Well, that’s the only issue about what’s novel left. So I only have to search for that one little piece because you basically already conceded to the examiner that everything else is found in the art, right? So, yeah, of course, it’s a lot easier. So you know that’s not something that I usually think about as a prosecutor, but it totally makes sense now, you know, after talking to some litigators about it, why that might be happening. So the next thing that we did is we took the data from that prior slide, and we broke it up by tribunal. And so we did it just broadly, looking at either district court or the PTAB. All right, we could have gone, could have went further and looked at like, oh, how does you know District Court in Texas compared to other places, things like that. But for this first cut in our research, we just went at the highest level, so breaking up by PTAB versus the District Court. So what I want to emphasize here in the data is we’re seeing two things. So the first line, look at the solid black line, and this is invalidity at the District Court. So we’re just now taking the red line here from from the prior slide and breaking that up by tribunal. So only look at invalidity, the green line, the validity line is gone now, okay? So if you just look at invalidity, we still see the same upward trend. It’s actually even more pronounced here, going from 30 ish percent up to 40% so now we’re seeing like a 10% jump in the invalidity rate as we’re going from zero to five office actions. So the trend becomes even more pronounced to District Court. But then interestingly, for some reason, at the PTAB, the trend kind of falls away. And so we get this approximately flat line, the blue dashed line at the top, where the invalidity rate of the PTAB is just kind of approximately hanging out around 40% consistently, and not varying as much based on the number of office actions. So you know this, this is also surprising. And I first want to start with Megan. Here is someone who’s handled litigations in both tribunals, I believe. Can you first just explain, you know, for those of us that aren’t as familiar with the differences between these tribunals, like you know how, how are things handled differently between at the PTAB versus District Court that might explain why we’re seeing this different in the difference in the correlations?
Megan White 27:27
Yes, absolutely. So, interestingly, the PTAB uses broader claim construction standard. So if it’s broader and you can have a broader meeting for certain claim terms than in district court that are often more narrowly construed, then you have more opportunities to use prior art, right? The challenger is able to more broadly apply different pieces of prior art. So I think that is a huge difference between the PTAB and the District Court. The other thing is the audience, right? So PTAB judges are special. They deal with patent topics all the time. They deal with technology all the time. And in district court, it’s a jury. And so I think that that may have a big insight, too, into how things are adjudicated and who the audience is, and also the standards that may provide a little bit of an insight into the different aspects of of the lines here.
Matthew Avery 28:28
So, yeah, definitely. You know, the breadth of claim construction seems like be a big factor in causing these differences. One thing that I struggle trying to really understand is this PTAB trend line being flat and what do you — what do you think this flat trend line says about the importance of the file wrapper, so the number of office actions to the PTAB panel. You know, my understanding is that the PTAB looks at the file wrapper pre institution, but once it’s instituted, that the file wrapper is not as important anymore. So maybe you could talk about that a bit.
Megan White 29:11
Yeah, absolutely. So, I mean, I thought about this a couple different times, and I feel like the line indicates that it’s fairly neutral, right? These judges are perhaps not allowing the length of the file rep or the amount of prior art to influence their decisions. And also, there’s a high standard for even filing a petition in the PTAB. You know, if you’re trying to invalidate a patent, you want to make sure you have really good prior art. And so perhaps that provides the flat line of, hey, only the really best petitions are going to be issued there. And it doesn’t matter if it’s one office action or five. And so it’s really just, is there the really good prior art? And perhaps there’s no correlation. So I think it means the PTAB doesn’t care about. About the length of the file wrapper and all of the ways of attack that you have in district court from a longer file wrapper and paints it more neutrally, such that the difference and the length doesn’t have as much as an impact on the result.
Matthew Avery 30:14
Yeah, yeah, that’s interesting. So you know, looking at now the district court trend line. You know, some have noted that, you know, it’s the lot the data doesn’t like align that great necessarily, with the trend line, and that you could view this maybe differently as like a step function. So if you look at office action 0, 1, 2 and 3, it’s kind of hovering around 30 ish percent. And then at the fourth office action, it jumps up to a 40 ish percent invalidity rate and stays there at the fifth. So maybe rather than drawing this straight trend line, we should be drawing this as a step function, where it’s like, flat from zero to three and then jumping up 10% once you hit four and five. So if we assume that, like mapping it is or charting it is a step function is actually correct, like, what? What does that say about what’s happening at the patent office that might be causing this step change. And Keith, I know you thought about this a little bit.
Keith Jurek 31:25
Yeah, this goes back to the count system, I think. So the examiners are only allocated a certain amount of time, and as you have to file additional RCE, that amount of time decreases, so by the time that you reach your second or your first RCE, and then your second RCE, the amount of time really goes down fast. So the system doesn’t really incentivize examiners to spend a lot of time looking at cases. And so I’m — I was giving that preamble to allow the practitioners to start doing the math on terms of what that fourth office action would signify. That is, on average, that is an allowance that is after your second RCE, meaning the examiner has very, very little time in order to review the arguments on the merits, prepare or prepare a rejection, get it checked, do all of that work that’s necessary in order to issue the rejection, as opposed to filing or submitting a Notice of Allowance where they can have familiarity with the material, so they can kind of lean on that a bit. But they resist the way that the count system is structured, it really incentivizes a quick resolution at that point. And I think, you know, I’ll speak anecdotally here, that does seem to be a turning point for a lot of difficult cases where the examiner, maybe that’s they feel more emboldened, maybe they feel that they have more negotiating authority when they know the case better, but they may make very strong suggestions at that point that will help you get across the line, or it may just be simply the way the system is set up, they’re just unable to spend the time that they need to. But I do think it comes down to the count system, as far as what explains a jump like that. That would be my guess.
Matthew Avery 33:06
And I think this maybe gets to a point that Daisy was making earlier — by the time you’re getting up to this fourth action, and they give you an allowance at this point, you’re just maybe wearing them down. And, you know, they’re just like getting tired of this case, maybe I’ll just, you know, getting allowance now, it’s been, been long enough, right? And so maybe these are less meritorious allowances, right? You’re just wearing the examiner down, which is maybe not how we want to have the system actually work, right? So the next thing that we did with the data is we broke it up by tech center, and so we looked at all the different tech centers, except for design patents, right? So this is all the tech centers where utility patents are going. And you know, one thing I want to point out here is that the sample sizes, because now we’re splitting up eight different ways, the sample sizes start getting smaller. Some of these tech centers have, you know, particularly small sample sizes, the data starts getting a lot messier. So we’re not showing the raw data points here, because the chart would be impossible to follow. We’re just showing the trend lines. But these trend lines definitely have, you know, they’re weaker trends for sure. But you know something that’s interesting is that for almost all the tech centers, the trend lines still follow the general trend, which is that more office actions leads to a slight increase in the likelihood of it invalidity. However, for two of these tech centers, the trend lines went in the opposite direction, so TC 2100 and TC 3700. So TC 2100 deals with primarily software patents, and then 3700 is like mechanical and medical devices. So Daisy, I wanted to ask you, as someone who does a lot of software related patents, to at least comment on what you think might be going on here with TC 2100 and why do you think it’s it’s breaking trend?
Daisy Yau 35:15
Yeah, that’s a great question. Um, so with 2100 I expect the patent applications to face more one on one issues. That means that more office actions does not necessarily indicate the density of art, which we talked about earlier, but rather it could be the occurrence of one on one rejections. So this, in turn, means that the responses to Office actions do not necessarily include narrowing amendments that are intended to overcome art they might have arguments or amendments intended to overcome one on one instead. So because of that, the more OAS is actually a process that helps improve the claims, rather than a reflection of the state of the art. I think it’ll be interesting to observe the trend going forward, because currently, as we speak, there appears to be a shift in the US PTO to disfavor, one on one, rejections. More current data reveals that there is a drastic reduction in PTAB, 101 rejections. And so we can see in the future, going forward, whether the data really does show that 101 is the underlying cause of this trend based on future data?
Matthew Avery 36:45
Yeah, I’d be really interested to see how, you know, examination and moving away from giving one on one rejections at the patent office, you know, if we actually see that play out, how that’ll impact, you know, patent quality going forward. And, you know, it’s interesting your points about, you know, I like what you’re saying about one on one rejections here, maybe below the explanation. Certainly we know that from data that TC 2100 has one of the highest rates of one on one rejections. So, you know another thing I want to ask in the article. We don’t have the numbers up here, but in Table nine of our the article, which was linked at the start of the slide deck, we saw that average invalidity rates didn’t align with the average allowance rates by tech center, right? Which is something that we thought might happen, that, you know, tech centers with high allowance rates are maybe going to have high invalidity rates because they’re, they’re just, you know, letting out more, you know, allowing more bad patents, right? And that tech centers with lower allowance rates maybe that means they’re examining them more thoroughly, and thus those should be harder to to invalidate. And so when you just look at the average numbers those those numbers did not actually align. So what do we think that that means about the relationship between allowance rate and, you know, the thoroughness of examination and patent quality here,
Daisy Yau 38:23
great question, another strange, strange data set, right? Because there is a a hypothesis or expectation that where we expect, but it’s going against that expectation. So the invalidity rates of technology centers do not align with the allowance rates of technology centers. Now, if we look at where those high invalidity rates are, they are in 2100 2400 2600 these are computer architecture, software, networking, communications, again, these are the technology areas that might be more heavily influenced by one on one issues than 102103 issues. So these technology centers would them having special behavior is not out of expectation we could do again, more future research to see if, within a technology center itself, the expected correlation still exists, right? Because instead of looking at all the TCS at once, look at a single TC and my guess, again, hypothesis is that the correlation would still exist. But what take away from this, this table, I think that is particularly interesting, is that it does affirm, perhaps, that different technology centers should. Have different allowance rates. We shouldn’t expect the same allowance rates from this from different TCS, and this is because of the underlying differences in the technology and the laws. So we could consider whether TC should have a uniform invalidity rate, and thereby building more trust in patents in the court system and to the general public. But as for allowance rates, perhaps they should be different, and perhaps allowance rates could be monitored or calibrated to achieve uniform target and validity rates across technology centers. I can speak a little bit more about this. This idea later.
Matthew Avery 40:36
Yeah, in this is and maybe when Wayne gets back on at the end, you know this idea of consistency of quality control among different art units or tech centers, right? Should we expect? You know, the same allowance rate in TC 1600 which is dealing with like biotech and in chemistry, you know, as TC 2100 which is, which is software, right? And they definitely have different, you know, different allowance rates. You know, TC 1600s quite low. TC 2100 is relatively high. But then, you know, if you look on the chart here, like the invalidity rate, like TC 1600 very low invalidity rate across, you know, across the whole chart here. And so they seem to be allowing pretty high quality patents versus TC, 2100 relatively high in validity rates across the board. And so they are allowing relatively low quality patents, which is which is interesting. So one last thing about invalidity to talk about is we parsed out the data by examiner toughness. So again, remember, we’re using patent advisors, measure of examiner toughness. They break up examiners into three categories, green being the easiest, yellow being medium difficulty, and red being the most difficult or examiners with the lowest allowance rate. And what we see for all these trend lines is that they still all generally follow the same trend as we saw before, invalidity increases with the number of Office actions. But the trend is most dramatic with the green examiners, where we see it jumping up from like 36% up to 51% so that’s a pretty big change over the number of OAS, whereas for the tough examiners the red ones, the line is relatively flat, slight increase, but almost Flat, going from, you know, at 00 office actions, 29% to, you know, five office actions is actually drops back down to 29% again. There’s some noisy data at the fourth office action there, where it jumps up to 39 but basically this red line is almost flat at the 30 ish percent, regardless of the number of Office actions. I guess the first question there, and I’m going to pose this one to Keith, you know. So we see that for the easy examiners, patent quality is getting a lot worse with the increasing number of rejections, and it’s pretty constant for hard examiner. So why do we think that we’re seeing such a big difference between easy and hard examiners. Is it something about the quality, you know, the amount of thoroughness that you know easy examiners versus hard examiners? I should say red and green, right, rather than putting some type of value judgment there. So you know, red versus green examiners are providing with their examination over time. For my money,
Keith Jurek 43:49
I think this is one of the more interesting results to see compared to the overall trend line. So I’ve thought about this a lot. I have a number of theories to offer, starting with one that’s slightly uncharitable. It could be that diminishing examination quality, right? If we’re using this as a proxy, if we’re using the invalidity rate as a, as a, as a proxy for examination quality, it could be diminishing examination quality is a hallmark trait of a green examiner in this classification system, right? It could be that you’ll notice that the invalidity rate is clustered relatively close. It’s within a 5% band for the for the lower number of Office actions. But then, as it goes, that gap expands. And it could just be that right examiners, they’re more consistent, and they apply their standards more consistently. And while green examiners, they, you know, let things slide a little bit. It could also be, again, taking an uncharitable read of it could be that there are, is that applicants are reacting to the green examiners. By it, because the reactions are less, they’re contributing less to the overall validity of. The patent because they’re worse the they’re just worse. Written rejections could be that the examiners are not providing a really thorough explanation of the of the art, thorough explanation of their positions, and that applicants are taking positions that they maybe would have not taken with a red examiner, with a green examiner. I also think that there are some possible charitable reads to the situation. I don’t want to put blame on anybody, but you’ll really notice that it’s green examiners really going against type here, right? It’s sort of a strange thing to think about. You think the easiest examiners, these cases, are going significantly longer than the average for them. They’re they’re really holding on to these cases for a long time. Why would they be doing that? Maybe these are just really, really bad patents, like, maybe this is your set that breaks that, that that provides evidence of that myth of the bad patent, right? If the if an examiner who is fully inclined to give you something at two office actions is holding you until five, something’s got to be going on there, right? So it could be evidence of that. It could be evidence of the bullying effect, right? The applicants just holding on to just hammering the examiner over and over and over again. So I think there’s a couple. It could be all those mixed in together. Obviously, every individual case is going to be different, but I think that there’s something to be said there for the Hallmark traits of the type of examiner and the the quality of the fundamental cases for the for the long, the long week held green examiner cases.
Matthew Avery 46:29
Yeah, I like this idea of, if my case gets assigned to a green examiner and I still haven’t gotten an allowance by the fourth or fifth action, maybe it’s just a bad, you know, bad application in the first place, right? So, you know, Daisy, I want to ask you next about these trends here, and what you think that these trends are telling us about consistency of examination between, you know, easier and tougher examiners and and there was something you pointed to table 11 in the article, where we looked at how allowance the average allowance rates among these cohorts of examiners compared to their invalidity rates, right? And we see that they do, in fact, align that green examiners, which have the highest allowance rate, also have highest average invalidity rate, and red examiners with the lowest allowance rate have also the lowest invalidity rates. So someone, What’s that telling us about the consistency of examination?
Daisy Yau 47:34
Okay, great question. So in terms of the consistency, it does seem to indicate that the red examiners are more consistently thorough with their examinations regardless of the number of Office actions. In other words, additional office actions are not needed for a tough examiner to provide a thorough examination. The correlations basically match our expectations. Overall examiners that are tougher have lower invalidity rates. Contrasting this with the earlier table that we looked at regarding the technology center allowance rates, this table 11 seems to affirm that allowance rates are a good proxy for invalidity rate. And so something is weird about the technology center kind of data. So again, maybe it does suggest we can aim for some uniform, targeted invalidity rate, and maybe the technology centers can can consider that as a uniform target, but yeah, that’s for consideration.
Matthew Avery 48:47
Yeah, yeah. I like this idea that, you know, maybe for the toughest examiners, this explains this flat line that like they are doing a thorough examination up front, and perhaps they’re only giving you allowance when it’s actually meritorious, right? Which is why we see a pretty consistent invalidity rate across the board. So let’s you know, move on to talking about infringement. Next. Talked a lot about invalidity. We’re going to move to the other the other outcomes more quickly. I think the the results there a little more straightforward or less interesting. So first, let’s talk about infringement. So recall that for infringement, we hypothesize that more office actions is going to lead to narrower claims, which should lead to a lower likelihood of infringement, right? So here’s the data we got. Again, we’re back to showing infringement and non infringement both graphed together, so no infringement is it’s the green line. Infringement is the red dashed line on top, and so they are mirror images of each other, summing up. 100% at each at each data point. So what we see, and I think we what we want to do, is focus on the infringement line. The red line going down is that, as we get more office actions, the likelihood of infringement this red line goes down and down right? So basically, this is aligning with our hypothesis that more office actions is leading to narrower claims and a, you know, lower likelihood of of infringement, dropping from, you know, about like mid 60 ish percent down to 45% once we get up to five or more office actions. So, so starting with with Megan here, right? You know, we know that each round of amendments is adding, you know, approximately 28 words to each independent claim. So do we think that this incremental narrowing and it’s kind of explaining everything about what we’re seeing in this trend line. Or do you think there’s something else going on here?
Megan White 51:09
So I think there are nuances. I will say, when we were looking at invalidity, I was saying it’s nice to have your gut feeling checked by data, but it’s also nice to have data confirm your anecdotal experience, and I think what you said, Matt is spot on. If you assume that for every office action you are adding amendments, sometimes you just argue, right. But assuming that each office action you are adding, like you said, on average, 28 words, that adds to a longer claim, and that does make it more difficult for infringement if you’re adding true substance to overcome an office action, it’s a meaty, meaty amendment. And so then putting on your litigator hat when you go down the line to prove infringement as a patent owner, it’s harder because you have more boxes you need to check. So I know that there are nuances, and the trend line isn’t perfect, but I do, I do think that our anecdotal experience and our hypotheses are confirmed by this downward trend line of more office actions equals more words in the claim equals more difficult to show infringement.
Matthew Avery 52:14
Generally, having a trend line that matches my hypothesis makes me feel better about that I actually analyze the data correctly. And certainly the results we were looking at before threw me into some turmoil about, like, maybe I’m screwing up my analysis completely. Did I flip these numbers? What’s going on so and of course, we did, like, triple, quadruple, quintuple, check all of our data to make sure it was all right. So, you know, Daisy, I wanted to ask you next right is a, you know, portfolio manager like, you know, as cases are getting more and more office actions, and, you know, we are seeing the likelihood of infringement dropping by, you know, from like 60 ish percent down to 45 ish percent, like almost a 20% drop in the likelihood of infringement. You know, how does that factor into your thinking about, like, the value of, you know, patents in your portfolio, and perhaps, like for pending applications, like the value of continuing prosecution?
Daisy Yau 53:19
Yeah, that’s a great question. So in terms of portfolio management, portfolio pruning is a process of strategically removing poor patents from our portfolio to reduce costs and to just build a cleaner need easier to navigate portfolio to use for offensive or defensive purposes. Traditionally, portfolio pruning is based on factors such as the technological field, the claims go forward, citations, continuity, families, things like that. But I think based on this really interesting data, see the significant infringement rate drop here, adding the number of OAS as a factor to consider in portfolio, pruning would be valuable. Yeah, but keep in mind, I guess, that the purpose of the portfolio does matter. Pruning is more important for offensive portfolios, for portfolios that are kept for defensive purposes or for valuation purposes or other purposes, other factors might be important as well, right?
Matthew Avery 54:29
And of course, like each patent still has to be evaluated individually, right? These are just maybe signals that are helping us kind of evaluate it at a high level. So, you know, Keith, I want, I was hoping you could also maybe comment briefly on this weird data point that we’re seeing at one office action where the likelihood of infringement starts at like 62% and then goes up to 66% and then, and then drops down. Um. And you know what’s going on with 00, office actions? Why is, why is your office actions worse?
Keith Jurek 55:09
You know, in this instance, I’m not going to over, overthink it. I think it really does come down to that common assumption that if you get an allowance on your first action, you filed way too narrow, that your your initial claims, you know they, they captured your invention, but that they, they were not, we’re not as broad as they could have been. I think that that there’s a strong narrative there that carries forward, that sort of lines up with the assumption there. And maybe there’s something interacting here, having to do a disrecord of nailing down claim construction as well. If you don’t have a file history to guide you, it sort of becomes more of an open book, potentially, where, I don’t know there may be some games to ship that helps to explain that. You know, percentage point difference there. But my Yeah, my initial assumption is just it’s, you file too narrow, and you, you failed to capture the competitor products because you filed too narrow.
Matthew Avery 56:05
Yeah. So I think that gets back to, you know, I think what a lot of us learned when we got trained is like, yeah, you want to file, try and file broader claims. And yeah, if you got a first action allowance, 00, OAS, you might have screwed up there and filed something that is a little too narrow. So we’re going to go through the next few slides pretty quickly. Here. Again, we did a breakdown by Tech Center, and again, we saw that the general trend that we saw previously holds up for almost all the tech centers. So more office actions is leading to, you know, a lower likelihood of finding infringement again. For some reason, TC 20 130 700 are deviating. Might be the same reasons as for, you know, what we saw in the invalidity so, just brief comment there, and then we also broke these down by examiner toughness. So I did want to get a few thoughts here. So what we’re looking at now is just the lines for infringement, right? So basically, the red line from our original chart now broken up by examiner toughness, and we see for green and yellow examiners that that they generally follow the same trend as the general trend where more office actions is leading to a lower likelihood of infringement. But for our red examiners, again, they’re deviating, and we’ve got kind of a, you know, somewhat of a flat line, certainly a much reduced trend, where the number of Office actions is having, you know, a lot less, or arguably, no impact on on the likelihood of infringement. Um, so you know, Keith, maybe just starting with you. What do you think the the green trend line and maybe the yellow one also are, are telling us about how you know easier examiners are examining these cases and how that impacts infringement.
Keith Jurek 58:29
So my initial read on this is that the the the non right examiners in this cohort are they’re triggering, triggering amendments and responses that may be more tactical than strategic, where the applicant is responding to the office action, but not responding in the context of the broader invention, the broader value of a claim. So the thinking I can imagine is, you know, this examiner has such a high allowance rate, why am I struggling? I just need to overcome this prior art. I just need to move beyond this 112 rejection in the hopes of getting the allowance next and again, if you’re extending a case along that far, we spoke to the quality of the patent at that point, but so that could mean that an applicant is more willing to take larger amendments, more significant amendments, that would naturally cut the infringement value of the of the claim, whereas with, with, compared to red examiners, you know, you could be arguing more because you want to really make your point, because you’re really fighting for this, when you think that there’s a good
Matthew Avery 59:31
chance for it, yeah, yeah. So in Stacy, I want to ask you, you know, particularly, just focusing on the red examiners to, you know, keys point like, you know, maybe we’re just arguing more. Do you think there’s a difference in how we are handling our responses to Office actions from from Red examiners? Yeah.
Daisy Yau 59:51
So the red trend line is pretty flat, and it indicates that the red examiners have more consistent. I. Infringement results. I think that, to echo Keith’s point, it shows that the patents allowed by red examiners are not allowed necessarily because of narrowing amendments. It can be a different type of amendments or arguments are being made. Maybe there are more amendments to increase clarity. Another idea might be that the amendments were made to shift the scope or direction of the claims, so rather than just adding more limitations that just narrows the claims, you’re shifting the scope or direction to better address the examiners on point rejections. And one more idea might be that the rejections by red examiners are more varied or diverse. They’re not just because of 102103 they might have more 1011 12 rejections because the red examiners are more able to use a diversity of rejection types.
Matthew Avery 1:01:02
Yeah, yeah. So maybe because we’re getting more types of rejection that that is forcing us to, you know, respond, respond differently. And maybe, rather than just making kind of simple, incremental amendments to overcome, you know, prior art, maybe we have to do, you know, a lot more shift scope, do other things to advance a case with these tougher examiners. So that’s the end of our infringement analysis. Here. We’re going to talk about an enforceability it’s going to be relatively quick, because the data is, Well, it’s interesting, but it’s also not interesting. So here’s unenforceability. So the red line on the bottom is showing our trend in unenforceability, going from zero to five office action. You see, it’s basically a flat line at zero, right? And then no unenforceability. It’s basically a flat line at, you know, 100, 100% right? You know, you break it down, you zoom in, and we see, you know, it’s, it’s kind of a messy trend line about bouncing around, maybe around 1% you know, if you use just statistics, you know, our statistical model tells us that, like, oh yeah, these are the best trend lines in your whole, whole study, right? These are totally following. Look how close those data points are to the trend lines, right? So this is where a little common sense has to come in and tell you, like, well, also the sample sizes here, particularly for unenforceability, are so small that maybe this is all just noise, right? So to you know that the total you know, these sample sizes, you can see, are much smaller, because unenforceability is litigated much. It’s not in every case right versus infringement, of course, and in validity are going to be litigated almost every case, right, certainly infringement, so, but unenforceability is not necessarily going to be raised every case. So the sample size, just for the total data set, is a lot smaller, and then the number of actual findings of unenforceability is minuscule. So a lot of these buckets maybe only have, like, represent one or two cases, maybe three cases of unenforceability being found in the total data set, right? So these are all small. The total size of unenforceability here, I think, is like 21 total findings of unenforceability across the entire data set. So you could either say that, yeah, hey, there’s, you know, we’ve got this great trend line, or you could say, yeah, maybe it’s all noise, and these trends don’t mean, mean anything. But let’s, let’s, you know, give it a good Beth interpretation for the moment and say, like, Okay, we’ve got, you know, somewhat flat, flat trend line here where the number of Office actions isn’t really making a difference on, you know, the likelihood of unenforceability. So if that’s so, like, what do we think? And maybe Megan, you can comment on this. What do you think this means about the relationship between mistakes by prosecutors and inequitable conduct?
Megan White 1:04:12
I tend to agree with you, Matt, that I think this is a lot of noise and just too small of a sample set, but giving it its day, right, our hypothesis one of them was, Okay. More rejections lead to more opportunities for prosecutors to basically, like, engage in equitable conduct and do bad acts, right? So they find a really great piece of prayer and they don’t submit it. The longer it comes, you know, you the longer you have to find the prior art, perhaps longer prosecution you alluded to this at the beginning, allows them to do the right thing and provide the right material, or for the inventor to finally disclose it, or something like that. So more opportunities, more office actions, more opportunities for IDs is more opportunities for the examiner to find prior art. Perhaps that can. Keeps it fairly neutral in the ultimate findings of an equitable conduct. Yeah.
Matthew Avery 1:05:06
And Keith, what do you think similarly, this flat line might mean for, you know what the examiner is doing so, or, you know, actions at the patent office and how that might be impacting unenforceability?
Keith Jurek 1:05:23
Well, I think that the flat line indicates there’s no real correlation as to the actions and the amount of time or the amount number of actions. So it seems like the errors are being corrected in house before the patent is allowed. So I think that is this may be evidence of the system working as intended. As far as that goes. One interesting data point to look at, though, is, you know, there is a bump at one for as little as it is there is a bump at one office action. So maybe there’s something there that’s worth digging into. My suspicion, having looked at the tables, is that that may just be like one case that had just one really bad actor and that, like the patents were all issued around the same time, or something like that, because that actually, that that bucket right there represents more findings of unenforceability than all the rest of the buckets combined. So it’s, it’s a very difficult chart to interpret from that perspective, but I think giving it again, the good faith read is largely the issues come out in the wash and that we’re all professionals doing our jobs perfectly or ethically, I should say,
Matthew Avery 1:06:29
perfectly and ethically, as how I do my job. For sure, we again broke down unenforceability here by Tech Center. So now the data just gets so messy, because we’re taking 21 cases findings of unenforceability, parsing it eight ways. So some of these tech centers essentially have like zero findings of unenforceability. So anyway, I think that the key takeaway here is that there’s no real takeaway if you start parsing it down so so narrowly by by Tech Center. Similarly, we parsed it by examiner toughness. In here the data gets even smaller. So here we dropped from 21 findings of uninforceability down to, I think, only 14, because in our data set, some examiners don’t actually have a toughness assigned to them in in patent advisor. And so if toughness wasn’t assigned to the examiner, then that got excluded from this part of the the analysis. So now we’re looking at even fewer cases. So data here is even messier and narrower. Let’s wrap up here with maybe a few takeaways from from our panelists. And you know, first thing I want to ask is, just like, you know, from the data that we we’ve slowed like, what is it that surprised you most about the results above? And maybe Daisy we can start with, you
Daisy Yau 1:07:54
sure, I think the number one thing that was presented was the most surprising, which is the most the positive correlation between the number of OAS and the invalidity rates overall, combined that with the counter correlation for TC to the 20 130 700 that was surprising as well. Why are these two technology centers behaving differently? To me, I think that it seems to indicate that there are two opposing forces driving these trend lines. One is the 102 and 103 issues, they push for a positive correlation, and it appears that the 101 issues pushes for negative correlations. So if I’m handed a patent, then maybe the first question I should ask is, Does it face more significant 101 ish hurdles, or does it face more significant 102103 hurdles? And thereby use that to see how the number of OAS should signal the strength or weakness of the patent.
Matthew Avery 1:08:58
Yeah. And then Keith or Megan, any other surprising things from the results that you wanted to comment on?
Keith Jurek 1:09:08
I’ll be very brief, the trend lines for red examiners, I think, really fascinating, really interesting. Gives some hope for if you’re facing tough examiners. Generally speaking, the red examiners are ones who have very low allowance rates. If you’re facing tough examiners, it gives you some hope that maybe there’s a light at the end of the tunnel.
Megan White 1:09:27
And just to add to that, too, thinking about the examiner in assertion, right? So we typically look up the examiner like Keith was saying, we go through prosecution like, oh, this examiner’s red is so tough and they’re never going to allow it, but on the assertion side, when the company gets to the point of, Hey, are we going to assert it? Then looking back at who the examiner was and and using that as another data point of what patent is, is best to assert, right? So I think that would be an interesting data point for litigators to look at before they assert a patent.
Matthew Avery 1:09:59
Yeah. Yeah. And so I want to take just have you guys talk briefly about some some takeaways here, Keith, I wanted to return to your point about having some hope for red examiners and and maybe just comment on that a little more like you know, what’s your takeaway there, if you’re getting cases in front of red examiners, how that is maybe now, you know, in view of this data, making you rethink how you’re guiding your portfolio.
Keith Jurek 1:10:28
Yeah, it’s a really, it complicates, I think the work there a little bit. Because previously, if you were in front of a red examiner, my instinct was to do what we could, to get away from that, you know, fight for it, do what you can, but really look to really narrow down on the subject matter and either just drill down and be happy there, or file cons, file related subject matter and different in different with different examiners, or do what you can to try and land in front of different examiners. And I think there’s still some merit there to trying to steer away from the particularly tough examiners. But one way to read this data is to say you may not even want to do that, right? Because, if your goal at the end of the day isn’t to have a patent, but to have a valid and valuable patent, meaning one that reads on on another product, maybe you do want to be in front of the most difficult examiners, right? It goes to the point of, why do you want this patent? You know, what’s your what’s the value you’re going to extract out of there, and if it’s going to be something you want to assert, if you’re doing competitor litigation, maybe you do want to be in front of the toughest examiner’s because steel sharp and steel. So I think that there’s, there’s a lesson to be taken away from there in terms of really being a long think towards why you want these patents, and being more strategic as opposed
Matthew Avery 1:11:42
to being tactical. I like steel sharp and steel. That’s a great analogy. Daisy, I know you had some thoughts about, you know, what these findings implicate about, like USPTO policies and maybe things that perhaps should be changed at the patent office.
Daisy Yau 1:12:02
Jeff, so I think, as I mentioned earlier, maybe we could aim for more uniformity amongst technology centers for the invalidity rate. So in that case, you would set a target invalidity rate. Maybe it would be the average invalidity rate overall for USPTO, or it could. We could even piggyback on the red examiner’s invalidity rate being so consistent at around 30% and use that. I’m not sure which what is the right target as of now, but theoretically, if we have a target, then we could basically calibrate the each Technology Center’s allowance rate to align with that target in validity rate. Does that make sense? So if, if a technology centers in validity it was higher than target, then perhaps the allowance rate should be lower, right? So, um, if it was higher than the target, that means that too many in the in the valid uh patents are getting uh allowed, so the allowance rate should be lower. Does that make sense for the calibration? That’s just one idea. Yeah. And then yeah. One more idea is to look at the point system again, that that’s what I mentioned earlier. Like, is there unwanted incentive that that drive examiners to cave in at some point and allow bad patents? So we could, we could look into that policy that USPTO point system, and consider whether that needs adjustment, but definitely the current fee schedule with the increased fees for subsequent RCE was probably unpopular when it was pushed out, but perhaps there is a merit to that, because we’re seeing that the lengthier prosecutions are producing less strong patents. So perhaps the current fee schedule with increasing fees for subsequent RCS makes sense. From my perspective, we definitely want to keep low quality patents out of the system.
Matthew Avery 1:14:15
Yeah, see, we got Wayne back on and so we’re going to wrap up here.
Wayne Stacy 1:14:19
No, I actually wanted to weigh in. So if you go back to that last slide, give you another, another read on it. So this is the trial lawyer in me that looks at the second bullet point. There’s a is I listen to the presentation and see the data. There’s a presumed disassociation between invalidity and infringement, that those are two separate things. They’re decided separately, and of course they are in my patent law class academically, but there’s good data that show. Those juries run this like the ancient Roman Emperor up or down, and they dump everything at once. So the lower standard of burden is on infringement. Infringement is easier to conceptualize than validity. So as a fact finder, if I say infringed, well, then I’m more likely to go with validity. And you know, it statistically shows that juries link the two. So how do you how do you break that out and see if you’re getting some unfair linking from juries? Because if you look at that, you know the these worst performers. I can see that with a if you’ve got five office actions and 100 extra words, infringement gets pretty difficult to prove. If they vote against you on infringement, it’s 445 I’d like to go home. It’s Friday afternoon. Then I just take out the validity issues too. So how do you back that out?
Matthew Avery 1:16:07
Yes, that is a good critique of the data in terms of, like, confounding of the results possibly right, because validity. And you know, we’re analyzing these as separate, separate factors, and they may, in fact, be tied, you know, like we looked at invalidity standing alone at the PTAB, right? So, you know, infringement is not being judged there at all. So that’d be one way, you know, I guess that you could try and parse these apart. Of course, the problem there is that, you know, also, the way the PTAB works is very different from from District Court, you know, it’s not a jury, it’s professional judges. So, you know, it’s a good question, and it’s certainly something that I’d love to think through more for future research. How to, you know, do the statistics there to try and figure out how to, you know, break apart that data and and make it so that we’re not getting, you know, confused by, perhaps these issues being tied together by juries. I was
Megan White 1:17:14
just gonna say one other point to that is, you talk about linking and juries are unpredictable, right? Which is why things settle so much. But sometimes they link it in a different way, where it may impact the results Oppositely, right? So it’s like, I don’t like that party, I want them to lose, so in non infringement and invalidity, right? So they’re and zero, right? Like they just say, I don’t care. I don’t like that person, I want them to lose. And so I hear you what you’re saying of like, infringement is conceptual, and if it’s infringe, then it probably is valid. But I’ve seen the opposite to where it’s just like, nope. Like, I love your thumbs up, thumbs down. Roman Empire, it’s just like, nope, nope, nope. That party loses on all counts. So I agree. I think it’s a really interesting question, and and an issue with having a jury be a fact finder is, are they really understanding the key questions? And sometimes they are, and sometimes they’re not.
Keith Jurek 1:18:08
And there may be, there may be a way to get an answer there, looking at sorry went up very brief, by looking for instances where this would require a particularly dutiful research assistant, but looking for instances where the decision was taken out of the jury’s hands, right looking where the where it wasn’t true. It wasn’t tried before a jury, where it was a J mall or something like that, where you could sort of start disentangling it. I imagine that’s gonna be really difficult to get out of in any data set, but I think is a question that can be answered. It’s just, how much of Matt’s free time does he want to dedicate to answering it? It’s a very different point there
Wayne Stacy 1:18:39
will and if you look at this, this analysis, with just the cases tried to a jury that returned a verdict, it might actually give some illustration. And that’s what, six, 700 in this timeframe. So that might, that might illustrate it. And the reason I asked that is, I’m really curious, because, speaking of thumbs up, thumbs down. That’s how the whole PTAB seems to be going these days. So you know, if we’re looking at what to do going forward, here’s a system with the PTAB. Here’s a system without the PTAB. So I’m happy to volunteer more of Matt’s free time to answer all of our questions.
Megan White 1:19:21
I feel like every time Matt gives this presentation, he just comes away with more questions and more research to do. So eventually he’s just gonna hit a point where he’s like, I can’t do this anymore. Well, I love it.
Matthew Avery 1:19:32
Want to have just some final commentary here, just thinking from this, from you know, more of a litigator perspective, and, you know, maybe making just Megan, you could give us some of your thoughts on like, what are your takeaways as a litigator from the data above?
Megan White 1:19:52
Yeah, so, so two points on the patent owner side. So I mentioned this before, but looking at the examiners that that in when you’re a. Assessing which patents to assert, or whether to assert a patent, understanding that there is likely a difference between a patent that was granted by a red examiner and one by a green examiner. I think that’s another data point that people who are asserting patents can take a look at. The other one is what we’ve always suspected is lower number of Office actions, not zero, but lower number of Office actions is a better candidate for assertion because it’s more likely to have a shorter claim and more likely to be able to find a fringe mint. So I think those are both two really important data points and on the defendant side, right? So the we talked about invalidity a lot, but typically, when you see a patent that there’s a lot of prior art along File History, you get excited about claim construction and non infringement, but it’s a bummer on invalidity, and so this data really flips that on the head of, hey, perhaps we should lean more into validity based on this data. You know, the more office actions doesn’t necessarily mean you have fewer opportunities or fewer art out there, it might in heart and defendants at some point too. So I thought that, again, was super interesting for the data to flip our typical understanding on its head.
Matthew Avery 1:21:14
Yeah, so we’re quite a bit over time. Now I’m going to have Wayne wrap us up. If you have any final comments.
Wayne Stacy 1:21:23
Wayne, no, this is interesting, and I’m I’m curious to I’m curious to see your next round of research on this, so we can go ahead and schedule the follow up. Give you a couple weeks to bring this together. But I really like this. I think this is something that’s interesting for anybody that’s thinking about assertion or valuation of larger portfolios trying to work their way through it. When you can’t look at every claim you’ve got to look at some of the larger data. So very, very nice work, and it’s nice to see actual data entering the patent debate these days. So I appreciate you all putting it together, making this happen.
Matthew Avery 1:22:09
Yep, thanks. Thanks. Wayne, thanks Megan Daisy and Keith for going along for the ride today. I appreciate it. Thanks for having me. Thank you guys. Thank you. Take care. Bye.