Introduction
This Article analyzes over 89,000 patents litigated over a twenty-year period to determine how the number of office actions to allowance during prosecution impacts rates of invalidity during subsequent litigation. Many practitioners would hypothesize that a patent with a longer prosecution history will be less likely to have an inherent validity problem. Presumably, the longer a patent was in prosecution, the more likely an examiner will have found relevant prior art, and the more likely the applicant will have narrowed the claims to avoid the art found by the examiner. Thus, more office actions should lead to narrower allowed claims, lessening the chance the court finds it anticipated or obvious under § 102 or § 103 when litigated. Contrary to expectations, the analysis of litigation outcomes shows that an increase in the number of office actions correlates with a heightened risk of invalidation. These surprising results are discussed in detail below.
Data & Analysis
All data analyzed for this Article was extracted from PatentAdvisor,[1] which is powered by Lex Machina.[2] PatentAdvisor has litigation data for patents that have completed litigation from March 1, 2000, onward, provided they have electronic file histories. The litigation data includes results from U.S. district courts and from the Patent Trial and Appeal Board through April 2021.[3] This resulted in a raw data set with 89,248 rows. We then excluded cases where there was no finding of either infringement, validity, or enforcement (for example, a case that was dismissed or ended in settlement). This ultimately leaves us with 10,174 rows from 6,877 cases, which includes 6,382 outcomes from district court cases and 3,792 outcomes from PTAB cases.[4]
Results
Next, we analyzed our data to see if the number of office actions correlates with litigation outcomes. Because the number of cases with 5 or more office actions is significantly smaller than cases with 0 to 4 office actions, these cases were grouped into a “5+” bucket.
Figure 1: Plot of Invalidity and Validity Rates Overall
R-squared: 0.65, RMSE: 0.025, Total Sample Size: 7633
Figure 1 is a plot of validity outcomes for litigated patents in terms of percentage of total outcomes where there was a finding of validity.[5] The trendlines in Figure 1 show a positive correlation between the likelihood of invalidity and the number of office actions, apparent by the R-squared value of 0.65, and a close fit of the data to the trendlines, with an RMSE of 0.025. In other words, the more office actions a patent application receives before allowance, the more likely the issued patent will be found invalid during litigation. While the data shows only a small increase in the likelihood of invalidity — from 35% at 0 office actions to 41% at 5+ office actions — this trend is significant because it contradicts the hypothesis of this Article — that more rejections lead to narrower claims that are less likely to be found invalid.
While the reasons for this unexpected trend are not clear, there are a few possible explanations. First, it is possible that a higher number of office actions is indicative of a patent space that is more crowded with prior art.[6] Thus, even for patents that were significantly narrowed during prosecution to overcome multiple office actions, because the patent space is more crowded, litigants may be able to easily find alternative prior art to use when challenging the novelty and non-obviousness of the patent under §§ 102 and 103, respectively. Second, it is possible that as the number of office actions increases, the likelihood of the applicant adding new matter to the claims increases as the applicant attempts to overcome rejections. Every time the applicant amends the claims, they risk adding claim elements that lack support in the specification in violation of written description and enablement requirements of § 112, thus increasing the likelihood of invalidity. Unfortunately, the data extracted from Patent Advisor for this Article does not include information on the specific types of statutory rejections the patents received during prosecution.[7] As such, it cannot be determined if one, both, or neither explanation is correct.
Figure 2: Plot of Invalidity Rates by Tribunal
Next, the invalidity data was separated by tribunal. District courts and the PTAB have different procedures and standards for claim construction. Notably, the PTAB applies the “broadest reasonable construction” standard, which is the same standard used by examiners during prosecution.[8] By contrast, district courts will construe claims using their ordinary and customary meaning, in the context of the specification and file history, and read to preserve validity.[9] Thus, which tribunal a patent is challenged in should have an impact on validity outcomes.
The trendlines in Figure 2 show a striking difference between validity outcomes in district courts compared to the PTAB. Like the general trendlines in Figure 1, the trendline for district court outcomes shows a positive correlation between the likelihood of invalidity and the number of office actions, apparent by the R-squared value of 0.75, and a close fit of the data to the trendline, with an RMSE of 0.047. This aligns with the general trend for invalidity observed for all patents illustrated in Figure 1 — the more office actions a patent application receives before allowance, the more likely the issued patent will be found invalid during litigation. By contrast, the trendline for PTAB outcomes shows essentially no correlation between the likelihood of invalidity and the number of office actions, with a low R-squared value of 0.023, and a close fit of the data to the trendline, with an RMSE of 0.022. These results show that patents are more likely to be found invalid at the PTAB compared to district court, and that these patents are more consistently found invalid, regardless of the number of office actions and corresponding thoroughness of examination of those patents.
While the reasons for the difference between district court and PTAB outcomes are not clear, several potential explanations warrant consideration. Firstly, as noted above, the PTAB uses a broader claim construction standard, which should make it easier for challengers to invalidate claims based on novelty or obviousness grounds. Secondly, the PTAB uses a lower standard than district courts for finding invalidity. In district court, invalidity must be proven by “clear and convincing” evidence, which gives patents a relatively strong presumption of validity in litigation.[10] In contrast, invalidity at the PTAB only needs to be proven using the lesser “preponderance of evidence” standard, which significantly diminishes the presumption of validity during these administrative proceedings.[11] Thus, the higher invalidity rates at the PTAB for all patents, regardless of the number of office actions associated with them, should be expected. Finally, the procedural differences between PTAB and district court proceedings contribute to distinct adjudicative approaches. Once a patent is instituted for review at the PTAB, the panel does not give deference to the examination history and instead focuses on identifying examiner errors.[12] Consequently, the panel’s scrutiny may not be influenced by the thoroughness of the examination process, potentially amplifying the likelihood of invalidity findings. By contrast, district court proceedings may be influenced more by the examination history, wherein the frequency of rejections may signal proximity to prior art, facilitating a more robust invalidity defense.[13]
Figure 3: Plot of Invalidity Rates by Examiner Toughness
Finally, validity data was analyzed with respect to examiner “toughness,” which is shown in Figure 3 and Table 3. This Article hypothesizes that the thoroughness of examination is expected to correlate with the toughness of the examiner. For toughness, this Article borrows the definition used by PatentAdvisor, where examiners are categorized as “Red” (i.e., the toughest examiners with relatively low allowance rates and a high average number of office actions to allowance), “Yellow” (i.e., moderately tough examiners), and “Green” (i.e., the easiest examiners with relatively high allowance rates and a low average number of office actions to allowance).[14] The classification of an examiner as being Green, Yellow, or Red is based on a propriety calculation by PatentAdvisor, which they call an “ETA measurement.”[15] According to PatentAdvisor, the calculation is based on a number of factors, including the examiner’s allowance rate; years of service with the USPTO; and total number of issued patents, abandoned applications, and pending applications.[16] These factors are used to calculate a single value between 0 and 99, which is what PatentAdvisor calls an ETA measurement.[17] A Green examiner has an ETA measurement between 0.1 and 2.5 (inclusive). A Yellow examiner has an ETA measurement between 2.6 and 5.9 (inclusive). A Red examiner has an ETA measure of 6.0 and above.[18] This Article adopts an identical definition of “Green,” “Yellow,” and “Red” examiners.
The average allowance rates for each type of examiner compared to the invalidity rates for patents examined by that type of examiner are shown in Table 4 below. As expected, Table 4 shows that the average allowance rate for each type of examiner positively correlates with the respective invalidity rate for litigated patents examined by those examiners.
As seen in Figure 3, for patents that face the toughest examiners (“Red”), there is almost no correlation between the likelihood of invalidity and the number of office actions (R-squared = 0.065). The likelihood of invalidity stays approximately constant at 31% as the number of office actions increases. Taken together, these trends suggest that these tough examiners provide a more consistently thorough examination, 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.
In contrast, for patents examined by the easiest examiners (“Green”), there is a strong positive correlation between the likelihood of invalidity and the number of office actions (R-squared = 0.83). Finally, for patents examined by Yellow examiners, there is also a significant positive correlation between the likelihood of invalidity and the number of office actions (R-squared = 0.74). These results align with hypotheses of this Article. Examiners that are tougher will be more likely to thoroughly scrutinize patent applications at every stage of examination and less likely to give an allowance. Thus, if these tougher examiners do give an allowance, it suggests the application survived a more rigorous examination process. As a result, it follows that patents granted by Red examiners are less likely to be found invalid when litigated.
Conclusion
The data presented in this Article reveals discernible trends with profound implications for patent prosecution and litigation strategies. Notably, as the number of office actions increases, patents face a heightened risk of invalidation, showing a small positive correlation between prosecution length and invalidity rates. This raises questions about the role prior art plays during lengthy prosecutions. The data extracted for this Article does not include information on the specific types of statutory rejections the patents received during prosecution, but it is possible that a higher number of office actions is indicative of a patent space that is more crowded with prior art, leading to a higher incidence of invalidity.
The results of this Article also show that as average allowance rates decrease from Green examiners to Yellow and Red examiners, invalidity rates similarly decrease. This tracks with the notion that the tougher an examiner (e.g., Red examiners), the more likely it is that the patent underwent rigorous examination, which accordingly would lead to lower invalidity rates. Meanwhile, patents facing validity challenges under less tough examiners (Yellow and Green examiners) saw increasing invalidity rates that one might expect from a less rigorous examination.
In the complex landscape of patent prosecution and litigation, these insights offer actionable guidance for applicants, patent owners, and accused infringers seeking to navigate the intricate dynamics of the patent system.
*This article is excerpted from the authors’ recent article in volume 38, issue 1 of the Harvard Journal of Law & Technology titled “The Myth of “Bad” Patents: Impact of Prosecution Length on Patent Litigation Outcomes” and is republished with permission. The full article is available at https://ssrn.com/abstract_id=5123132.
[1] PatentAdvisor, https://go.patentadvisor.com/.
[2] Lex Machina, https://lexmachina.com/.
[3] The data extracted from PatentAdvisor also includes data from the ITC. However, the Case Outcome field for all ITC cases is listed as “N/A” in PatentAdvisor. As such, ITC cases were not included in the analysis.
[4] As noted previously, our analysis only includes cases from District Court and PTAB cases, and does not include patents litigated at the ITC.
[5] This graph comprises district court cases with outcomes “Invalidity” and “No Invalidity,” and PTAB cases with outcomes “Final Decision: All Claims Unpatentable,” “Final Decision: All Claims Upheld,” and “Institutional Decision: Denied Institution.” “Final Decision: All Claims Unpatentable” maps to “Invalidity,” “Final Decision: All Claims Upheld” maps to “No Invalidity,” and “Institutional Decision: Denied Institution” maps to “No Invalidity.” “Institutional Decision: Denied Institution” maps to “No Invalidity” because it is the primary step where the PTAB considers prosecution history, and because of this, if the board denies the institution, it is equivalent to the board deeming the patent valid.
[6] As used here, “patent space” indicates the available analogous art in the technical field of the patent claims.
[7] This may be an interesting area of investigation for future researchers. After discussing these results with numerous patent litigation experts, the Authors’ educated guess is that the first explanation — that more office actions are indicative of a more crowded patent space — is likely correct.
[8] See In re Cuozzo Speed Techs., LLC, 793 F.3d 1268, 1276–77 (Fed. Cir. 2015) (“This court has approved of the broadest reasonable interpretation standard in a variety of proceedings, including initial examinations, interferences, and post-grant proceedings such as reissues and reexaminations . . . . [W]e have cited the long history of the PTO’s giving claims their broadest reasonable construction.”); Cuozzo Speed Techs., LLC v. Lee, 579 U.S. 261, 283 (2016) (“The Patent Office is legally free to accept or reject such policy arguments on the basis of its own reasoned analysis. Having concluded that the Patent Office’s regulation, selecting the broadest reasonable construction standard, is reasonable in light of the rationales described above, we do not decide whether there is a better alternative as a policy matter.”).
[9] Phillips v. AWH Corp., 415 F.3d 1303, 1312 (Fed. Cir. 2005) (“We have frequently stated that the words of a claim ‘are generally given their ordinary and customary meaning.’”); id. at 1327 (“[W]e have acknowledged the maxim that claims should be construed to preserve their validity.”); id. at 1317 (“In addition to consulting the specification, we have held that a court ‘should also consider the patent’s prosecution history’ if it is in evidence.”).
[10] Manual of Patent Examining Procedure § 2286 (9th ed. 2023) (“Specifically, invalidity in a district court must be shown by “clear and convincing” evidence, whereas in the Office, it is sufficient to show unpatentability by a “preponderance of evidence.” Since the “clear and convincing” standard is more difficult to satisfy than the “preponderance” standard, deference will ordinarily be accorded to the factual findings of the court where the evidence before the Office and the court is the same.”).
[11] Id.; see also 35 U.S.C. § 316(e) (“In an inter partes review instituted under this chapter, the petitioner shall have the burden of proving a proposition of unpatentability by a preponderance of the evidence.”).
[12] See Cuozzo Speed Techs. LLC v. Lee, 579 U.S. 261, 279 (2016) (explaining that the “basic purpose[]” of inter partes review is “to reexamine an earlier agency decision”).
[13] See, e.g., Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co., Ltd., 535 U.S. 722, 733 (2002) (“Prosecution history estoppel requires that the claims of a patent be interpreted in light of the proceedings in the PTO during the application process.”); see also Pacific Coast Marine Windshields Ltd. v. Malibu Boats, LLC, 739 F.3d 694, 700 (Fed. Cir. 2014) (applying Festo).
[14] See PatentAdvisor, https://go.patentadvisor.com/tools/online-manual.php?fn=ETA.pdf (explaining that PatentAdvisor measures examiner toughness using a proprietary metric that takes “into consideration the examiner’s pending portfolio, how long they have been at the Patent Office, number of office actions written and a number of other factors”). PatentAdvisor does not disclose exactly how the various factors are used to determine their ETA metric.
[15] Id.
[16] Id.
[17] Id.; see also Patent Prosecution Analytics No Longer Just a Nice to Have, LexisNexis (Aug. 31, 2021) https://www.lexisnexisip.com/wp-content/uploads/2022/06/Patent-Prosecution-Analytics-No-longer-a-nice-to-have-SLIDES.pdf.
[18] See PatentAdvisor, supra note 16.
[19] Average allowance rate is calculated by summing each individual examiner’s allowance rate of that type and dividing by how many examiners there are of that type.
[20] To calculate “Invalidity Rate,” we took the total number of patents that were marked invalid or not invalid in our dataset. We then separated them by examiner toughness, and then we calculated the respective percentage of patents that were deemed invalid by dividing the total number of invalid patents by the sum of invalid and not invalid patents for that examiner type.
