Opinion
18 Civ. 05290 (CM) 19 Civ. 04693 (CM) 20 Civ. 10195 (CM)
12-21-2023
GEIGTECH EAST BAY LLC, Plaintiff, v. LUTRON ELECTRONICS CO., INC., Defendant.
DECISION AND ORDER
GeigTech East Bay LLC (“GeigTech”) seeks to preclude Lutron Electronics (“Lutron”) from raising an affirmative defense of invalidity based on 35 U.S.C. § 325(e)(2), i.e., the PostGrant Review (“PGR”) estoppel provision. Specifically, GeigTech argues that Lutron is estopped from asserting: (1) grounds already raised in Lutron's prior PGR petitions; and (2) grounds that Lutron could have reasonably raised during PGR. Lutron counters that GeigTech has not met its burden to establish that PGR estoppel applies.
Lutron is correct insofar as the Kirsch and Cid Quintas patents are concerned. In order to obtain the advantage of PGR estoppel, GeigTech has the burden to prove that the patents were or could reasonably have been discovered in a prior search. On this record, it has not satisfied that burden.
However, Lutron does not dispute that the other patents that are the subject of this motion - the Colson, Fraczek, Mitsuhiro, Quill and Nichols patents - either were raised or reasonably could have been raised in post grant review. The same is true of claims arising under 35 U.S.C. §§ 101 and 112. Therefore, PGR estoppel applies, and Lutron may not rely on these patents/issues at trial.
I. PGR Estoppel Presents a Question of Law for Resolution by the Court
As an initial matter, I must first decide whether PGR estoppel is a question of law or fact. Lutron argued in its brief that there is a genuine issue of fact over whether PGR estoppel applies. However, this position seemed nonsensical to me. Genuine issues of material fact are resolved by the ultimate fact finder, which in this case will be a jury. But judges, not juries, decide what issues parties are or are not legally barred from raising. The District of Delaware concluded as much in Innovative Memory Sys., Inc, v. Micron Tech., Inc., No. CV 14-1480-RGA, 2022 WL 4548644, at *5 (D. Del. Sept. 29, 2022), and the court's reasoning in the case seemed correct to me. I asked the parties to clarify their positions.
In their responsive letters, GeigTech agreed that whether GeigTech could argue PGR estoppel presented the court with a question of law, which the court should decide. Lutron danced around the question but ultimately agreed that the issue was not for a jury to decide - it insisted that there was no genuine issue of fact precluding summary judgment in its favor on the question, but agreed that, were the court to find a genuine issue of fact, any ensuing trial would be a bench trial.
So, it appears we are all in agreement that the court will decide whether Lutron can be precluded from asserting invalidity due to PGR estoppel. And I conclude that GeigTech has not met its burden to prove that Kirsch and Cid Quintas are patents that Lutron actually knew about or that a skilled searcher conducting a diligent search reasonably could have been expected to discover during post-grant review (assuming this presents a question of law) - or even to raise a genuine issue of fact (if it presents an issue of fact for the court to rule one), since GeigTech did not bother to present the expert testimony needed to raise such an issue. Therefore, Lutron is not estopped to assert Kirsch and Cid Quintas at trial. It is estopped from asserting the other five patents and Sections 101 and 112.
II. Relevant Procedural History
The court assumes the parties' familiarity with the facts and the procedural history of the case. An extensive discussion of both is available in the court's August 12, 2021, Decision and Order Denying Plaintiffs Motion for Preliminary Injunction and Staying the Remainder of the Case, (Dkt. No. 141), and May 5, 2022, Claim Construction Decision. (Dkt. No. 163).
On September 20, 2023, the court issued a decision and order on multiple motions to . strike, motions to exclude the opinions and proposed testimony of multiple experts and motions for summary judgment. (Dkt. No. 308). On October 10, 2023, GeigTech asked the court to clarify its ruling on the issue of PGR estoppel. (Dkt. No. 331). But the court had made no such ruling. GeigTech had raised and briefed this argument in a first plaintiffs motion for summary judgment, but that motion was stricken. As a result, all arguments raised therein but not raised (and briefed) on the second motion were deemed forfeited. Because GeigTech failed to adequately raise the PGR estoppel issue in its second plaintiffs motion for summary judgment, GeigTech's second motion for summary judgment on PGR estoppel grounds was denied, and the court deemed PGR estoppel abandoned as an issue on summary judgment.
However, the court has since concluded that it is necessary to decide the issue of estoppel in limine. (Dkt No. 343).
III. Relevant Facts
A. The Patents
GeigTech holds the rights to the patents-in-suit - U.S. Patent Nos. 10,294,717 (the “‘717 Patent”) and 10,822,872 (the “‘872 Patent”). (Dkt. No. 202, ¶ 21). The ‘872 patent states that it is a continuation of several prior patents filed by GeigTech, including U.S. Patent No. 9,237,821 (the ‘“821 Patent”) and the ‘717 patent. Id. at ¶ 18.
The ‘717 Patent is classified as “E06B 9/50” and “A47H 1/13” under the Cooperative Patent Classification (“CPC”) system. Id. at ¶ 34. E06B is the classification for “Fixed or Movable Closures for Openings in Buildings, Vehicles, Fences, or Like Enclosures, in General, e.g. Doors, Windows, Blinds, Gates.” (Dkt. No. 201-14). The subclassification “9” refers to “Screening or protective devices for (wall or similar} openings, with or without operating or securing mechanisms; Closures of similar construction.” Id. A47H 1/13 is the subclassification for “Brackets or adjustable mountings for both roller blinds and drawable curtains.” (Dkt. No 206, ¶ 34).
Cid Quintas is a Spanish patent titled “Combined multiple mechanism for shades, curtains and the like with complementary housing structure.” (Dkt. No. 201-30). Cid Quintas was published in 2009 and is classified as E06B 9/56 under the CPC. Id.
Kirsch is a United States patent issued in 1923 and titled “Window Shade Roller Fixture.” (Dkt. No. 201-31). Kirsch is classified as E06B 9/50 under the CPC. (Dkt. Nos. 201-29, 201-31).
B. The Searches
1. The Global and Desai Searches
In August 2018, prior to PGR, Lutron engaged third-party search vendor Global Prior Art Inc. (“Global”) to conduct a prior art search for the ‘821 Patent. (Dkt. No. 207-31, ¶ 3). In December 2018, Lutron did the same for the ‘687 patent application (which eventually became the ‘717 Patent). TcZat ¶ 8. Global's searches did not yield Cid Quintas or Kirsch. Id. at ¶ 13.
In 2019, Lutron prepared to file PGR proceedings against the ‘717 Patent. (Dkt. No. 20732, ¶ 3). As part of its preparation, Lutron asked Nirav Desai to review Global's search results and conduct his own prior art search. Id at ¶¶ 2-5. Desai's search also did not yield Kirsch or Cid Quintas. Id. at ¶ 5. Desai and his team spent approximately 25 hours conducting this search. Id. 2.
The PGR Proceedings
On January 2, 2020, Lutron filed two petitions for PGR with the Patent Trial and Appeal Board. The PGR petitions challenged the validity of the claims of GeigTech's ‘717 patent. In both PGRs (PGR2020-00013 and PGR2020-00012), Lutron asserted the Colson, Fraczek, Mitsuhiro, and Quill patents against the ‘717 Patent. (Dkt. No. 202, ¶¶ 40, 43). Lutron also argued that the asserted patent claims were invalid under 35 U.S.C. § 112. Id. at ¶ 40. Lutron did not assert the Kirsch or Cid Quintas patents. The final decision in the PGR proceeding was issued in August of 2020.
3. The 2020 Koptiw Search
In the summer of 2020 - whether before the PGR's decision or not, the record does not reveal - Lutron had patent attorney Michael Koptiw (“Koptiw”) perform yet another set of prior art searches for the '717 Patent. (Dkt. No. 207-33, ¶ 4). Koptiw is a graduate of the Stevens Institute of Technology and Rutgers Law School. He holds five patents of his own and specializes in complex patent litigation, invalidity analysis and infringement studies. He has an impressive resume and multiple publications. See Attorney Profile: Michael A. Koptiw, https://www.condoroccia.com/attomey/michael-a-koptiw/. Most important, he has conducted hundreds of prior art searches over sixteen years in practice. (Dkt. No. 207-33, ¶ 3).
For his 2020 searches, Koptiw used a third-party commercial prior art search tool called Innography - an artificial intelligence powered search engine that uses text clustering analysis without requiring exact keyword matches. Id. at ¶ 6. The Innography dataset contains 48 million full-text English patent documents and 14 full-text translated jurisdictions, including Spain. Id.
Koptiw used Innography's keyword and semantic search feature to conduct the search. Koptiw selected keywords he derived from the ‘717 Patent claims, including, but not limited to: “motorized,” “roller shades,” “automated,” “electrical drive,” “passage,” “motor in tube,” “concealed,” “obscured,” and “wires.” Id. at ¶ 8. Koptiw decided - given the language in the ‘717 Patent claims and file history - that these terms would return the most relevant prior art. Id.
Koptiw's initial search returned approximately 4,000 results. Koptiw then utilized Innography's analytical tools to craft subsequent searches, in order to obtain a smaller subset of patents to review. Koptiw then “deduped” (i.e., removed duplications among) the entries of this subset and eliminated family members. Id. at ¶ 10. Neither Kirsch nor Cid Quintas was in the resulting final set. Koptiw spent approximately 23 hours conducing this search. Id. at ¶ 12.
Only later did Koptiw learn that references in the Innography datasets from before 1970 (which included Kirsch) were not text searchable.
4. The 2021 Koptiw Search
In 2021, after the PGR proceeding was over, Lutron asked Koptiw to perform yet another prior art search for the ‘717 Patent. Id. at ¶ 13. This time, Koptiw conducted a classification search using the following CPC codes: A47H1/13 (brackets or adjustable mountings for both roller blinds and drawable curtains); E06B9/68 (Operating devices or mechanisms, e.g., with electric drive); and E06B9/42 (parts or details of roller blinds, e.g, suspension devices, blind boxes (brackets or adjustable mountings for roller blinds and drawable curtains)). Koptiw did not conduct a search using the classification “E06B9,” as he believed the resulting universe would have been too large. Koptiw did not search by “E06B9/50” (one of the ‘717 patent's two subclassifications), because he thought it would not return the most relevant prior art.
Koptiw placed his search results into project files, eliminated duplicate entries (“de-duped” the search results), and then eliminated patents that were members of the same family. Neither Kirsch nor Cid Quintas was among the results he obtained.
Additionally, in order to combat Innography's pre-1970 text search issue, Koptiw created thumbnail issues of the patent figures in his search results. While reviewing those images, Koptiw came across a patent (which Koptiw does not identify) that required further investigation (for reasons Koptiw does not explain). Koptiw then conducted an assignee and portfolio search for related references of that patent; one of those references was Cid Quintas.
The additional procedures that yielded Cid Quintas were not part of the search protocols Koptiw normally uses when conducting prior art searches.
Koptiw spent approximately 28 hours conducting this search.
5. The 2021 Lincoln IP Search
In September 2021, Lutron asked Desai to perform a subsequent prior art search for the ‘717 Patent and ‘872 Patent. (Dkt. No. 207-32, ¶ 6). Desai engaged a different third-party search firm, Lincoln IP, to conduct this search. Id. at ¶ 7. Desai provided Lincoln IP with the '717 Patent, the ‘872 Patent and the results of Lutron's prior searches. Id. Lincoln IP's search found neither Kirsch nor Cid Quintas.
So only Koptiw's 2021 search turned up Cid Quintas. None of these searches turned up Kirsch; it is not clear to the court how Lutron eventually discovered Kirsch.
6. Lutron's Prior Art Claims
In the instant case, Lutron asserts certain prior art in its invalidity contentions against the ‘717 Patent, arguing that they “anticipate and/or render obvious the asserted claims” of the ‘717 Patent. (Dirt. No. 201-20, 21). According to the report of Lutron's expert Dr. Eric Maslen, the asserted claims of the ‘717 Patent are invalid (as anticipated or obvious) in view of: (1) Cid Quintas alone or combined with Fraczek or Nichols; (2) Kirsch alone or combined with Fraczek, Quill, or Nichols; and (3) Colson alone or combined with Fraczek, Nichols, or Cid Quintas. Id. at ¶¶ 21-26.
Lutron does not dispute that it was aware of the Colson, Fraczek, Mitsuhiro, Nichols and Quill patents while the PGR was pending. And in fact, four of the patents (Colson, Fraczek, Mitsuhiro, and Quill) were asserted during the PGR, albeit not in combination with either Kirsch or Cid Quintas (because Lutron's searches had not turned up either of these patents). The remaining unasserted patent (Nichols) was a Lutron-owned patent during that same PGR period, so Lutron was obviously aware of it.
7. GeigTech's Evidentiary Arguments
GeigTech asserts that the court should not consider evidence about Lutron's various prior art searches in deciding the PGR estoppel issue, because Lutron disclosed this evidence for the first time in its summary judgment papers. I reject that argument.
Rule 37(c)(1) of the Federal Rules of Civil Procedure provides that: “If a party fails to provide information ... as required by Rule 26(a) or (e), the party is not allowed to use that information ... to supply evidence on a motion, at a hearing, or at a trial, unless the failure was substantially justified or is harmless.” The purpose of Rule 37(c) is to prevent the practice of “sandbagging” an adversary with new evidence. Ebewo v. Martinez, 309 F.Supp.2d 600, 607 (S.D.N.Y. 2004).
However, preclusion is a discretionary remedy, even if “the trial court finds that there is no substantial justification and the failure to disclose is not harmless.” Design Strategy. Inc. v. Davis, 469 F.3d 284,297 (2d Cir. 2006). Further, as preclusion of evidence is a “harsh remedy,” it “should be imposed only in rare situations.” Cates v. Trustees of Columbia Univ, in City of New York, 2019 WL 1433709, at *3 (S.D.N.Y. Apr. 1, 2019) (citations omitted).
GeigTech is correct that Lutron did not previously disclose its prior art searches. But it had no reason to do so until GeigTech asserted PRG estoppel - which GeigTech did not do until it filed its Motion in Support of Summary Judgment on February 2, 2023. Moreover, GeigTech did not disclose any of its own prior art searches until it filed that Motion; it attached its own search strings and the search results as exhibits to that motion. So if I were to exclude Lutron's evidence I would also, in fairness, have to exclude GeigTech's - which would doom its effort to invoke PGR estoppel, since GeigTech has the burden of proof on the issue.
Accordingly, the court will consider evidence about both Lutron's and GeigTech's prior art searches for the purposes of deciding the instant motion.
IV. Legal Standard for PGR Estoppel
PGR estoppel is governed by 35 U.S.C. § 325(e)(2), which states:
The petitioner in a post-grant review of a claim in a patent under this chapter that results in a final written decision under section 328(a) . . . may not assert. . . that the claim is invalid on any ground that the petitioner raised or reasonably could have raised during that post-grant review.
Courts in this Circuit have not addressed the proper standard for determining whether a ground “reasonably could have [been] raised” during PGR. However, other courts interpret this as encompassing “any patent or printed publication that a petitioner actually knew about or that a skilled searcher conducting a diligent search reasonably could have been expected to discover.” Ironburg Inventions Ltd. v. Valve Corp., 64 F.4th 1274,1298 (Fed. Cir. 2023) (quotations omitted); Wi-LAN Inc. v. LG Elecs., Inc., 421 F.Supp.3d 911, 924 (S.D. Cal. 2019); SiOnyx, LLC v. Hamamatsu Photonics K.K., 330 F.Supp.3d 574, 602 (D. Mass. 2018); Clearlamp, LLC v. LKQ Corp., 2016 WL 4734389, at *8 (N.D. Ill. 2016); GREE, Inc. v. Supercell Oy, 2020 WL 4999689, at *2 (E.D. Tex. July 9, 2020), report and recommendation adopted, 2020 WL 4937111 (E.D. Tex. Aug. 24,2020). While some of these courts applied this standard in the context of an IPR estoppel claim, the PGR and IPR estoppel provisions are substantively identical in this context. See GREE, 2020 WL 4999689, at *3. Both parties argue under this standard, and the court will use it.
Finally, the burden is on GeigTech to prove, by a preponderance of the evidence, that they have met this standard. See Ironburg, 64 F.4th at 1299.
V. GeigTech's Motion to Estop Lutron From Relying on Kirsch and Cid Quintas Is Denied
Although the Federal Circuit has not refined exactly what facts or circumstances qualify as “a skilled searcher conducting a diligent search,” district courts have concluded that a skilled searcher would have been able to find a particular reference when the moving party (in this case, GeigTech): “(1) identif[ies] the search string and search source that would identify the allegedly unavailable prior art and (2) present[s] evidence, likely expert testimony, why such a criterion would be part of a skilled searcher's diligent search.” TrustID, Inc. v. Next Caller Inc., C. A. No. 18-172-MN, 2021 WL 3015280, at * 1 (D. Del. July 6, 2021) (emphasis added) (quoting Clearlamp, 2016 WL 4734389, at *9); Palomar Techs., Inc. v. MRSI Sys., LLC, 373 F.Supp.3d 322,332 (D. Mass. 2019); Bos. Sci. Corp. v. Cook Grp. Inc., No. 117CV03448JRSMJD, 2023 WL 1452172, at *6 (S.D. Ind. Jan. 31, 2023); Asetek Danmark A/S v. CoolIT Sys. Inc,, No. 19-CV-00410-EMC, 2019 WL 7589209, at *8 (N.D. Cal. Dec. 30, 2019); Pavo Sols. LLC v. Kingston Tech. Co., Inc., No. 814CV01352JLSKES, 2020 WL 1049911, at *4 (C.D. Cal. Feb. 18, 2020); Acceleron, LLC v. Dell, Inc, No. 1:12-CV-4123-TCB, 2020 WL 10353767, at *4 (N.D.Ga. Mar. 30, 2020), aff'd, No. 2022-1620, 2023 WL 4503189 (Fed. Cir. July 13,2023); M-I LLC v. FPUSA, LLC, No. 5:15-CV-406-DAE, 2020 WL 13413830, at *11 (W.D. Tex. Aug. 20, 2020).
GeigTech fulfills the first requirement, but not the second.
A. GeigTech Identifies a Search. String and Search Source That Would Identify the Kirsch and Cid Quintas Patents
GeigTech's presents the court with URL links to two searches conducted on Google Patents. Sparse though this evidence be, it is sufficient to establish that there is a search string and source that would identify Kirsch and Cid Quintas.
Kirsch Search
GeigTech identifies the following terms from the Kirsch patent: “window shade”, “roller”, “bracket”, “support”, “surface”, “window casing”, “wall” and “mounted.” (Dkt. No. 201-31). GeigTech chose these terms because they are the “same or similar” to the following terms identified in the ‘717 patent: “window shade”, “roller”, “curtain”, “blinds”, “motor”, “bracket”, “support surface”, “window casing”, “wall”, “ceiling”, “mount”, “key” and “clutch”. (See Dkt. No. 201-7). Searching Google Patents using these terms together with the search term “E06B9,” and filtering the search to patents before the '717 Patent's 2012 priority date, returns twenty-six results, including Kirsch. (Diet. No. 201, ¶ 67).
Cid Quintas Search
GeigTech identifies the following terms from the Cid Quintas patent: “blinds”, “curtains”, “motorized”, “bracket”, “support”, “roller”, “mount”, “key”, “ceiling” and “wall.” Id. GeigTech chose these terms for the same reason it chose the Kirsch search terms. Searching in Google Patents for these terms, along with the search term “E06B9,” and filtering by the 2012 priority date, returns thirty results, including Cid Quintas. Id.
GeigTech's evidence gives rise to some questions in my mind.
First, while GeigTech identifies “support” as a key term in the Kirsch Patent, and claims it as a search criterion, the URL of the actual search provided by GeigTech does not include that term.
Second, GeigTech does not identify “surface” as a term from Cid Quintas, and does not claim it as a search criterion, yet the URL indicates that “surface” was part of the actual search.
Third, GeigTech uses “E06B9” and “E06b9” as search terms for the Cid Quintas search, but only “E06B9” for the Kirsch search.
Fourth, the URL indicates that GeigTech uses both “mount” and “wall” twice in the Cid Quintas search.
However, obtaining answers to those questions would not change my conclusion that GeigTech has satisfied the first prong of its evidentiary burden, by providing the court with search strings that would have located the allegedly unavailable prior art.
B. GeigTech Does Not Offer Evidence to Establish That These Search Criterion Would Reasonably Be Part of a Skilled Searcher's Diligent Search
The second prong of GeigTech's burden is to explain why a skilled searcher's diligent search would have been equivalent to its searches. See M-I LLC, 2020 WL 13413830, at *12. GeigTech offers no such explanatory evidence.
GeigTech's “evidence” consists solely of the search strings that were selectively assembled by its own attorneys to bolster ex post facto arguments. To the extent that the court can infer anything from these search strings, it appears that GeigTech essentially “reverse engineered” its search terms with the benefit of hindsight. It knew what patents it wanted to find - Kirsch and Cid Quintas - so it constructed the search strings by looking for terms that it knew were in both those patents and in the ‘717 patents.
But anyone who knows what he is looking for can “reverse engineer” a successful search. GeigTech's burden is to demonstrate why a skilled searcher who was not aware of the Kirsch or Cid Quintas patents would have chosen that particular combination of terms when structuring a diligent search.
Because the choice of these search terms was clearly the product of hindsight, one cannot infer that a skilled researcher would have found the patents from the fact that these particular searches yielded the patents. The court's reasoning in Palomar Techs., Inc. v. MRSI Sys., LLC, No. 18-10236-FDS, 2020 WL 2115625, at *3 (D. Mass. May 4, 2020), is instructive:
To be clear, the standard cannot be that a reasonable search could have, or might have, discovered the disputed reference. It is almost always possible to construct a search scenario-particularly with the benefit of hindsight-under which a skilled searcher could locate a reference with a relatively small number of steps. Instead, the standard must account for the fact that the relevant databases are huge, the technology is often complex, and there are almost infinite ways to construct a search. The touchstone is reasonableness, not perfection. Accordingly, the appropriate standard for the objective prong is one of probability, not possibility: that is, whether it is more probable than not that a skilled searcher conducting a diligent search reasonably could have been expected to discover the disputed reference.
In Palomar, the moving party proffered an expert witness who opined, inter alia, on subjects such as: how he used classifications in his patent searches that turned up the relevant prior art, the most important steps in selecting the appropriate keywords and classifications in those searches, why the prior searches conducted by the opposing party were not reasonably diligent, examples of search strings which would have yielded the relevant patents, descriptions of searches actually conducted by the expert and an explanation of why those search results would be relatively manageable for a searcher to look through. Id. at *6-7. Despite this expert testimony, the court in Palomar still found this expert evidence insufficient, as “most importantly” all this evidence suffered from “the same basic issue: [it was] clearly influenced by hindsight analysis.” Id. at *10.
GeigTech's evidentiary record suffers from the same basic defect - its searches were constructed after the Kirsch and Cid Quintas patents had been otherwise located, which makes it easy to ensure that the chosen search terms would turn them up in a prior art search. Significantly, GeigTech's showing is far less convincing than was the case in Palomar, because GeigTech does not offer the court expert evidence at all, whereas the unsuccessful plaintiff in Palomar did offer expert testimony to try to explain its search process and why the searches conducted by its opponent were unreasonably inadequate. Lacking any evidence of that nature, this court cannot conclude that GeigTech has come anywhere close to meeting its evidentiary burden. GeigTech does not offer the court information about any “search process.” And its selection of terms appears to be in some instances inexplicable except by concluding that the contents of the Kirsch and Cid Quintas patents dictated the contents of the search strings. For example, GeigTech used “curtain,” “blinds” and “ceiling” in the search that yielded Cid Quintas; but the words “curtains” and “blinds” do not appear in the ‘717 patent, and “ceiling” is only used once in a patent running to twenty-six pages and literally tens of thousands of words. Without some expert testimony explaining why these terms - which appear rarely in the patent that Lutron's prior art searchers were working off - would have been chosen by a skilled and diligent searcher who had never heard of Cid Quintas, the court has no basis to conclude that a search not employing the terms “curtain,” “blinds” and “ceiling” was unreasonable or less than diligent. Nor can the court conclude as a matter of logic that a search employing those terms was necessarily reasonable or diligent, since the source for those terms was obviously the Cid Quintas patent itself.
Moreover, GeigTech does not explain why it did not include terms such as “passage”, “wiring” or “electrical”, or offer opinion testimony about why a skilled searcher conducting a diligent search would not have included those terms. This is significant because the words “wiring” and “electrical” are mentioned repeatedly throughout the ‘717 Patent and its abstract, but do not appear in either Kirsch or Cid Quintas, while “passage” appears just once in each of those patents but multiple times in the ‘717 patent. It seems logical that a reasonable prior art searcher would look for patents containing words of such import that they were repeatedly used; GeigTech offers no evidence to refute that logic. Nor does GeigTech explain, for example, why its use of “mounted” (which appears once in the ‘717 patent but seven times in Kirsch and thirteen times in Cid Quintas) instead of “mounting” (a word used nineteen times in both the ‘717 patent and in Cid Quintas but not at all in Kirsch) - or even the simpler and broader word “mount” - was more reflective of what a skilled searcher would do if he knew nothing of Kirsch or Cid Quintas and was conducting a reasonable and diligent search for prior art relating to the ‘717 patent.
GeigTech argues that Lutron's prior art searches were inadequate. But its argument is essentially conclusory. Its only real support for that argument is that Lutron's searches failed to turn up Kirsch and Cid Quintas. The law is clear, however, that failure to find a reference does not without more render a search inadequate or prove that a “a skilled searcher conducting a diligent search” would have discovered Kirsch and Cid Quintas. See Ironburg, 64 F.4th at 1299. But GeigTech relies on nothing else.
In sum, GeigTech has not demonstrated that its search strings consisted of terms that a skilled searcher who was unaware of either Kirsch or Cid Quintas would have used in order to conduct a diligent search for prior art that anticipated the ‘717 patent. GeigTech “does not provide sufficient details, such as an explanation of the search process ... to enable the Court to assess whether those searches were diligent.” See Innovative, 2022 WL 4548644, at *4. The fault lies with GeigTech, which has the burden to prove that its search terms would be part of a skilled searcher's diligent search.
That being so, Lutron is not estopped to rely on those two patents.
I could stop at this point, because I have found that GeigTech has failed to meet its burden of proof. But I will address other arguments that it makes in support of its motion. They are: (1) the eventual discovery of Kirsch and Cid Quintas means they could have been discovered earlier; (2) the deposition testimony of Lutron's POSA establishes that the patents could have been discovered earlier; and (3) the fact that the two patents shared certain classifications with the ‘717 patent means that a reasonably diligent search would have uncovered them earlier. None of these arguments alters my ultimate conclusion.
C. Koptiw's Eventual Discovery of the Cid Quintas Patent Does Not Establish That Lutron Could Have Reasonably Discovered Kirsch or Cid Quintas Before PGR
Because Koptiw's final search in 2021 discovered Cid Quintas, GeigTech argues that Lutron could reasonably have discovered Cid Quintas sooner - by conducting a skilled search -and asserted it in the PGR (which was over by the end of the summer of2020). But early discovery does not necessarily follow from ultimate discovery.
GeigTech misunderstands the relevant inquiry. “The inquiry into what a skilled and diligent searcher would reasonably have discovered is ultimately concerned with what the searcher of ordinary skill would find through reasonable diligence and not what an actual researcher in fact did find through whatever level of diligence she exercised.” Ironburg, 64 F.4th at 1299 (emphasis in original). Estoppel does not preclude the use of any prior art that a parly eventually discovered. A party must discover a reference in order to assert it as a defense; the PGR estoppel standard takes this into account by adding the term “reasonably” before “could.” See Bos. Sci. Corp., 2023 WL 1452172, at *8. “Scorched earth” tactics or extraordinary measures that result in a patent's eventual discovery are not necessarily relevant “to a determination of what would have been discovered by an ordinarily skilled searcher acting with merely reasonable diligence.” Ironburg, 64 F.4th at 1299.
Other courts have repeatedly rejected GeigTech's argument that eventual discovery necessarily means earlier discovery. “[T]he mere fact that defendants later found [the reference] and included it in their . . . invalidity contentions . . . [does not independently] show that a reasonably diligent search would have uncovered it [pre-petition].” Si Onyx, 330 F.Supp.3d at 602. Koptiw's eventual discovery of Cid Quintas does not independently establish GeigTech's burden. GeigTech still must present factual evidence showing that a skilled searcher reasonably would have found Cid Quintas at an earlier date. Other than its hindsight search string, it offers no such evidence. And it certainly does not establish that a skilled searcher would have found Kirsch, which none of the prior art searches discussed in Lutron's papers turned up. See id at 603.
That is not to say Koptiw's 2021 search is immaterial. The circumstances of Cid Quintas' discovery are relevant to determining what a reasonably diligent search would and would not find. See Wi-LAN, 421 F.Supp.3d at 924. However, I take into account the fact that Koptiw - an individual whose credentials easily establish him as an expert in the field of prior art searches finally discovered Cid Quintas using methods he considers atypical and unusual, i.e., searching thumbnail imagery and narrowing search results by patent classification. Koptiw explained that he used these methods because the 2021 search was his second prior art search for the ‘717 Patent, and he did not think it appropriate simply to repeat the search he had already conducted - a search that was guaranteed to produce the same result.
It is also significant that Koptiw did not initially discover Cid Quintas in his second search. It was only after he ran classification searches that failed to yield the patent that Koptiw decided to search through the thumbnail imagery of the search results' figures which led to the discovery of a different patent, which in turn led to the discovery of Cid Quintas.
Finally, while the search that found Cid Quintas was Koptiw's second prior art search, it was Lutron's fifth; the three searches that Koptiw did not conduct also failed to yield Kirsch and Cid Quintas. It is of great significance to this court that the search that was conducted by a professional patent firm after the discovery of Cid Quintas did not turn up Cid Quintas - even though the search firm was provided with the results of Koptiw's thumbnail imagery, which had finally (if indirectly) yielded Cid Quintas.
Thus, the court does not find the fact that Lutron eventually discovered Cid Quintas to be dispositive of the PGR estoppel question.
D. Dr. Maslen's Expert Report and Deposition Testimony is Irrelevant to the PGR Estoppel Inquiry
GeigTech argues that the report and testimony of Lutron's expert Dr. Maslen establishes that Lutron knew or reasonably should have known about Cid Quintas and Kirsch prior to or during the PGR. I do not find GeigTech's arguments persuasive.
Lutron retained Maslen, on a date neither Maslen nor Lutron provides, as its technical expert on patent non-infringement and invalidity. Maslen offers two expert reports. The first, his invalidity report, was issued on November 22, 2022, and the second, his noninfringement report, issued on December 22, 2022. In these two reports, Maslen opines that (1) Elements of the ‘717 Patent are not infringed by the Palladiom shading system; and that (2) the asserted claims of the '717 Patent are invalid as anticipated or obvious in view of (i) Cid Quintas alone or combined with Fraczek or Nichols, (ii) Kirsch alone or combined with Fraczek, Quill, or Nichols, and (iii) Colson alone or combined with Fraczek, Nichols, or Cid Quintas. (Dkt. No. 218-1 ¶¶ 21-23; Dkt. No. 218-2 ¶¶l 77, 202).
Geigtech's first argues that Maslen's report's background section includes numerous purported prior art references, which Maslen asserts are within “the knowledge of a person of ordinary skill in the art.” However, neither Cid Quintas nor Kirsch is included among these references. So Maslen's statement does not suggest that Cid Quintas or Kirsch would, could or should have been found earlier.
Next, Geigtech points out that Maslen agreed during his deposition that it would be “reasonable for someone looking for prior art... to come across Kirsch.” (Dkt No. 202, ¶ 71). But Maslen is an engineering consultant who has been involved in the field of mechanical engineering for over forty years. He is offered as a “Person Having Ordinary Skill in the Art” (“POSA”) - not as an expert in prior art searches. A POSA's qualifications include “(1) 1-2 years of experience designing or installing powered roller blinds or 3-4 years of designing electro-mechanical consumer products; and (3) at least an associate's degree in mechanical engineering, industrial design, or related fields.” (Dkt. No. 201-21). Dr. Maslen did not include anything about experience with prior art searches in his definition of a POSA. That makes sense, since a POSA like Dr. Maslen testifies about issues like non-infringement and invalidity - not prior art searches.
Maslen's expertise does not extend to prior art searches. He admits as much. Lutron used patent attorneys and prior art search firms, not mechanical engineers, to search for prior art. Maslen did not conduct any prior art searches; he testified that he did not even “know exactly what a routine prior art search is.” (Dkt. No. 206). Lutron provided Dr. Maslen with the prior art references that he used in reaching his conclusions. And at this deposition, Maslen was not discussing prior art search procedures or practices when he made this statement.
So Maslen's statement falls far short of the “expert evidence” GeigTech would need to satisfy its evidentiary burden.
E. Kirsch and Cid Quintas' Shared Classification With the ‘717 Patent Does Not Render Lutron's Searches Unreasonable or Not Diligent, Nor Does It Establish That GeigTech's Searches Were Reasonable or Diligent
The CPC is a classification system used by the United States Patent Trademark Office (“USPTO”), jointly developed with the European Patent Office (“EPO”) and put into use as of January 1, 2013. The CPC harmonizes the USPTO and EPO's prior patent classification systems into a common system. (Dkt. No. 201-25). The CPC system is organized into nine sections, which are divided into various classes and subclasses. (Dkt. Nos. 201-27, 201-28). A patent examiner assigns a classification to the patent application based on its contents. Some older patents that predate the CPC system have been classified as well, e.g., Kirsch. The '717 Patent, Cid Quintas and Kirsch all have the same section (E), the same class (E06), the same subclass (E06B), and the same main group (9) under the E06B subclass.
GeigTech argues that because Kirsch, Cid Quintas and the '717 Patent share this classification, a skilled and diligent search would have found them. (Dkt. No. 200,19). In support, GeigTech claims that “Anyone searching by the E06B classification of the '717 Patent would . .. reasonably expect to find both Kirsch and Cid Quintas, especially because Kirsch, Cid Quintas, and the '717 Patent also share the same main group (9) under the E06B subclass.” (Dkt. No. 200, .. ¶21).
It certainly seems like shared classification could be relevant to constructing a prior art search. But GeigTech offers no evidence that it is always the best or most diligent way to find prior art, or that skilled prior art searchers always or even generally rely on shared classification when constructing their prior art searches. In fact, GeigTech's own “reverse engineered” search string for Cid Quintas did not include the A47H1 classification, although Cid Quintas shares that classification with the ‘717 patent - indeed, it was Koptiw's addition of the A47H1 classification to his search that yielded the reference that ultimately (if indirectly) led Lutron to Cid Quintas. GeigTech's failure to employ classification in its own searches is equivalent to an admission that an expert prior art searcher would not necessarily choose to rely on shared classification in constructing a diligent and reasonable search.
The reasons for that are explained by Mr. Koptiw, who (unlike any witness proffered by GeigTech) is a skilled prior art searcher. Koptiw claims that “conducting prior art searches using CPC classifications are not as effective as conducting searches on keywords and are oftentimes inaccurate because placing a patent within a class and subclass involves a series of judgments by the reviewer, a process that inevitably results in inconsistencies.” (Dkt. No. 207-33, ¶ 16). According to Koptiw, a keyword search based on a combination of terms that appear in a patent's claims (and associated synonyms) is typically the best approach to find prior art with the highest relevance. Thus, Koptiw does not normally use classifications as part of his prior art searches. That is the only expert testimony in the record on this subject.
Additionally, if using shared classification does not significantly narrow a field to a searchable number of possible prior art references, it stands to reason that it would not be a particularly useful way to conduct a prior art search. GeigTech's own submission to the court makes this point. On Google Patents, a search under the “E06B9” classification returns more than 100,000 results, (Dkt. No. 207-33), - far more than any prior art searcher could possibly examine in a reasonable search. The subclassification “E06B9/68” (Operating devices or mechanisms, e.g, with electric drive) returns approximately 36,000 results, the subclassification “E06B9/42” (parts or details of roller blinds, e.g., suspension devices, blind boxes (brackets or adjustable mountings for roller blinds and drawable curtains) returns approximately 26,000 results.
Although Kirsch and the ‘717 patent share a “E06B 9/50” subclassification, neither party offers evidence as to the number of patents in that classification.
Of course, one might think that a skilled searcher would use both keywords and classifications. And clearly they sometimes do. Koptiw used classifications when he conducted his first 2021 search. Yet that search, using classifications, did not locate Cid Quintas or Kirsch! Moreover, Koptiw discovered Cid Quintas, not through the “E06B9/68” subclassification that GeigTech emphasizes, but through a patent classified as “A47H1/13,” which is the ‘717 patent's second subclassification. As noted above, this classification - which has but 1600 entries, a far more searchable number is not identified by GeigTech as part of a reasonable search for Cid Quintas.
So I cannot conclude that simply doing an E06B9/68 classification search at the outset of the search process would have yielded the prior art patents. Lutron's actual search history, which is quite extensive, suggests the opposite.
If GeigTech wished to raise a question about either the diligence of its searches or the lack of diligence in Lutron's, it would most appropriately have done so by introducing expert evidence. However, GeigTech has not presented any evidence from an expert tending to show that use of classifications is always part of a reasonable prior art search, and why. The court will not assume . that a search is unreasonable simply because it does not use classifications. Nor will the court assume a search is reasonable simply because it uses classifications - especially when the only expert evidence I have (from Koptiw, who is an expert in conducting patent searches) is to the contrary.
VI. GeigTech's Motion is Granted as to the Colson, Fraczek, Mitsuhiro, Quill, and Nichols patents, and to claims under 35 U.S.C. §§ 101 and 112.
GeigTech argues - and Lutron does not dispute - that Lutron asserted the Colson, Fraczek, Mitsuhiro, and Quill patents against the ‘717 Patent in a PGR that resulted in a final written decision. (Dkt. No. 201, ¶¶ 40,43). Accordingly, Lutron is estopped from arguing that the asserted claims are invalid over these patents.
The Nichols patent is a Lutron-owned patent, and Lutron was the owner of Nichols at the time of the ‘717 PGR petition. (Dkt. No. 201, ¶44; Dkt. No. 201-23). Accordingly, Lutron “actually knew about” the Nichols patent, and “could have raised” the patent in its PGR petition. Thus, Lutron is estopped from asserting it at the trial of this case.
Finally, in the ‘717 PGR, Lutron argued that the asserted claims were invalid under 35 U.S.C. § 112. (Dkt. No. 201, ¶ 40). Lutron does not dispute this, nor does it give a reason why it could not have reasonably raised a 35 U.S.C § 101 claim in the PGR. Accordingly, Lutron is estopped from challenging the asserted claims under 35 U.S.C. §§ 101 and 112.
The Clerk of Court is directed to remove the motion at Docket 332 from the court's list of open motions. This constitutes the opinion and order of the court. It is a written opinion.