NXP USA Inc v. Impinj Inc

[5] UNITED STATES DISTRICT COURT 6 WESTERN DISTRICT OF WASHINGTON AT SEATTLE [7] NXP USA, INC., and NXP B.V., CASE NO. 2:20-cv-01503-JHC [8] Plaintiffs, ORDER RE: MOTION TO EXCLUDE [9] CERTAIN OPINIONS OF LAUREN R. v. KINDLER AND NXP’S MOTION IN [10] LIMINE IMPINJ, INC., [11] Defendant. [14] There are two motions before the Court. First is NXP’s motion to exclude certain [15] opinions of expert Lauren R. Kindler. Dkt. # 279; see also Dkt. # 339 (reply brief). Impinj [16] opposes the motion. Dkt. # 326. Second is one of NXP’s motions in limine. Dkt. # 458. The [17] Court previously reserved ruling on this motion in limine. Dkt. # 516 at 2. [18] The Court issued an order on June 2, 2023, setting a Daubert hearing. Dkt. # 513. That [19] order allowed the parties to submit additional materials and asked several questions. Id. Kindler [20] filed a declaration on June 6, 2023, Dkt. # 517, and NXP filed a supplemental response on June [21] 7, 2023, Dkt. # 518. The Court held a Daubert hearing on June 8, 2023. Dkt. # 521. [22] For the reasons below, the Court: [24] 1 (1) GRANTS NXP’s motion to exclude certain opinions of Lauren R. Kindler in part, 2 DENIES it in part, and STRIKES it in part as moot. Dkt. # 279. 3 (2) GRANTS in part and DENIES in part NXP’s second motion in limine. Dkt. # 458. 4 I 5 LEGAL STANDARDS 6 An expert witness may provide opinion testimony if “(a) the expert’s scientific, technical, 7 or other specialized knowledge will help the trier of fact to understand the evidence or to 8 determine a fact in issue; (b) the testimony is based on sufficient facts or data; (c) the testimony 9 is the product of reliable principles and methods; and (d) the expert has reliably applied the 10 principles and methods to the facts of the case.” Fed. R. Evid. 702. 11 “Before admitting expert testimony into evidence, the district court must perform a 12 gatekeeping role of ensuring that the testimony is both relevant and reliable under Rule 702.” 13 United States v. Ruvalcaba-Garcia, 923 F.3d 1183, 1188 (9th Cir. 2019) (quotation marks 14 omitted) (quoting Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 597 (1993)). But 15 “[u]nder Daubert, the district judge is a gatekeeper, not a fact finder.” Primiano v. Cook, 598 16 F.3d 558 , 564–65 (9th Cir. 2010) (citation and quotation marks omitted). “Shaky but admissible 17 evidence is to be attacked by cross examination, contrary evidence, and attention to the burden of 18 proof, not exclusion.” Id. at 564 . “In Daubert, the Court addressed the proper standard for 19 admitting expert testimony and emphasized that the focus ‘must be solely on principles and 20 methodology, not on the conclusions that they generate.” Apple Inc. v. Motorola, Inc., 757 F.3d 21 1286, 1313–14 (Fed. Cir. 2014) (quoting Daubert, 509 U.S. at 595 ), overruled on other grounds 22 by Williamson v. Citrix Online, LLC, 792 F.3d 1339 (Fed. Cir. 2015) (en banc). [24] 1 II 2 MOTION TO EXCLUDE CERTAIN OPINIONS OF KINDLER 3 A. Panduit Analysis for Lost Profit Damages 4 NXP argues that Kindler misapplied the Panduit factors used to calculate lost profit 5 damages. Dkt. # 279 at 10–12. But in a previous order, the Court held that NXP may not seek 6 lost profit damages. Dkt. # 452 at 13–21. 7 Accordingly, the Court strikes this portion of the motion as moot. 8 B. Reliance on Oliver for Valuation Opinions 9 “[W]here multi-component products are involved, the governing rule is that the ultimate 10 combination of royalty base and royalty rate must reflect the value attributable to the infringing 11 features of the product, and no more.” Ericsson, Inc. v. D–Link Sys., Inc., 773 F.3d 1201 , 1226 12 (Fed. Cir. 2014). Thus, “[w]hen the accused infringing products have both patented and 13 unpatented features, measuring this value requires a determination of the value added by such 14 features.” Id.; see also ResQNet.com, Inc. v. Lansa, Inc., 594 F.3d 860, 869 (Fed. Cir. 2010) 15 (“[A] reasonable royalty analysis requires a court to hypothesize, not to speculate. . . . [T]he trial 16 court must carefully tie proof of damages to the claimed invention’s footprint in the market 17 place.”). 18 In her report, Kindler concludes that she “understand[s]” that the accused FastID feature 19 accounts for approximately 1% – 2% of the value of Impinj’s products, while the accused 20 TagFocus feature accounts for approximately 3% – 4% of the value. Dkt. # 281-3 at 37. She 21 also concludes that the unaccused features would comprise “no less than 90% of the value of 22 Monza 4 relative to Monza 3.” Id. These conclusions are based on “discussion[s] with Mr. Ron 23 Oliver,” an Impinj employee. Id. Kindler also bumps the upper bound of her estimate of the [24] 1 value of both TagFocus and FastID to 5% each. Id. (“Assuming that (i) FastID accounts for 1% 2 – 5% and (ii) Tag Focus accounts for 3% – 5% of the value of Monza 4 relative to Monza 3. . .”). 3 NXP argues that Kindler’s apportionment opinions are unreliable and should be 4 excluded. Dkt. ## 279 at 12–16, 339 at 4–6. Specifically, NXP asserts that “Kindler was 5 required to apportion the value of the accused FastID and TagFocus features in a manner that 6 captures the value attributable to the asserted patents.” Id. at 12. But, NXP says, the 7 apportionment percentages that Kindler purportedly assigns to these features are not based on her 8 own analysis or conclusions. Id. Instead, NXP says that Kindler’s opinions are based entirely on 9 the apportionment figures provided to her by Impinj’s employee, Ron Oliver, which Kindler then 10 uses as the foundation of her analysis. Id. The Court agrees with NXP that Kindler’s 11 quantitative apportionment opinions are not based on a sufficiently reliable methodology and are 12 therefore subject to partial exclusion. But the Court believes her qualitative apportionment 13 opinions are sufficiently reliable, so the Court does not exclude those. 14 Calculating a reasonable royalty rate necessarily entails estimation. But an expert’s 15 apportionment analysis may not be “plucked out of thin air.” LaserDynamics, Inc. v. Quanta 16 Comput., Inc., 694 F.3d 51, 69 (Fed. Cir. 2012); see also id. (criticizing expert’s methodology in 17 which his “one-third apportionment to bring his royalty rate down from 6% per ODD to 2% per 18 laptop computer appears to have been plucked out of thin air based on vague qualitative notions 19 of the relative importance of the ODD technology.”). “Experts must follow some discernable 20 methodology, and may not be a black box into which data is fed at one end and from which an 21 answer emerges at the other.” NetFuel, Inc. v. Cisco Sys. Inc., No. 5:18-CV-02352-EJD, 2020 22 WL 1274985 , at *2 (N.D. Cal. Mar. 17, 2020) (quoting GPNE Corp. v. Apple, Inc., 2014 WL 23 1494247, at *4 (N.D. Cal. Apr. 16, 2014)). “While the Federal Circuit allows for ‘some 24 approximation’ in the reasonable royalty context, this ‘does not negate the Federal Circuit’s 1 requirement of sound economic and factual predicates for that analysis.’” Id. at *7 (quoting 2 Cornell Univ. v. Hewlett-Packard Co., 2008 WL 2222189 , at *2 (N.D.N.Y. May 27, 2008) 3 (Rader, C.J., sitting by designation) (quotation marks omitted)). 4 The Court concludes that Kindler’s quantitative apportionment analysis does not adhere 5 to a sufficiently reliable methodology. Kindler relies almost entirely on the apportionment 6 figures provided to her by Impinj employee, Ron Oliver. For example, her report states: “Based 7 on a discussion with Mr. Ron Oliver, I also understand that (1) FastID accounts for 8 approximately 1% – 2% and (2) TagFocus accounts for approximately 3% – 4% of the value of 9 Monza 4 relative to Monza 3.” Dkt. # 281-3 at 37. She also concludes—“[b]ased on a 10 discussion with Mr. Ron Oliver”—that the unpatented features would comprise “no less than 11 90% of the value of Monza 4 relative to Monza 3.” Id. Her report provides no meaningful 12 explanation of how she arrived at these figures aside from her conversations with Oliver. 13 During her deposition, Kindler stated that her apportionment estimates were provided to 14 her by Oliver: 15 I got [Oliver’s] reaction to Mr. Haas’s opinions but also his own independent evaluation as to the relative contribution of these features . . . And then I asked 16 him “Well, if you had to come up with a specific quantifiable contribution for FastID and TagFocus, like, what do you think would be a reasonable amount?” 17 And he gave me the 1 to 2 percent for FastID, and I think it was . . . 3 to 4 percent for TagFocus. . . . [W]e had a lengthy call talking through the various features of 18 Monza 4 relative to Monza 3 and what Mr. Haas and NXP’s technical expert had opined to and got his reaction to all that and his own independent assessment 19 based on his own expertise. 20 Dkt. # 281-7 at 15–16; see also id. at 19 (conceding that her “upper bound” for the benefit of 21 each allegedly patented feature was based on what “he [Oliver] concluded” about the relative 22 benefits of the product features). Kindler admitted that she could not explain how Oliver arrived 23 at his conclusions. See id. at 18 (“Q: On what basis did Mr. Oliver form his opinion that 24 TagFocus feature was worth the whole – the FastID feature? A: . . . Ultimately, he – I don't know 1 why. I mean, we didn’t get into the details as to why, but he concluded that the relative 2 contribution of TagFocus would be approximately 3 to 4 percent based on his experience.”); id. 3 at 23 (“Well, I said at the beginning I don’t know how he came up with the numbers.”); id. at 4 12–13, 18–19. She admitted that she did not understand why, for example, Oliver gave 5 TagFocus more weight than FastID. Id. at 18–19 (“Q: But you don’t know whether [the 6 frequency of use of the accused features was] the reason he opined to 3 to 4 percent for 7 TagFocus versus 1 to 2 percent for FastID? A: No. I’m – I’m connecting some dots, but no, I 8 don't know if that’s the reason.”). And she admitted that if Oliver’s conclusion that “at least 9 90%” of the products’ value came from unpatented features was incorrect, that would have 10 affected the outcome of her analysis. Id. at 17, 27 (“Now, the 90 percent could matter, if he’s 11 wrong on that, but I don’t have any reason to believe that he’s wrong on that.”). 12 Worse yet, Oliver—whose conclusions serve as the backbone of Kindler’s apportionment 13 analysis—admitted in deposition (in the co-pending Texas litigation) that he generally could not 14 quantitatively isolate the value of a given patented feature. For example, he said that he “cannot 15 assign to each feature individually a value and then further within each feature I cannot dissect 16 that down into individual innovations and the contribution from [the patents].” Dkt. # 341-1 at 17 10; see also id. at 8 (“Q: You said it would be hard to assign a percentage, but is it possible to 18 even assign a percentage? . . . A: That would be a question for the market. . . . [M]y only 19 response would be that that would be a question for the market, that Impinj does not have enough 20 information to answer.”); id. (“[I]t would be really hard to assign a – a percentage value to 21 that.”); id. at 6 (“Q: What is the methodology that Impinj undertakes to do that, you mentioned 22 [customer] comments, is there some – does that get formalized somehow in terms of a ranking or 23 a priority or a value? A: I don’t know that there is a formalized ranking. In my experience, there 24 have been discussions, sometimes involving sales personnel, sometimes involving marketing that 1 just on a qualitative basis look at what’s the perceived value. It is very hard to assign any kind of [2] a quantity in the ranking system.”).1 Indeed, when questioned about apportionment, Oliver 3 implied that he is not particularly well-suited to assess relative value of features. See Dkt. # 5194 1 at 19–20 (“[T]here are people that are closer to the market than I am that would be in a better 5 position to answer. Those would be marketing people or they would be members of the sales 6 team.”). And critically, because Impinj did not designate Oliver as an expert witness (Dkt. # 330 7 at 14), he did not prepare an expert report. So there is no documentary evidence of any kind 8 explaining how he reached his quantitative apportionment opinions. 9 To be sure, Kindler did explain that Oliver’s opinions were “consistent [] at least 10 qualitatively” with “documentary evidence” that she reviewed. Dkt. # 281-7 at 24. She also 11 explained that she had a “multistep discussion” with Oliver about his opinions, during which she 12 “tried to inquire further . . . to understand . . . how he arrived to the 90 percent figure.” Id. at 26. 13 And elsewhere in her report, Kindler explains that press releases and other documentary 14 evidence emphasized features of the accused products other than TagFocus and FastID, 15 suggesting that those features account for a limited share of the products’ total value. Dkt. 16 # 281-3 at 24–29. 17 But this does not cure Kindler’s unexplained reliance on the figures provided by Oliver. 18 See Netfuel, 2020 WL 1274985 , at *7 (“This is fatal; without explaining how he arrived at 19 ‘33%,’ the figure appears to have been ‘plucked out of thin air based on vague qualitative 20 notions of the relative importance of [the patented features.]’” (quoting LaserDynamics, 694 21 F.3d at 69)). Kindler does not explain how she came up with the 3–4% figure, the 1–2% figure, [22] 1 Oliver’s testimony in the co-pending California litigation was directed to different patents and, in some cases, different products. But his deposition testimony is generalizable: Nothing in his deposition 23 testimony suggests that he would have answered differently for the patents and accused products at issue in this case. And in fact, he was asked at this deposition about the accused FastID and TagFocus features. [24] 1 or the no-more-than-90% figure. The documentary materials relied on by Kindler may suggest 2 that NXP’s apportionment overstates the value of the accused features. But there is no evidence 3 in the report or deposition testimony that she did, in fact, rely on such evidence to calculate her 4 particular apportionment figures. Instead, the only reasonable interpretation of her report is that 5 she adopted the figures provided by Oliver without independent evaluation. And how Oliver 6 calculated those figures is anyone’s guess. See Guardant Health, Inc. v. Found. Med., Inc., No. 7 CV 17-1616-LPS-CJB, 2020 WL 2461551 , at *18 (D. Del. May 7, 2020) (requiring “an 8 evidentiary foundation for the particular percentage selected—i.e., for describing the 9 methodology used to reach that number”), report and recommendation adopted, No. CV 1710 1616-LPS-CJB, 2020 WL 5994155 (D. Del. Oct. 9, 2020); cf. Netfuel, 2020 WL 1274985 , at *9 11 (admitting expert opinion when the expert “explains the facts underlying his apportionment 12 percentages, how they connect to his apportionment percentages, and the methodology he used 13 to arrive at his apportionment percentages”). 14 To be clear, the Court does not imply that a damages expert’s reliance—even heavy 15 reliance—on a technical expert is inappropriate in apportionment analysis. Nor does the Court 16 imply that reliance on a party’s own employee is improper. See Apple Inc. v. Motorola, Inc., 757 17 F.3d 1286, 1322 (Fed. Cir. 2014) (“A rule that would exclude Apple’s damages evidence simply 18 because it relies upon information from an Apple technical expert is unreasonable and contrary 19 to Rules 702 and 703 and controlling precedent.”). To the contrary, a technical expert or a 20 party’s employee may be particularly well-positioned to understand the relative technological 21 benefits of the accused products. See, e.g., Realtime Data, 2017 WL 11574028 , at *5; Apple, 22 757 F.3d at 1322. And because there is often a connection between a feature’s technological 23 value and its economic value, economic apportionment may correlate—even strongly correlate— [24] 1 with the relative technological importance of the patented feature. See Realtime Data, 2017 WL 2 11574028, at *5. 3 But even if, as a general matter, damages experts may rely on party employees or 4 technical experts, on these facts, Kindler was not entitled to wholesale adopt the apportionment 5 figures of Oliver in the manner she did. 6 First, by Oliver’s own admission, he could “only assign relative rankings,” not “an exact 7 market value to that feature set.” Dkt. # 519-1 at 28 (emphasis added). It was unreasonable for 8 Kindler to rely nearly exclusively on quantitative estimates from an employee who admitted that 9 he could not provide quantitative estimates. The ipse dixit of an expert—itself relying on the 10 ipse dixit of a third-party—cannot alone provide a sufficiently reliable methodology. See Gen. 11 Elec. Co. v. Joiner, 522 U.S. 136, 146 (1997) (“[N]othing in either Daubert or the Federal Rules 12 of Evidence requires a district court to admit opinion evidence that is connected to existing data 13 only by the ipse dixit of the expert.”). 14 Second, if Kindler wanted to rely on Oliver’s apportionment estimates rather than 15 calculate her own, Kindler needed, at minimum, to ensure that Oliver’s methodology was 16 adequately explained, either in her report or elsewhere in the record. See NetFuel, 2020 WL 17 1274985, at *6 n.5 (“[B]ecause Mr. Bratic’s analysis depends on the validity of Dr. Rubin’s 18 analysis, in order for Mr. Bratic’s analysis to stand, Dr. Rubin’s figure must be ‘reliable.’”). For 19 example, Kindler could have explained the assumption that led Oliver (and her) to conclude that 20 FastID is less valuable than TagFocus, or how Oliver determined that True3D was worth 30– 21 35%, rather than, say, 20% or 40%. Without such explanations, the Court is left without 22 knowing whether Oliver’s estimates (and by extension, Kindler’s estimates) are 23 methodologically sound or amount to speculation. See id. at *9 (“[B]ecause Dr. Rubin failed to 24 provide the ‘methodology’ underlying his apportionment amount or explain how he arrived at 1 that figure based on the facts of this case, his apportionment opinion is not backed by ‘sufficient 2 facts or data’ or by ‘reliable principles and methods.’”). 3 Third, based on her deposition testimony, the most plausible inference is that Kindler 4 adopted Oliver’s quantitative figures without much independent analysis of her own. See Dkt. 5 # 341-1 at 18 (“Q: On what basis did Mr. Oliver form his opinion that TagFocus feature was 6 worth the whole – the FastID feature? A: . . . Ultimately, he – I don't know why. I mean, we 7 didn’t get into the details as to why, but he concluded that the relative contribution of TagFocus 8 would be approximately 3 to 4 percent based on his experience.”); id. at 23 (“Well, I said at the 9 beginning I don’t know how he came up with the numbers.”); id. 18–19 (“Q: But you don’t 10 know whether [the frequency of use of the accused features was] the reason he opined to 3 to 4 11 percent for TagFocus versus 1 to 2 percent for FastID? A: No. I’m – I’m connecting some dots, 12 but no, I don't know if that’s the reason.”). 13 The Court does not suggest that a damages expert’s apportionment opinions must be 14 perfect. Far from it: “Determining a fair and reasonable royalty is often . . . a difficult judicial 15 chore, seeming often to involve more the talents of a conjurer than those of a judge.” 16 ResQNet.com, 594 F.3d at 869 (quoting Fromson v. Western Litho Plate & Supply Co., 853 F.2d 17 1568, 1574 (Fed. Cir. 1988)); see also Aqua Shield v. Inter Pool Cover Team, 774 F.3d 766 , 771 18 (Fed. Cir. 2014) (explaining that royalty calculations often involve “approximation and 19 uncertainty”); Virnetx, Inc. v. Cisco Sys., Inc., 767 F.3d 1308, 1328 (Fed. Cir. 2014) (stating that 20 “absolute precision” is not required in apportionment, as “it is well-understood that this process 21 may involve some degree of approximation and uncertainty”); Whitserve, LLC v. Comput. 22 Packages, Inc., 694 F.3d 10, 31 (Fed. Cir. 2012) (“[M]athematical precision is not required.”). 23 Apportionment necessarily requires some approximation, and an expert should not be excluded 24 simply because she relies on proxies for relative value, provides only rough estimates, or 1 otherwise cannot perfectly quantify the relative value of features. NXP’s apportionment analysis 2 is undoubtedly imperfect, too. See Dkt. # 287-2 at 90–101 (relying, for example, on proxies to 3 determine valuation). But while the Federal Circuit permits “some approximation” in calculating 4 a reasonable royalty, this “does not negate the Federal Circuit’s requirement of ‘sound economic 5 and factual predicates’ for that analysis.” Cornell Univ., 2008 WL 2222189 , at *2 (Rader, C.J., 6 sitting by designation) (quoting Riles v. Shell Expl. & Prod. Co., 298 F.3d 1302, 1311 (Fed. Cir. 7 2002)). Nor can an expert’s calculations amount to little more than a “black box into which data 8 is fed at one end and from which an answer emerges at the other.” GPNE Corp., 2014 WL 9 1494247, at *4. 10 Here, Kindler does not explain how she landed on her 3–4% figure, the 1–2% figure, or 11 the no-more-than-90% figure besides the ipse dixit of Oliver. Oliver, in turn, provides no 12 methodology of his own for how he arrived at these figures. The Court recognizes that [13] qualitative inputs and subjective impressions can inform apportionment analysis.2 But the expert 14 must explain how those inputs lead to the calculation of the specific figures provided. Kindler’s 15 report fails to provide a reliable methodology for her quantitative apportionment conclusions, 16 and thus is subject to partial exclusion. 17 Impinj makes several arguments in response, but none is persuasive. 18 First, Impinj points to NXP expert David Haas’s reliance on NXP’s employee, Ralf 19 Kodritsch, for certain facts in his apportionment analysis. Dkt. # 330 at 16. It is not lost on the 20 Court that there are similarities between Haas’s reliance on Kodritsch and Kindler’s reliance on [22] 23 2 And indeed, Kindler reviewed documents—like third-party articles—in explaining why certain features like True3D are more valuable than others. [1] Oliver.3 After all, Kodritsch provides critical information to Haas about the relative value of 2 various features just as Oliver does. 3 But the Court believes there is an important distinction between Kindler’s reliance on 4 Oliver and Haas’s reliance on Kodritsch. Unlike Kindler, Haas relied on Kodritsch’s input about 5 the relative importance of features to customers before generating his own apportionment 6 figures. See, e.g., Dkt. # 287-2 at 99. Kodritsch does not provide the apportionment valuations 7 himself. And Haas also explains some assumptions behind Kodritsch’s conclusions (e.g., that 8 “inlay compatibility” was “not a specific product benefit,” but “simply a function of the fact that 9 all of the chips were built on the same silicon,” and that “memory write speed beyond a certain 10 point was not a feature that drove customer purchase decisions, especially among industrial 11 users”). Id. at 94 . 12 But more fundamentally, Haas explains in his report how he uses Kodritsch’s input to 13 arrive at his particular apportionment figures. Haas explains that he started with 14 features, 14 eliminated 5 that were present in the prior version of the product, discounted 3 that he believed 15 present limited economic value, and then applied some methodology (borrowing Kindler’s 25% 16 figure for read sensitivity, and dividing the remaining value equally among five other features) to 17 determine his ultimate apportionment figures. One can certainly take issue with Haas’s 18 methodology (and indeed, Kindler does)—it is debatable at best, for example, whether the 19 remaining five features warrant equal consideration, or whether True3D should not be given any 20 weight. But at the very least, one can discern (and evaluate) the methodology used, however 21 imperfect. NetFuel, 2020 WL 1274985 , at *7 (requiring a “discernable methodology”). By [22] 3 And as noted by Impinj’s counsel at the Daubert hearing, the Court recognizes that by telling 23 Haas that certain features have no value, Kodritsch was, in a sense, providing a quantitative assessment of a given feature, assigning it 0% value. [24] 1 contrast, there is no explanation about how Kindler (or Oliver) determined that TagFocus is 2 worth 3–4%, FastID is worth 1–2%, True3D is worth 30–35%, read sensitivity is worth 25–30%, 3 or memory options are worth 20%. Dkt. # 281-3 at 37. 4 Second, Impinj argues that Kindler could rely on Oliver’s opinions because under Federal 5 Rule of Evidence 703, “an expert may rely on otherwise inadmissible facts or data, such as 6 purportedly technical opinions of lay witnesses, if ‘experts in the particular field would 7 reasonably rely on those kinds of facts or data in forming an opinion on the subject’ and ‘their 8 probative value in helping the jury evaluate the opinion substantially outweighs their prejudicial 9 effect.’” Dkt. # 330 at 14 (citing Fed. R. Evid. 703). But Kindler does not rely on Oliver for the 10 “facts or data” he provides, but rather for his valuation opinions. And there is no indication that 11 an expert in the field would wholesale adopt an employee’s valuation opinions as their own 12 without questioning or even fully understanding the basis for those opinions. Kindler does not 13 incorporate Oliver’s technical conclusions as an input to her economic valuation opinions; she 14 adopts them in their entirety as her own economic apportionment conclusion. Kindler cannot 15 smuggle Oliver’s purportedly lay opinions (again, Oliver has not been identified as an expert) 16 into evidence under the guise of Rule 703 by reciting Oliver’s apportionment opinions. 17 Third, Impinj argues that Kindler could rely on Oliver even if he is a lay, rather than 18 expert, witness. Dkt. # 330 at 14. But however Oliver’s testimony is labeled, there must be 19 some discernable method—provided in Kindler’s report, in a document drafted by Oliver, or 20 elsewhere—that explains the methodology used to reach the apportionment figures. And 21 because Oliver was not designated as an expert, he produced no report in which he could provide 22 such an explanation. 23 Ultimately, assessing the relative value of the patented features in a product is one of the 24 core functions of a damages expert when calculating a reasonable royalty. This assessment 1 necessarily undergirds much of the overall reasonable-royalty analysis. Before outsourcing that 2 role to a third-party (and a non-expert, at that), the damages expert must provide some 3 explanation of the methodology used. See NetFuel, 2020 WL 1274985 , at *7 (barring expert’s 4 calculations that represent a “black box” without a “sound economic and factual predicate[]” or 5 “discernable methodology”). 6 Accordingly, the Court excludes Kindler’s quantitative apportionment opinions (and any 7 other quantitative opinions that directly rely on those apportionment opinions). But the Court 8 does not exclude Kindler’s qualitative opinions. Kindler may argue, for example, that Haas’s 9 royalty rate is too high (even far too high), that Haas’s equal apportionment of value among 10 features is improper, or that Haas otherwise failed to account for the value of features like 11 True3D. Kindler may use words that have carry some relative quantitative meaning (e.g., she 12 may opine that Haas “vastly overstates” the value of TagFocus, that the value of TagFocus to 13 customers is extraordinarily “small,” that True3D is the “primary driver” of demand, etc.). And 14 she may generally explain how Haas’s overvaluation of TagFocus leads to an inflated royalty 15 rate. The Court precludes Kindler only from offering a precise, numerical apportionment 16 percentage for the TagFocus feature (e.g., 3–4% or 3–5%). 17 C. Kindler’s Purported Lack of “Starting Point” for Reasonable Royalty Analysis 18 NXP says that Kindler applies each of the Georgia-Pacific factors to adjust upward or 19 downward her reasonable-royalty calculations. See Georgia-Pacific v. U.S. Plywood Corp., 318 20 F. Supp. 1116 , 1120 (S.D.N.Y. 1970); Dkt # 279 at 16–17. But, NXP says, her analysis lacks a 21 reliable starting point from which she could make those adjustments. See, e.g., Dkt. # 279 at 17 22 (“It is unhelpful to the jury to opine that certain facts would exert an upward pressure on the rate 23 without specifying upward from what.”). [24] 1 The Court disagrees in part. Throughout her report, Kindler often relies on the royalty 2 rates presented by NXP’s expert, Haas, as a starting point. See, e.g., Dkt. # 287-11 at 34. She 3 then adjusts Mr. Haas’s rates using the Georgia-Pacific factors. Id. (downwardly adjusting 4 Haas’s estimate of the value of the accused features). This makes sense: Her report is a rebuttal 5 report. The Court finds nothing improper about using the royalty rates provided by the opposing 6 party as a starting point of analysis. Cf. Dkt. # 287-2 at 95 (Haas borrowing Kindler’s estimates 7 as a proxy). 8 But as stated above, Kindler may not provide quantitative apportionment opinions. 9 Accordingly, when discussing any Georgia-Pacific factors relating to apportionment, Kindler 10 may not “adjust” Haas’s apportionment calculations with her own quantitative calculations (or 11 rather, Oliver’s). She may still qualitatively explain why she disagrees with Haas’s assessment 12 of any given Georgia-Pacific factor. 13 III 14 NXP’S MOTION IN LIMINE 15 NXP moves to preclude any evidence or argument from “Mr. Oliver or Ms. Kindler” [16] “regarding Mr. Oliver’s opinions on issues of valuation and apportionment.”4 Dkt. # 458 at 5–8. 17 The Court agrees in part and disagrees in part. For reasons similar to those outlined 18 above, Oliver may not present his quantitative valuation opinions. As a lay witness, Oliver may 19 not offer opinions that are based on “scientific, technical, or other specialized knowledge within 20 the scope of Rule 702.” Fed. R. Evid. 701. Oliver’s precise quantification of the relative value 21 of various features developed specifically for purposes of litigation is at least arguably within the 22 realm of expert testimony under Rule 702, at least when such opinions were not developed in the 23 4 The Court reserved ruling on this motion in limine pending the Daubert hearing. Dkt. # 516 at 2. [24] 1 course of the employee’s ordinary job duties. Even if not expert testimony, as described above, 2 || Oliver’s statements indicate that he cannot reliably quantify the relative values of various 3 features. So regardless of how his testimony is characterized, he is not qualified to offer precise, 4 || quantitative apportionment figures. 5 But Oliver may testify about facts that he knows. He can testify (consistent with other 6 || tules of evidence) that based on his role at Impinj, certain features are more important than 7 ||others. For example, he can testify about his observation that in “internal marketing g || presentations,” TagFocus is “further down in that list [of features].” Dkt. # 519-1 at 28. He can g testify as to any facts or opinions developed as a result of his role at Impinj (e.g., based on 10 conversations with customers). 11 So the Court grants the motion in limine in part and denies it in part. 12 IV 13 CONCLUSION 14 For the reasons stated, the Court: 15 (3) GRANTS NXP’s motion to exclude certain opinions of Lauren R. Kindler in part, 16 DENIES it in part, and STRIKES it in part as moot. Dkt. # 279. 7 (4) GRANTS in part and DENIES in part NXP’s second motion in limine. Dkt. # 458. [18] 19 Dated this 8th day of June, 2023. Soka 4. Chur 21 ohn. Chin United States District Judge [24] ORDER RE: MOTION TO EXCLUDE CERTAIN OPINIONS OF LAUREN R. KINDLER AND NXP’S MOTION IN