With respect to the Shugan conjoint analysis relied upon by Cockburn Google says this:
[Ed. Note: Here, courtesy of our friend Steve, is document 751 in text]
KEKER & VAN NEST LLP
ROBERT A. VAN NEST, #84065
CHRISTA M. ANDERSON, #184325
DANIEL PURCELL, #191424
KING & SPALDING LLP
DONALD F. ZIMMER, JR. — #112279
CHERYL A. SABNIS — #224323
KING & SPALDING LLP
SCOTT T. WEINGAERTNER
(Pro Hac Vice)
ROBERT F. PERRY
BRUCE W. BABER (Pro Hac Vice)
IAN C. BALLON — #141819
HEATHER MEEKER — #172148
GREENBERG TRAURIG, LLP
Attorneys for Defendant GOOGLE INC.
NORTHERN DISTRICT OF CALIFORNIA
ORACLE AMERICA, INC.,
SAN FRANCISCO DIVISION
Case No. 3 :10—cv—03 561 -WHA
GOOGLE'S REPLY IN SUPPORT OF
MOTION TO STRIKE PORTIONS OF
THIRD EXPERT REPORT BY IAIN
COCKBURN AND EXPERT REPORT
BY STEVEN SHUGAN
Dept.: Courtroom 8, 19th Floor
Judge: Hon. William Alsup
TABLE OF CONTENTS
||Dr. Cockburn's "independent significance" approach is
indeterminate and based entirely on Dr. Cockburn's subjective
||Dr. Cockburn's group and value approach is methodologically
flawed and rests on an assumption untethered to the facts of the
||Dr. Cockburn failed to apportion the value of the copyrights at
||Dr. Cockburn failed to conduct a claim-by-claim valuation of the
||Oracle's attempt to use Dr. Shugan's conjoint analysis to prove a
damages amount is unprecedented and methodologically flawed
||Oracle has failed to cite any case where any Court has
permitted a conjoint survey to be used to prove damages.
||Dr. Shugan's conjoint analysis is methodologically flawed
and produced illogical results
||Dr. Cockburn's econometric analysis uses unrepresentative
market data and rests on implausible assumptions
TABLE OF AUTHORITIES
|Allison v. McGhan Med Corp.
184 F.3d 1300 (11th Cir. 1999)
|American Gen. Life Ins. Co. v. Schoenthal Family, LLC
555 F.3d 1331 (11th Cir. 2009)
|Cayuga Indian Nation of New York v. Pataki
83 F. Supp. 2d 318 (N.D.N.Y. 2000)
|Finjan, Inc. v. Secure Computing Corp.
626 F.3d 1197 (Fed. Cir. 2010)
|Gen. Elec. Co. v. Joiner
522 U.S. 136 (1997)
|Georgia Pacific Corp. v. U.S. Plywood Corp.
318 F. Supp. 1116 (S.D.N.Y. 1970)
|Hein v. Merck & Co., Inc.
868 F. Supp. 230 (M.D. Tenn. 1994)
|Kumho Tire Co., Ltd v. Carmichael
526 U.S. 137 (1999)
|4, 5, 8|
|LG Display Co. Ltd v. AU Optronics Corp.
722 F. Supp. 2d 466 (D. Del. 2010)
|Lucent Techs., Inc. v. Gateway, Inc.
580 F.3d 1301 (Fed. Cir. 2009)
|Lust By & Through Lust v. Merrell Dow Pharms., Inc.
89 F.3d 594 (9th Cir. 1996)
|Macaluso v. Herman Miller, Inc.
01 CIV. 11496 (JGK), 2005 WL 563169 (S.D.N.Y. Mar. 10, 2005)
|Ruiz-Troche v. Pepsi Cola of Puerto Rico Bottling Co.
161 F.3d 77 (1st Cir. 1998)
Despite the Court's patience in allowing Oracle yet another attempt to put forth a viable damages opinion, Dr. Iain Cockburn's third report is not the charm, as a simple reading of the report itself and the concessions in Oracle's opposition brief confirm.
First, Oracle concedes that there is a hole in Dr. Cockburn's "independent significance" analysis big enough to drive a truck through. Although Dr. Cockburn's report uses that approach to calculate the value of the patents- and copyrights-in-suit at 25% and 12.5%, respectively, of the intellectual property bundle at issue in the 2006 Sun-Google negotiations, Dr. Cockburn admitted at deposition that he actually believes, and intends to tell the jury if asked, that the patents and copyrights in suit actually might account for the entire $685 million value of the bundle. That alone renders this approach indeterminate and inadmissible. But the approach has other fatal flaws. Dr. Cockburn admitted that he evaluated the totality of the evidence in the case and then, without applying any identifiable analytical methodology, picked a number that fit his view of those facts. No case Google has located, and certainly not the Finjan case Oracle mistakenly relies on, has ever endorsed such a subjective, idiosyncratic analysis.
Second, Dr. Cockburn's alternative "group and value" approach purports to calculate the value of the patents-in-suit as being somewhere in the broad range between 10.2% and 32.7% of the total 2006 IP bundle. Such a wide range would give the jury little guidance. But that range also depends entirely on Dr. Cockburn's assumption that the distribution of value within the Sun patent portfolio is the same as the distribution identified in three patent-value studies. Unfortunately, all of those studies are inapposite, looking at the broadest possible context—a random sampling of all patents, owned by all patentees, related to all technology areas. None of the studies looked at a single party's narrow patent portfolio covering only one technology area, like the Sun portfolio at issue here. Essentially, Dr. Cockburn was given a fruit basket and asked to estimate the percentage of the basket's value attributable to the apples in the basket. For some reason, he attempted to answer that question by looking at the distribution of value among all groceries in every department in the supermarket. Dr. Cockburn offers no basis for justifying his failure to use a benchmark that fits the narrow category of intellectual property at issue here.
Third, Dr. Cockburn still does not have any meaningful sense of what copyrights were on the table in the 2006 Sun—Google negotiations. Out of millions of lines of source code and other valuable works, he assigns a specific value to only the 37 API packages at issue. For the rest of the copyrights, he erroneously uses Sun's projected engineering costs to develop additional code in the future as a proxy for the value of existing code. Even though he concedes Google spent tens of millions of dollars independently developing code it could have acquired from Sun, he refuses to give Google credit for those costs, which Google would have avoided in a partnership.
Fourth, Oracle concedes that Dr. Cockburn valued only the patents-in-suit as whole units and did not subtract the value of any claims of those patents that Oracle decided not to assert or asserted but abandoned. Again, this is contrary to the Court's instruction that Oracle value patent claims, not patents as a whole—and again, Oracle's error inflates its purported damages.
Fifth, Oracle cannot identify a single case (and Google is unaware of any) where a court has approved the use of a conjoint analysis to calculate damages in litigation. Not only is Dr. Cockburn's use of Dr. Steven Shugan's conjoint analysis unprecedented, Dr. Shugan concedes that his methodology is so unreliable that a full quarter of the participants in his study purported to prefer a smartphone costing $200 to an identical phone costing $100, or at least be ambivalent between the two, a result which makes no sense in the world of real consumers.
Sixth, Dr. Cockburn's econometric study is unreliable for several reasons—most notably, because it assumes that an Android phone that didn't use Sun's patents (and purportedly would perform much worse) would sell for the same price as a real Android phone that performs well. This assumption makes no sense either—and again, it significantly inflates Oracle's damages.
All these aspects of Dr. Cockburn's analysis should be excluded at trial.
A. Dr. Cockburn's "independent significance" approach is indeterminate and based
entirely on Dr. Cockburn's subjective judgment.
Oracle's Opposition does nothing to fill the largest hole in the "independent significance" approach: its essential indeterminacy. Dr. Cockburn stated in his report that [REDACTED]
[REDACTED] Cockburn Report ¶¶ 423, 671. [REDACTED] Oracle justifies Dr. Cockburn's plan to inflate his report's apportionment figures at trial by arguing that Dr. Cockburn's report stated only [REDACTED] Oracle Opp. at 7. But Dr. Cockburn should not be permitted to state only a [REDACTED]
The indeterminacy of Dr. Cockburn's independent significance analysis is not a bug; it is an intended feature, designed to give Oracle the flexibility to ask the jury for hundreds of millions of dollars more than the figures in Dr. Cockburn's report. Indeed, even the basis of the independent-significance analysis is unclear, with Dr. Cockburn having offered wildly varying descriptions of that analysis at different times. Oracle insists that Dr. Cockburn's conclusions are "based on objective evidence, not a 'subjective judgment.'" Oracle Opp. at 6 (emphasis in original). But Oracle and Dr. Cockburn have not even managed to be consistent about what evidence Dr. Cockburn considered. In his report, [REDACTED] Oracle similarly argued in its Opposition that Dr. Cockburn's independent significance analysis relies on the conjoint and econometric analyses. Oracle Opp. at 6 n.1. This inconsistency is symptomatic of the approach's inherent subjectivity and total lack of rigor. [REDACTED]
Oracle makes two unsuccessful attempts to justify Dr. Cockburn's stab in the dark. First, Oracle relies on one paragraph in Finjan, Inc. v. Secure Computing Corp., 626 F.3d 1197 (Fed. Cir. 2010). This reliance is misplaced. In Finjan, the defendant challenged a royalty award by arguing about the sufficiency of evidence under many of the Georgia-Pacific factors. The language Oracle quotes is from the court's brief discussion of Georgia-Pacific factor 13, where it ruled that the expert's opinion that the patented inventions were "fundamentally important to the product" was probative under factor 13. Id. at 1211. Oracle argues that Dr. Parr, the expert in Finjan, "offered no algorithm or formula for his apportionment of profits and no quantitative evidence." Oracle Opp. at 5. But Dr. Parr's method did not involve apportioning profits. Dr. Parr simply opined that the patented inventions were fundamentally important to the product for purposes of Georgia-Pacific factor 13, one of 15 non-exclusive factors bearing on a royalty rate calculation. The heart of Dr. Parr's analysis did not rely on "fundamental importance"; instead, he offered a detailed calculation of the defendant's operating margin for its patented product based on the defendant's financial data. Id. at 1209. Nothing in Finjan authorizes Dr. Cockburn to [REDACTED] Mullen Decl. Ex. A (Cockburn Dep.) at 134:2-3, 138:9-10.
Second, Oracle appears to argue that the independent significance approach is not subject to standards of scientific reliability at all, but should be judged solely based on Dr. Cockburn's experience. Oracle Opp. at 7. This is legally incorrect, and none of the three cases Oracle cites support it. In Lucent Techs., Inc. v. Gateway, Inc., 580 F.3d 1301 (Fed. Cir. 2009), the Federal Circuit reversed a damages award as not based on substantial evidence. Id. at 1324. In doing so the court noted the uncontroversial proposition that the reasonable royalty calculation is subject to "an element of approximation and uncertainty." Id. at 1325. But tolerating "an element" of uncertainty is no justification for allowing an expert to testify to an almost infinite apportionment range based on an ever—shifting list of evidence.
In Kumho Tire Co., Ltd v. Carmichael, 526 U.S. 137 (1999), the Supreme Court affirmed
the district court's decision to strike an expert's testimony because it was "connected to existing data only by the ipse dixit of the expert." Id. at 157. The Court noted briefly that this rule might not apply to cases where "the relevant reliability concerns may focus upon personal knowledge and experience." Id. at 150. Contrary to Oracle's suggestion, Oracle Opp. at 7, nothing in Kumho Tire even remotely suggests that the "valuation of property" is an example of expert testimony depending solely on personal knowledge or experience, much less the valuation of the highly technical intellectual property at issue here, which depends significantly on assumptions provided by technical experts. In fact, courts have applied Daubert's scientific reliability requirement to expert valuations of property. In Cayuga Indian Nation of New York v. Pataki, 83 F. Supp. 2d 318, 323-26 (N.D.N.Y. 2000), the court struck the testimony of an expert appraiser's valuation of property because his methodology and data were suspect—even though the expert was well—qualified on paper.
The third case Oracle cites, American Gen. Life Ins. Co. v. Schoenthal Family, LLC, 555 F.3d 1331 (11th Cir. 2009), considered the type of personal experience-based expert testimony mentioned in Kumho Tire and is nothing like this case. In Schoenthal, the court allowed an expert to testify about general standards in the insurance industry based on his experience working in that industry. Id. at 1338. This type of expertise could only come from personal knowledge and experience as there is no scientific methodology to capture insurance industry standards. No case that Oracle cites, and none of which Google is aware, authorizes an expert to value patents based solely on a resume, the totality of evidence in the case, and bare conclusions.
This case is too complex and the stakes too high for Oracle and Dr. Cockburn to fall back on generic assertions of "experience." The independent significance approach is indeterminate and gives the jury no meaningful guidance. The Court should strike it altogether. At the very least, the Court should bar Dr. Cockburn from presenting apportionment percentages greater than the 25% patent and 12.5% copyright percentages purportedly justified by his report.
B. Dr. Cockburn's group and value approach is methodologically flawed and rests on
an assumption untethered to the facts of the case.
Dr. Cockburn's "group and value" approach rests entirely upon his guess that the patent-
value distribution in Oracle's self-selected and self—ranked group of 569 "Java-related" patents is the same as all patent-value distributions everywhere, regardless of patent owner, patent type, or portfolio size. Dr. Cockburn's guess is not reliable science. The group and value approach should be stricken because Dr. Cockburn offers no reliable way to apportion the value of the five patents-in-suit amongst all of the other IP at issue in the hypothetical license negotiation.
[REDACTED] Cockburn Rep. ¶ 408. The sole basis for these figures is three surveys showing similar patent-value distributions among other groups of patents. Id. at ¶¶ 405, 408, 412, Ex. 34. But Dr. Cockburn offers no analysis whatsoever explaining why those surveys are a good fit for the Sun patent portfolio here, such that the distribution of value among Sun's patents could be assumed to be similar to the distributions observed in the surveys. Contrary to Oracle's attempt in its opposition to shift the burden to Google of disproving Dr. Cockburn's logic, it is Dr. Cockburn's obligation to justify each aspect of his analysis. Lust v. Merrell Dow Pharms., Inc., 89 F.3d 594, 598 (9th Cir. 1996) ("It is the proponent of the expert who has the burden of proving admissibility."). In any event, Google explained in detail in its Motion to Strike why the Sun portfolio is nothing like the patents analyzed in those surveys: namely, (1) Sun's portfolio was comprised of United States-issued patents, (2) owned by a single company, (3) in a narrow technology area, and (4) specifically selected as relevant to a smartphone like Android. Dkt. No. 718 at 9-10. Oracle now says the differences between Sun's portfolio and the survey patents are immaterial. Dkt. No. 737 at 11-12. Oracle is wrong for at least three reasons.
First, as a preliminary matter, Oracle mischaracterizes Google's challenges to the work done by the five Oracle engineers. Google deposed each of the five engineers upon whom Dr. Cockburn relied to categorize and rank the 569 patents.1
Four of those five engineers had "assisted lawyers in analyses to determine whether Android infringes Oracle's patents." Kessler
Decl. ¶ 25; Plummer Decl. ¶ 25; Rose Decl. ¶ 25; Wong Decl. ¶ 25. Google explained in its Motion that the engineers' familiarity with the patents-in-suit naturally would have colored their evaluation of those patents (and related patents) relative to the lesser known (or unknown) patents in the review set. Dkt. 718 at 7-8.2
Oracle responds that allegations of bias are points for cross-examination and are "not cognizable" in a Daubert motion. Oracle Opp. at 10.
Google agrees that the engineers' prior work in preparing this lawsuit is excellent fodder for cross-examination, but Oracle is wrong to suggest this evidence is irrelevant to Google's Daubert motion. To the contrary, the process by which the engineers selected and ranked patents for Dr. Cockburn goes to show how different the Sun patent portfolio analyzed by Dr. Cockburn is from the random groups of patents analyzed in the three patent-value surveys, and thus why those surveys are not a reliable baseline for the distribution of value within Sun's portfolio. Whereas the patents in the surveys cited by Dr. Cockburn were randomly selected from a pool of every patent under the sun (and thus certain to run the gamut from completely worthless to highly valuable), the 569 patents here were specifically selected for their purported relevance to a mobile smartphone platform. Oracle used a term search to narrow the patents at issue from Sun's entire 2006 patent portfolio (presumably consisting of thousands of patents) to 1300 patents. Then Oracle pared that list down to 569 patents by having one of the engineers look at each patent individually. Oracle Opp. at 8-9. Oracle's position on the patent-value surveys would be somewhat more plausible if Oracle were arguing the surveys could predict the distribution of value among all Oracle patents, but in creating its pool of 569 patents, Oracle put its thumb on the scale by weeding out all patents supposedly having little or no value to Google.
Second, Dr. Cockburn makes no effort to tie his assumption on patent-value distribution to the facts of this case. He merely cites to surveys showing a "skewed" value distribution and
then maps the upper and lower bound of that distribution onto the Sun portfolio. That is not expert analysis; it's tracing. It may be true that some patents in Sun's portfolio are much more valuable than some others, but the existence of "skew" is not a proxy for a damages figure. Dr. Cockburn's own sources show that the amount of skew can significantly vary from one group to another. For example, between his revised report and the new declarations filed with Oracle's Opposition, Dr. Cockburn notes the following possible skew distributions:
- 10% of the patents hold 48%-93% of the value. (Cockburn Decl. ¶ 6.)
- 10% of the patents hold 95% of the value.(Cockburn Decl. ¶ 6.)
- 10% of the patents hold 88% of the value. (Cockburn Decl. ¶ 6.)
- 5% of the patents account for 60% of the value. (Norton Decl. Ex. A.)
- 3.9% of the patents hold 67.9% of the value. (Cockburn Rep. Ex. 34.)
- 3.9% of the patents hold 77.1% of the value. (Cockburn Rep. Ex. 34.)
- 3.9% of the patents hold 91.9% of the value. (Cockburn Rep. Ex. 34.)
- 1%-2% of the patents hold "more than 50%" of the value. (Cockburn Decl. ¶ 9.)
- 1% of the patents hold 42%-78% of the value. (Cockburn Rep. ¶ 406.)
Dr. Cockburn made no attempt to analyze where the distribution in the Sun portfolio might fall on this spectrum of options. He simply observed a high point, a low point, and concluded that the actual distribution must be somewhere in between. Cockburn Rep. at ¶ 408 [REDACTED] Daubert requires more. See, e.g., Kumho Tire, 526 U.S. at 151-52 (noting that Daubert's gatekeeping requirement is intended to "make certain an expert, whether basing testimony upon professional studies or personal experience, employs in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field").
Third, Oracle incorrectly argues that LG Display Co. Ltd v. AU Optronics Corp., 722 F. Supp. 2d 466 (D. Del. 2010), supports the methodology underlying Dr. Cockburn's group and value approach. Oracle Opp. at 11-12. The sole support Oracle provides for this position is a declaration from LG's expert in that case—Jonathan D. Putnam—that summarizes a "Lorenz graph" that purportedly shows that "5% of the patents must account for 60% of the portfolio's
value." Id.; Norton Decl. [Dkt. No. 738], Ex. A. Oracle relies on a footnote to the Putnam declaration, which states only: "As I explained in the Appendix to my supplemental report, the distribution most closely related to the patents-in-suit is for 'the electronics industry (excluding Japan),' reported in Table 5 of M. Shankerman, "How valuable is patent protection," RAND Journal of Economics, 29(1), pp. 77-107 (1998)." Norton Decl. [Dkt. No. 738], Ex. A at 7 n.4. Oracle's filings do not attach the "Lorenz graph," the Putnam supplemental report, the Putnam appendices, or the article upon which Putnam apparently relied. Id. Moreover, as is apparent from that pithy footnote, Mr. Putnam, unlike Dr. Cockburn, looked at a patent-value distribution model that marginally related to the same industry—"electronics"—as the infringing products. Id. In any event, the district court opinion does not address Mr. Putnam's opinions about patent distribution in any meaningful way. See LG, 722 F. Supp. 2d at 472-73. At best, the LG case shows that an expert in an unrelated case, assessing an unknown patent portfolio, concluded that some patents in that portfolio were more valuable than other patents, and the court allowed the expert to testify as to that conclusion. LG does not justify what Dr. Cockburn has done here.
For all of these reasons, Dr. Cockburn's group and value approach should be stricken.
C. Dr. Cockburn failed to apportion the value of the copyrights at issue.
Oracle tries to minimize, but cannot ignore, the central flaw in Dr. Cockburn's copyright apportionment: his fundamental ignorance of the universe of copyrights at issue. Dr. Cockburn Stated in his report that [REDACTED] Cockburn Rep. ¶ 346. But the proposed Sun-Google partnership would have given Google access to far more copyrighted works, including the source code underlying the Java Virtual Machine and over 100 additional core Java libraries which, after all, provide all of the functionality for which the APIs serve only as an index. In his deposition, [REDACTED] Mullen Decl. Ex. A (Cockburn Dep.) at 151:22-152:4. Oracle attempts to justify Dr. Cockburn's failure to understand the extent of Sun's Java-related copyrights by arguing that Dr. Cockburn "certainly does know, and analyzes, what is relevant: what copyrighted material . . . would have been included in the 2006
Bundle and provided value to Google." Oracle Opp. at 16 (emphasis in original). Even this is not true. In his deposition [REDACTED] Mullen Decl. Ex. A (Cockburn Dep.) at 153:21-22. [REDACTED] Id. at 154:7-16. He then admitted that [REDACTED] Id. at 155:21-156:2. Dr. Cockburn's ignorance of the other 100+ APIs and millions of lines of code that would have been licensed to Google makes clear he made no serious effort to understand what copyrights were at issue. He certainly did not do a work-by-work or group-by-group analysis of the copyrights, as he purported to do on the patent side.
Not only did Dr. Cockburn fail to identify the universe of copyrights at issue, he purports to account for the value of the non-API copyrights in a way that makes no sense and fails to give Google credit for engineering costs it would have avoided had it partnered with Sun. Dr. Cockburn opines that the value to Google of all copyrights other than the API specifications (whatever those might be) would be subsumed in Sun's projected future engineering costs. But Sun's cost projections relate to the creation of new and different intellectual property, and were properly deducted separately from the value of the bundle because it would be economically wrong to give Sun the benefit of lost revenues without deducting the associated costs. Cockburn Rep. ¶¶ 48, 370. The offset for Sun's future costs does not include the value to Google of the Sun code already in existence—e.g., the API implementations, JVM, documentation, and other source code. The potential Sun-Google partnership would have involved a division of labor, with Sun's existing code potentially giving Google a head start on Google's share of the work. [REDACTED] Oracle Opp.at 15; Cockburn Rep. ¶¶ 366-67, 372-82 & EX. 30. But [REDACTED]
Instead, he took another shortcut that inflates his damages calculation, bundling the value of the old code to Google into the projected future cost to Sun of writing new and different code.
D. Dr. Cockburn failed to conduct a claim-by-claim valuation of the patents-in-suit.
Oracle still stubbornly refuses to acknowledge what this Court has repeatedly held: Dr. Cockburn must identify the value of the patents in suit attributable to each of the individual asserted claims. Dr. Cockburn simply did not do this. As he admitted in his deposition, [REDACTED] Mullen Decl. Ex. A (Cockburn Dep.) at 90:5-10. Dr. Cockburn cannot know whether the entire value of a patent rests in the asserted claims without making any effort to value the unasserted claims.
Oracle justifies Dr. Cockburn's failure by arguing that "[p]arties to an actual licensing negotiation would not have evaluated every claim . . . and there is no reason to do so now." Oracle Opp. at 18. Once again, Oracle is fighting the hypothetical that the Court directed it to use, this time in its January 9, 2012 Order. There, the Court made clear that "[a]n infringer of one claim is compelled by law to pay for a license, via the hypothetical negotiation, for the specific invention represented by that claim but it is not required to pay for a license for the other specific inventions not infringed. Therefore, the hypothetical negotiation must be focused only on negotiating a compulsory license for each claim infringed, not for the entire patent." Jan. 9, 2012 Order [Dkt. No. 685] at 9. By failing to break out any value that might have been located in the unasserted claims, Dr. Cockburn failed to value only the asserted claims.
Oracle also argues that Dr. Cockburn did value the unasserted claims because "it is Prof. Cockburn's opinion that the claims would have no additional value to Google, above and beyond that of the asserted claims." Oracle Opp. at 18 (emphasis in original). Tellingly, Oracle provides no citation for this supposed opinion. It certainly appears nowhere in Dr. Cockburn's report. See Cockburn Rep. ¶¶ 493-533. In his report, [REDACTED] Now, to justify his omission, Oracle attributes to Dr. Cockburn an opinion he never expressed, that the unasserted claims add no additional value. This is not a true claim-by-claim analysis, and creates
a serious risk of forcing Google to pay a license fee for patent claims it did not infringe.
E. Oracle's attempt to use Dr. Shugan's conjoint analysis to prove a damages amount
is unprecedented and methodologically flawed.
1. Oracle has failed to cite any case where any Court has permitted a conjoint
Oracle fails to identify even a single case where a court in any jurisdiction allowed the use of conjoint analysis to prove damages. Oracle says "it is settled law that survey—based studies are admissible for just such a purpose." Oracle Opp. at 19. But the "settled law" Oracle cites has nothing to do with conjoint surveys. Oracle Opp. at 19 (citing Lucent Techs., Inc. v. Gateway, Inc., 580 F.3d 1301, 1333-34 (Fed. Cir. 2009)). In Lucent, the Federal Circuit merely noted that, "depending on the case," a plaintiff might use a survey to show that an infringing invention is frequently used by consumers, since that tends to prove the invention is valuable. Id. at 1333-34; accord Georgia-Pacific Corp. v. US. Plywood Corp., 318 F. Supp. 1116, 1120 (S.D.N.Y. 1970) (Factor 11 is "The extent to which the infringer has made use of the invention; and any evidence probative of the value of that use."). A simple survey designed to test the frequency with which consumers use an invention may well be reliable and helpful to the jury. But that would be very different from a conjoint survey that purports to isolate the relative importance of a single feature (or set of features) in a multi-feature product and then predict the market share attributable to that feature. Oracle cites no cases allowing the use of conjoint surveys to prove damages because, at least as far as Google is aware, there is no such case.
survey to be used to prove damages.
Oracle attempts to compensate for this lack of authority by arguing that "[m]any experts have concluded that conjoint analysis is a proper way to calculate intellectual property damages." Oracle Opp. at 19. As support for this position, Oracle cites to a multiple-hearsay declaration from Dr. Shugan, who in turn cites to other academics' vague descriptions of "litigation matters" where conjoint surveys purportedly were "use[d]." Shugan Decl. [Dkt. No. 740] at ¶¶ 14-15. Dr. Shugan offers no case citations for these "litigation matters," much less orders or transcripts admitting conjoint surveys into evidence, or, if so, for what purpose. Id. None of this is evidence, and Oracle's reliance on it only reveals the weakness of its argument. There is a good reason why conjoint analysis is not used to prove damages: it is a marketing research tool, not a
damages methodology of sufficient rigor to be used in complex intellectual property litigation.
2. Dr. Shugan's conjoint analysis is methodologically flawed and produced
Contrary to Oracle's repeated insistence, Google has never argued that the many serious methodological flaws in Dr. Shugan's conjoint analysis are for the jury. In its motion to strike Dr. Leonard's report, Oracle unsuccessfully argued that Dr. Leonard lacked the knowledge and expertise to critique Dr. Shugan's work. Dkt. No. 558 at 13-14. Because Google had not yet filed a Daubert motion on Dr. Shugan, Google's response was limited to explaining Dr. Leonard's critiques of Dr. Shugan and explaining why Dr. Leonard was qualified to dispute Dr. Shugan's analysis before the jury. Dkt. No. 581 at 14-17. Google never conceded that any aspect of Dr. Shugan's analysis was sufficiently methodologically sound to go to a jury in the first place. The reason Google asked the Court permission to file this motion is because it is not.
The key methodological flaw with Dr. Shugan's conjoint analysis, which neither Oracle nor Dr. Shugan (in his most recent declaration) disputes, is that survey participants obviously failed to follow the survey's critical instruction to hold constant all non-specified features of the smartphones being evaluated. Oracle and Dr. Shugan argue only that, even though respondents obviously implied additional, non-specified features into the smartphones tested in the survey, each respondent was probably internally consistent in making similar adjustments for similar classes of smartphones. Oracle Opp. at 22 (quoting Shugan Decl. ¶ 33) ("[t]here is no reason to believe that the respondents who do enrich the value of the price or brand with variables not included in the conjoint study vary their evaluation of price or brand between the 16 choice sets from which they choose their preferred smartphones."). This is just after-the-fact guesswork. There is nothing in Dr. Shugan's report addressing the issue whether individual respondents were consistent in implying the same additional features and making similar price adjustments for every tested class of smartphone. Indeed, there is no reason not to believe that respondents varied those feature and price adjustments across the choice sets—for example, by assuming that a higher-priced iPhone had additional features not found in a lower-priced iPhone, while not making that assumption for an Android phone. No one, certainly not Dr. Shugan, knows for
sure. The only thing that is clear (and conceded by Oracle) is that respondents did not follow the survey's key instruction to hold all non—specified features constant. With that concession, there is no way to determine what value the respondents placed on the specified features. Since answering that question was entire point of the conjoint study, it has no probative value at all.
The clearest evidence that Dr. Shugan's methodology is faulty is his nonsensical results, which Oracle desperately but unsuccessfully seeks to disguise. Citing Dr. Shugan's reply report and declaration, Oracle insists that a "proper reading [of Dr. Shugan's results] indicates that 8.8%, not 24%, of respondents were insensitive to the $100 price increase." Oracle Opp. at 22. This is a mischaracterization of what Dr. Shugan said. In both his reply report and declaration, Dr. Shugan says that while 24% of people preferred a price of $200 over a price of $100, only 8.8 percent of them had a preference strong enough that a "diligent statistician" would conclude that they would prefer a $200 phone over a $100 phone rather than be indifferent between the two. Shugan Decl. 11 39; Shugan Reply Rep. at 19. Thus even if Dr. Shugan is correct, his analysis still predicts that 24% of the population would be indifferent between paying $100 and $200 for an identical phone. That result is no less absurd and no more a justification for use of the conjoint study in this case. Oracle cryptically tries to justify this failure of common sense by arguing that "the ability of a Bayesian model to predict aggregate consumer behavior is not tested by focusing on individual outlier cases." Oracle Opp. at 22. But we are not talking about a handful of easily dismissed "individual outlier cases" here; we are talking about a full quarter of the survey population giving responses no real-world consumer ever would give.3
F. Dr. Cockburn's econometric analysis uses unrepresentative market data and rests
on implausible assumptions.
Oracle defends Dr. Cockburn's use of the eBay data solely on the basis that extrapolating
from sales of used phones to new phones is acceptable. But neither Oracle nor Dr. Cockburn have justified using eBay data specifically in the cell phone market, where most phone purchases take place as part of a package that includes phone service and a data plan. There is no reason to think that consumers' decisions in purchasing cell phones as part of such a package mirror consumers' decision in purchasing unlocked cell phones on eBay independent of any carrier.
Oracle also argues that the unrealistic assumptions underlying Dr. Cockburn's conversion of his econometric analysis to market share should go to the jury. But contrary to Oracle's argument, courts often strike expert testimony that is based on unreasonable assumptions, see, e.g., Macaluso v. Herman Miller, Inc., 01 CIV. 11496 (JGK), 2005 WL 563169 (S.D.N.Y. Mar. 10, 2005), including in the damages context, see, e.g., Hein v. Merck & Co., Inc., 868 F. Supp. 230, 233-35 (M.D. Tenn. 1994).
Oracle fails to justify Dr. Cockburn's assumption that every bidder's bid on eBay for a phone would decrease while the winning bids for the same phone, which would remain the same. Oracle attempts to defend this assumption by arguing that Google would have no control over the prices set by OEMs and carriers. This just emphasizes the disconnect between the prices at which phones sell on eBay versus through carriers and shows why eBay is a different market from carriers. Further, contrary to Oracle's assertion, Google is not assuming that the price of Android phones would go down. It is Dr. Cockburn that makes that assumption, but only when helpful to inflating damages. Dr. Cockburn's model adjusts eBay bids based on his econometric results. But within the context of eBay auctions it is incoherent to assume that every bid will go down but that the winning bid, which is determined by the second-highest bid, remains constant.
For all of these reasons, Google respectfully requests that the Court strike the specified portions of Dr. Cockburn's Third Expert Report and Dr. Shugan's Expert Report.
Dated: February 28, 2012
KEKER & VAN NEST LLP
/s/ Robert A. Van Nest
By: ROBERT A. VAN NEST
Attorneys for Defendant GOOGLE INC.
1 After Google deposed the engineers, Oracle filed seven-page declarations from each engineer. Dkt. Nos. 741-745. The declarations are nearly identical to one another in describing the work each engineer did to assist Dr. Cockburn.
3 Oracle insists that Daubert itself requires the Court to turn a blind eye to the survey's irrational results. Oracle Opp. at 23. Oracle is wrong. In applying Daubert, the Supreme Court has clarified that "conclusions and methodology are not entirely distinct from one another." Gen. Elec. Co. v. Joiner, 522 U.S. 136, 146 (1997). Indeed, courts regularly test experts' methods by the reasonableness of their conclusions. See, e.g., Allison v. McGhan Med. Corp., 184 F.3d 1300, 1315 (11th Cir. 1999) ("Joiner made it clear that although principles and methodology were the focus [of Daubert], the court was not precluded from looking at conclusions."); Ruiz-Troche v. Pepsi Cola of Puerto Rico Bottling Co., 161 F.3d 77, 81 (1st Cir. 1998).