decoration decoration
Stories

GROKLAW
When you want to know more...
decoration
For layout only
Home
Archives
Site Map
Search
About Groklaw
Awards
Legal Research
Timelines
ApplevSamsung
ApplevSamsung p.2
ArchiveExplorer
Autozone
Bilski
Cases
Cast: Lawyers
Comes v. MS
Contracts/Documents
Courts
DRM
Gordon v MS
GPL
Grokdoc
HTML How To
IPI v RH
IV v. Google
Legal Docs
Lodsys
MS Litigations
MSvB&N
News Picks
Novell v. MS
Novell-MS Deal
ODF/OOXML
OOXML Appeals
OraclevGoogle
Patents
ProjectMonterey
Psystar
Quote Database
Red Hat v SCO
Salus Book
SCEA v Hotz
SCO Appeals
SCO Bankruptcy
SCO Financials
SCO Overview
SCO v IBM
SCO v Novell
SCO:Soup2Nuts
SCOsource
Sean Daly
Software Patents
Switch to Linux
Transcripts
Unix Books
Your contributions keep Groklaw going.
To donate to Groklaw 2.0:

Groklaw Gear

Click here to send an email to the editor of this weblog.


To read comments to this article, go here
Oracle v. Google - Google Takes Exception to Third Damages Report [as text]
Monday, February 20 2012 @ 04:00 PM EST

Last Friday Google filed a motion to substantially strike both the third damages report submitted by Dr. Cockburn on behalf of Oracle and the conjoint analysis report submitted by Dr. Shugan on behalf of Oracle. (718 [PDF; Text]) If successful, this motion will virtually preclude the Oracle damages experts from testifying at trial. PJ did a fine job covering the flaws in these reports identified by Google. Now we consider the likelihood of success of the motion.

First, we need to recall that Judge Alsup put Oracle on a very short leash in allowing it an almost unprecedented third try to file an appropriate damages report:

On the following conditions, as to the items previously stricken, Dr. Ian Cockburn may have a third try to calculate damages so long as his methodology conforms to the prior rulings herein.

- Oracle Gets a Spanking & a 3rd Try to Come Up With a Usable Expert Damages Report

Here are the points, and the only points, that Cockburn was to correct in this third attempt:
  • Struck Dr. Cockburn’s opinion apportioning to the patents-in-suit 30% of the value of Sun’s $100 million demand to Google and assigning to the copyrights at issue 15% of the value of that demand because Dr. Cockburn failed to understand or account for the full panoply of things Google would have received had it accepted Sun’s demand. Jan. 9, 2012 Order [Dkt. No. 685] at 7-9.
  • Barred Dr. Cockburn from offering any testimony on a reasonable royalty, either as to patent or copyright damages, because of his failure to apportion the value of the patents and copyrights at issue. Id.
  • Precluded Dr. Cockburn from apportioning the value of Oracle’s patents among the asserted claims of those patents—something Dr. Cockburn has never attempted to do in either iteration of his report. Id. at 9-10.
  • Ruled that, because Dr. Cockburn had never valued the handful of code files that Oracle alleged Google to have literally copied, he may not offer damages testimony on those files at trial. Id. at 10.
  • Struck Dr. Cockburn’s testimony on future damages because Dr. Cockburn failed to calculate damages past the end of 2012 and failed to account for the varying expiration dates of the asserted patent claims. Id. at 10-11.
  • Excluded evidence of certain licenses and settlements of litigation between Sun and Microsoft, along with any testimony from Dr. Cockburn (or, presumably, any other Oracle witness) about that evidence. Id. at 11-12.

- Document 697 [PDF; Text]

In summary, in his third attempt at a damages report Dr. Cockburn was to focus solely on apportionment, first between the patents and copyrights held by Sun at the time of the 2006 negotiations, then between the copyrighted code at issue versus the copyrighted code not asserted, and finally, between the patent claims asserted and all of Oracle's (Sun's) unasserted patent claims.

So where does Cockburn start? Not at the $100 million valuation of the 2006 negotiations where we last left off. No, he starts, as PJ pointed out, by trying to inflate the base number to a point that will get his damages numbers back to an acceptable (to Oracle) level. That's not what he was told to do.

Even when Dr. Cockburn does turn to attempting to allocate values, he does not (according to Google) use any sound method to achieve the allocation. In other words, given a third attempt to get it right, Google believes Oracle has utterly failed again. It is hard to imagine that, having his patience worn thin by Oracle's first two bungled attempts, Judge Alsup will now show any greater patience or acceptance of this third report which is, to a large extent, not responsive to the direction given by the court.

And, as if this weren't bad enough, Oracle now faces losing all of its asserted claims in yet another (the third of seven) patents. On February 7, 2012, the USPTO issued a Final Rejection of claims 1, 5-7, 11-13, 15, and 16 of U.S. patent number 5,966,702. Those rejected claims include all seven (1, 6, 7, 12, 13, 15, 16) asserted by Oracle in this action. Oracle has two months in which to respond to this Final Rejection or it becomes permanent. Moreover, of the four other asserted patents, all of the asserted claims in three of those are also facing a possible final rejection. To date, only the four asserted claims of the '520 patent have been allowed following reexamination, and those are all subject to a limitation stated by Oracle in the reexamination which will likely cause them to not be infringed.

Oracle should think long and hard about whether it wants to persist on the issue of patent infringement or, for that matter, any infringement at all. Those failed settlement discussions probably look a lot more attractive to Oracle right now.


*****************

Docket

718 - Filed and Effective: February 2, 2012
MOTION
Document Text: - MOTION to Strike Portions of Third Expert Report by Iain Cockburn and Expert Report by Steven Shugan; Memorandum of Points and Authorities in Support Thereof filed by Google Inc.. Responses due by 3/2/2012. Replies due by 3/9/2012. (Van Nest, Robert)


*****************

Document

718

KEKER & VAN NEST LLP
ROBERT A. VAN NEST - #84065
[email]
CHRISTA M. ANDERSON - #184325
[email]
DANIEL PURCELL - #191424
[email]
[address]
[phone]
[fax]

KING & SPALDING LLP
DONALD F. ZIMMER, JR. - #112279
[email]
CHERYL A. SABNIS - #224323
[email]
[address]
[phone]
[fax]

KING & SPALDING LLP
SCOTT T. WEINGAERTNER (Pro Hac Vice)
[email]
ROBERT F. PERRY
[email]
BRUCE W> BABER (Pro Hac Vice)
[address]
[phone]
[fax]

IAN. C. BALLON - # 141819
[email]
HEATHER MEEKER - #172148
[email]
GREENBERG TRAURIG, LLP
[address]
[phone]
[fax]

Attorneys for Defendant
GOOGLE, INC.

UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA

ORACLE AMERICA, INC.,
Plaintff,
v.
GOOGLE INC.,
Defendant.

Case No. 3:10-cv-03561-WHA

GOOGLE'S NOTICE OF MOTION AND
MOTION TO STRIKE PORTIONS OF
THIRD EXPERT REPORT BY IAIN
COCKBURN AND EXPERT REPORT BY
STEVEN SHUGAN; MEMORANDUM OF
POINTS AND AUTHORITIES IN
SUPPORT THEREOF

Dept.: Courtroom 8, 19th Floor
Judge: Hon. William Alsup


TABLE OF CONTENTS

I. INTRODUCTION 1
II. THE THIRD COCKBURN REPORT 3
III. ARGUMENT 4
  A. Dr. Cockburn's "independent signifcance" approach to patent
apportiomnent is based entirely on Dr. Cockburn's say-so and is a
disguised resurrection of the "25 percent rule" that the Federal Circuit
struck down in Uniloc
4
  B. Dr. Cockburn's "group and value" approach to patent apportionment is
biased and based on inapposite studies of patent value.
7
  C. Dr. Cockburn failed to apportion the full value of the copyrights that
would have been included in the 2006 bundle
11
  D. Dr. Cockburn failed to conduct a claim-by-claim analysis, and hence
failed to attribute any value to the unasserted claims.
13
  E. Dr. Shugan's "conjoint" analysis is unreliable, methodologically
flawed, and the results defy common sense
14
  1. Conjoint analysis is not an accepted basis for calculating
damages
15
  2. The conjoint survey's methodology was flawed and unreliable 17
  a. The design of Dr. Shugan's conjoint survey 17
  b. The conjoint survey's fatal methodological flaws 17
  F. Dr. Cockburn's econometric analysis and calculation of market share
are based on inapplicable data and unrealistic assumptions.
20
  1. Overview of Dr. Cockburn's econometric analysis and market
share calculations
21
  2. Dr. Cockburn's econometric analysis is based on
unrepresentative data.
22
  3. Dr. Cockburn's calculation of market share is based on
unreasonable assumptions about price and consumer choice.
23
IV. CONCLUSION 25

i


TABLE OF AUTHORITIES

 
Federal Cases
 
Apple, Inc. v. Motorola, Inc.
No. 11 C 8540 (N.D. Ill.)
17, 18
Boucher v. US. Suzuki Motor Corp.
F.3d 18 (2d Cir. 1996)
23
Fractus, SA. v. Samsung Electronics Co.
N0. 6:09-cv-0203, Dkt. No. 896 (E.D. Tex. April 29, 2011)
18
Johnson Elec. N Am. Inc. v. Mabuchi Motor Am. Corp.
103 F. Supp. 2d 268 (S.D.N.Y. 2000)
19, 22
McLaughlin v. Am. Tobacco Co.
522 F.3d 215 (2d Cir. 2008)
15, 16
Medical Instrumentation and Diagnostics Corp. v. Elekta AB
No. 397-CV-00271, 2001 WTL 36151641 (S.D. Cal. Dec. 12, 2001)
23, 25
Riles v. Shell Exploration and Prod. Co.
298 F.3d 1302 (Fed. Cir. 2002)
14, 15
Schwab v. Philip Morris USA, Inc.
449 F. Supp. 2d 992 (E.D.N.Y. 2006)
15
Uniloc USA, Inc. v. Microsoft Corp.
632 F.3d 1292 (Fed. Cir. 2011)
1, 4, 5, 7, 8
 
Federal Rules
 
FED. R. CIV. P. 23(b) 15
Federal Rule of Evidence 702 22

ii


NOTICE OF MOTION AND MOTION TO STRIKE

PLEASE TAKE NOTICE that Defendant Google Inc. ("Goog1e") hereby moves to exclude portions of the opinions and testimony of Oracle America, Inc.'s ("Oracle") damages experts Dr. Iain M. Cockburn and Dr. Steven Shugan. This Motion is based on the following memorandum of points and authorities in support, the Declaration of David Zimmer ("Zimmer Decl.") and accompanying exhibits, the entire record in this matter, and on such evidence as may be presented at any hearing of this Motion, on a date and at a time to be detennined by the Court.

Dated: February 17, 2012

KEKER & VAN NEST LLP

By: s/ Robert A. Van Nest
ROBERT A. VAN NEST
Attorneys for Defendant
GOOGLE INC.

1


MEMORANDUM OF POINTS AND AUTHORITIES

I. INTRODUCTION

Oracle's damages expert Dr. Iain Cockburn's third attempt to formulate a viable damages report remains riddled with fatal errors. In response to the Court's rejection of his methodology for separating the value of the patents and copyrights at issue in this case from the remainder of the licensing bundle at the heart of the 2006 Sun—Goog1e negotiations, Dr. Cockburn now offers two alternative apportionment methods. Both are legally improper.

First, Dr. Cockburn purports to evaluate the "independent significance" of the patents-in- suit and copyrights to Google, but this "analysis" is smoke and mirrors. Dr. Cockburn throws into the hopper every piece of evidence he can identify showing that Google considered certain obvious features—speed, memory, and number of applications—important to a smartphone platform and that the inventions at issue improve those features. Based on this evidence as a whole, and without using any quantitative algorithm or formula, Dr. Cockburn simply concludes that the asserted patents are worth "at least" an even 25% of the total bundle, and the copyrights worth exactly half that much. This is a subjective guess, not an objective methodology; it is not replicable by anyone other than Dr. Cockburn. Indeed, Dr. Cockburn's "at least" 25% conclusion effectively tries to resurrect the "25 percent ru1e" that the Federal Circuit struck down just last year in Uniloc. Further, Dr. Cockburn admitted at deposition that he included the caveat "at least" to preserve his flexibility to argue a much higher percentage to the jury. He flatly stated that, having calculated the total value of the licensing bundle to be $597.5 million, he might tell the jury that the patents were worth at least 50% of that total—and that the copyrights might be so important to Android that they could be worth 100% of the bundle. This is as unforrned and speculative as expert testimony can get. It cannot possibly assist the jury.

Second, Dr. Cockburn's alternative apportionment methodo1ogy—the "group and value" approach—asked a group of five Oracle engineers to identify all the mobile Java-related patents in Sun's 2006 portfolio, separate those patents into categories, rank the categories in importance, and then rate each patent's value on a three—point scale. Four of the five Oracle engineers involved in this process conducted infringement analyses of the patents-in-suit in preparation for

1


this lawsuit, and admitted in deposition that they spent next to no time compiling their rankings and were influenced by their prior work on this case. Further, the engineers did no quantitative testing to confirm their conclusions, and Mark Reinhold, the Oracle engineer who headed up the team, admitted there was no way to translate the engineers' judgment of comparative value into the hard numbers required for a damages analysis. Dr. Cockburn tries to paper over this gap with three studies finding that, in general, a small percentage of issued patents account for a large percentage of the aggregate value of all patents. But none of these studies looked at a single company's patent portfolio in a narrow technology area, such as Sun's mobile Java portfolio, and two of the three studies evaluated European patents, not U.S. patents. The studies are Dr. Cockburn's sole basis for concluding that the top 22 patents in the Sun portfolio account for up to 92% of the total value of the portfolio. With three of the patents-in-suit—the '104, '205, and '720—placed among the top 22 by the very engineers who selected litigation patents at the outset, Dr. Cockburn calculates the value of the six patents-in-suit as falling in the broad range between 10.2% and 32.7% of the $597.5 million bundle. Dr. Cockburn has no reasonable basis for this conclusion. His group and value approach also should be stricken.

Dr. Cockburn's newest opinion has other disqualifying flaws as well. His apportionment analysis as to the copyrights at issue fails because he did not value any of the copyrighted material included in the Sun-Google negotiations other than the remaining 37 API specifications on which Oracle bases its copyright claim. Dr. Reinhold confirmed at his deposition that Sun and Google were negotiating for a partnership including a license to all of Sun's mobile Java related copyrights—among other things, the source code implementing the Java virtual machine and all of the class libraries associated with the APIs. Yet Dr. Cockburn still has no idea what was included in that universe of copyrights, and made no effort to value any of the other copyrights besides the APIs. Instead, he simply assumed that the value of that presently existing Sun intellectual property was subsumed into the projected cost of future Sun engineering expenses to be incurred during a Sun-Google partnership. There is no economic basis for using projected future engineering costs as a proxy for fair market value of copyrighted works. Moreover, despite repeated orders from this Court instructing him to do so, Dr. Cockburn

2


still refuses to value the patents on a claim-by-claim basis. This is particularly important here, where Oracle initially sued Google on 132 patent claims. Presumably Oracle had a Rule 11 basis for asserting all those claims, so all of them must have some value (even leaving aside the potentially valuable other claims of those patents that Oracle never asserted). Yet Dr. Cockburn never separated out the value of the unasserted—or asserted but abandoned—claims, nor did he deduct the value of the unasserted claims from the value of the patents-in-suit as a whole.

Finally, Dr. Cockburn also relies on his own econometric study and a "conjoint" analysis conducted by another Oracle expert, Dr. Steven Shugan. Both of these analyses are based on unrealistic assumptions and suffer from serious methodological flaws. They should be stricken. And, because Dr. Cockburn's "independent significance" approach and copyright apportiomnent rely on the conjoint and econometric analyses, they should be stricken for that reason as well.

II. THE THIRD COCKBURN REPORT

Dr. Cockburn's third report begins, as his second report did, with the negotiations Google and Sun conducted in early 2006 for a technology partnership to develop a mobile smartphone platform. As before, Dr. Cockburn uses as his monetary starting point Sun's initial February 2006 demand, which he calculates at $98.7 million, rather than Sun's final demand in April 2006 of $28 million. February 3, 2012 Cockburn Report ("Cockburn Rep.") ¶¶ 5-6. He then performs the following adjustments:

  • He adjusts the starting point upward by $557.2 million to account for convoyed sales Sun projected to make as part of its partnership with Google, resulting in a subtotal of $655.9 million. Id. ¶¶ 37-41.

  • Although Sun's initial demand contained a cap on Sun's ability to share in Google's revenue from the partnership, he removes that cap to adjust for Sun's loss of compatibility and control caused by Google's development of an independent platform. This adjustment adds a further $27.8 million, leaving the subtotal at $683.7 million. Id.

  • He then apportions the value of the patents and copyrights at issue in the suit as a percentage of the total. He uses two alternative apportionment methodologies—the "group and value" and the "independent significance" approaches. Id. ¶¶ 42-68.

    Under the group and value approach, Dr. Cockburn first adjusts downward by $86.15 million to account for projected engineering expenses Sun would have incurred as part of a partnership with Google. Id. ¶ 48. He assumes that this $86.15 million would have captured the value of (1) all copyrighted materials other than the APIs at issue, including source code and Java mobile class libraries;

    3


    and (2) all Sun engineering know-how and trade secrets. Id. ¶ 49. Next, he concludes that the JAVA trademark and Java brand was worth nothing to Google, and performs no further downward adjustment for that intellectual property. He similarly assigns no value to the fact that a partnership with Sun would have given Google access to relationships with OEMs and other Sun partners. Id. ¶ 50.

    This leaves him with a total of $597.5 million, which he contends accounts for the value of (1) Sun's Java mobile patent portfolio; and (2) the asserted copyrighted APIs. Id. 11 51. Based on a qualitative analysis by Oracle engineers, and three studies regarding the distribution of value among patents generally, he concludes that the six patents-in-suit are worth somewhere between 10.2% and 32.7% of the total, or between $69.5 million and $223.7 million. Id. ¶ 5. Based on Dr. Shugan's conjoint analysis, which suggests that consumers value the availability of applications (the Android feature purportedly enabled by the copyrighted APIs) about half as much as speed (the Android feature allegedly enabled by the asserted patents), he sets the value of the copyrights at exactly half the value of the patents—between 5.1% and 16.4% of the total, or between $34.7 million and $111.9 million. Id. ¶¶ 6, 54.

    Under the independent significance approach, Dr. Cockburn evaluates the totality of the evidence as to the importance of certain performance features to Google and to consumers, and concludes that the patents are worth "at least" 25% of the total, or at least $170.9 million. Id. ¶¶ 5, 60-68. Again relying on Dr. Shugan, Dr. Cockburn values the copyrights at exactly half the value of the patents, or "at least" 12.5% of the total, or at least $85.5 million. Id. ¶¶ 6, 60-68.

  • Dr. Cockburn then performs further downward adjustments to his alternative patent calculations to exclude damages for extraterritorial infringement, past damages for Sun's and Oracle's failure to mark its products, and damages for non-accused devices. The results are final patent-damages figures of $17.7 million to $57.1 million under the group and value approach and at least $43.7 million under the independent significance approach. Id. ¶ 5.

  • Accordingly, Dr. Cockburn's alternative total damages figures for both patent and copyright infringement are (1) between $52.4 million and $169 million under the group and value approach (assuming the Court requires all the deductions described above for extraterritorial infringement, marking, and non-accused devices); and (2) at least $129.2 million under the independent significance approach.

III. ARGUMENT

A. Dr. Cockburn's "independent significance" approach to patent apportionment is
based entirely on Dr. Cockburn's say-so and is a disguised resurrection of the "25
percent rule" that the Federal Circuit struck down in Uniloc.

Prior to the Federal Circuit's ruling last year in Uniloc USA, Inc. v. Microsoft Corp., 632 F.3d 1292 (Fed. Cir. 2011), experts often used, and courts frequently approved, the so-called "25 percent rule," which was "a tool that has been used to approximate the reasonable royalty rate that the manufacturer of a patented product would be willing to offer to pay to the patentee during a hypothetical negotiation." Id. at 1312. Under the rule, courts would simply assume, as

4


a matter of rough justice, that hypothetical negotiators would agree to a royalty rate equal to 25 percent of the licensor's expected profits for a product that incorporates the intellectual property at issue. Id. at 1312-13. In Uniloc, the Federal Circuit sounded the death knell for this sort of subjective approximation, concluding that the "25 percent rule of thumb is a fundamentally flawed tool for determining a baseline royalty rate in a hypothetical negotiation." Id. at 1315. It thus held that "[e]vidence relying on the 25 percent rule of thumb is thus inadmissible under Daubert and the Federal Rules of Evidence, because it fails to tie a reasonable royalty base to the facts of the case at issue." Id.

Unable to rely directly on the 25 percent rule or any similar rule of thumb, Dr. Cockburn surreptitiously resurrects it through what he describes as the "independent significance" approach. Cockburn Rep. ¶¶ 421-459. Under this approach, as he described it in his deposition, Dr. Cockburn [REDACTED] Declaration of David Zimmer in Support of Google's Motion to Strike ("Zimmer Decl.") Ex. A (Cockburn Dep.) at 135:6, 137:17-20.

According to his report, the evidence Dr. Cockburn reviewed included contemporaneous documents regarding the importance and value to Google of key performance characteristics, contemporaneous documents regarding the availability of non-infringing alternatives, Oracle's benchmarking studies, the opinions of Oracle technical expert John Mitchell and Google technical expert Owen Astrachan, and the opinions of Oracle's engineers. Cockburn Rep. ¶ 421. Although Dr. Cockburn did not mention them at all in the "independent significance" approach section of his report, Dr. Cockburn stated in his deposition that [REDACTED] Zimmer Decl. Ex. A (Cockburn Dep.) at 135:7-136:21.

In his report, Dr. Cockburn offers no explanation as to how he gets from this evidentiary

5


miscellany of data to his bottom-line 25% figure. In his deposition, Dr. Cockburn admitted that he had essentially made up that number because it felt right to him:

[REDACTED]

Id. at 139:2-18. Incredibly, Dr. Cockburn also admitted his 25% number was just a floor, and he might well assert an undisclosed higher number once he gets on the witness stand:

[REDACTED]

Id. at 142:7-143:3 (emphasis added). It gets worse—Dr. Cockburn also said that he might advise the jury that the copyrighted APIs could constitute 100% of the value of the 2006 partnership. Id. at 166:12-20 [REDACTED]

6


[REDACTED]

Not only is Dr. Cockburn's bottom-line apportionment percentage under the independent significance approach entirely undefined, the only thing connecting that conclusion to its purported evidentiary basis is Dr. Cockburn's subjective judgment, which by definition cannot be replicated or verified by another expert. This is no different from the guesswork that used to underlie the 25 percent rule, before the Federal Circuit's ruling in Uniloc. The Court should not allow Dr. Cockburn to resurrect the 25 percent rule by [REDACTED]

B. Dr. Cockburn's "group and value" approach to patent apportionment is biased and
based on inapposite studies of patent value.

In conducting his "group and value" approach, Dr. Cockburn relied on Dr. Reinhold and four other Oracle engineers to review Sun's portfolio of 569 mobile Java-related patents. The four engineers on whom Dr. Reinhold relied had all been involved in analyzing patents as part of this litigation, and did not recuse themselves from evaluating any of the patents-in-suit on which they had worked. Zimmer Decl. Ex. H (Rose Dep.) at 79:14-81:23, 84:24-86:8; Zimmer Decl. Ex. I (Wong Dep.) at 83:15-85:21, 88:1-95:6, 132:18-133:11; Zimmer Decl. Ex. J (Kessler Dep.) at 68:5-71:11; Zimmer Decl. Ex. K (Plummer Dep.) at 32:22-38:25, 50:25-52:22. The engineers first divided the 569 patents into 22 technology groups. Cockburn Rep. ¶¶ 391-92. The engineers then identified four metrics for evaluating the usefulness of a technology group to a mobile smartphone platform—operating speed, startup speed, memory footprint, and security— and ranked the 22 groups according to the four metrics. Then they tallied up those rankings, with the result being an aggregate ranking of the usefulness of the 22 groups. Finally, they separately rated each individual patent on a scale of l to 3 in terms of the benefits each would allegedly provide to a smartphone platform. Id. ¶¶ 393-94. The rating of the 569 patents was done over just two days. See, eg., Zimmer Decl. Ex. J (Kessler Dep.) at 33:12-34:5. One engineer [REDACTED] Zimmer Decl. Ex. K

7


(Plummer Dep.) at 37:3-25. The rating was based solely on a spreadsheet including the title, abstract, inventors and issue or filing date. See, e.g., Zimmer Decl. Ex. K (Plummer Dep.) at 32:13-21; Zimmer Decl. Ex. J (Kessler Dep.) at 40:6-41:21. It was therefore inevitable that the engineers would favor the patents they had already analyzed as part of this case. Christopher Plummer, one of the engineers, admitted he relied on his previous work in rating the patents:

[REDACTED]

Zimmer Decl. Ex. K (Plummer Dep.) at 101:25-102:14. The other engineers admitted having similar knowledge, but claimed it didn't influence them. See, e.g., Zimmer Decl. Ex. I (Wong Dep.) at 94:25-96:2 [REDACTED] Zimmer Decl. Ex. H (Rose Dep.) at 84:6-13 [REDACTED]

Dr. Cockburn then synthesized these questionable group rankings and the patent ratings by isolating all the patents in the purportedly top three technology groups with a "1" rating. The result was a group of 22 patents that he asserted constitute the most valuable 4% of Sun's mobile Java portfolio. Cockburn Rep. ¶¶ 397, 408. This group of 22 patents included three of the patents-in-suit—the '720, '205, and '104. Id. ¶ 397.

But because the engineers' analysis was entirely qualitative, Dr. Cockburn next faced the challenge of translating these value comparisons into the hard numbers required for a damages

8


calculation. The Oracle engineers themselves confirm that they had no technical basis for translating their qualitative judgment into quantitative valuations. Dr. Reinhold confirmed that the engineers did no quantitative assessment, and that such an assessment would require significant and repeated performance testing of each patent's functionality. Zimmer Decl. Ex. B (Reinhold Dep.) at 40:8-12, 41:20-42:11, 43:12-25, 47:13-20. Even then, Dr. Reinhold testified, no quantitative assessment of all the various patents would be possible given that some patents were substitutes for, or complements to, others in the portfolio. Id. at 96:7-97:9; id. at 105 [REDACTED] In the end, although Dr. Reinhold believed that the patents in the top-ranked group were each more valuable than any of the patents in the second-ranked group, he could not offer even a guess as to the ranking among these top 22 patents. Id. Dr. Reinhold conveyed this opinion to Dr. Cockburn. Zimmer Decl. Ex. A (Cockburn Dep.) at 125.

Undeterred, Dr. Cockburn tries to bridge the gap between the engineers' vague judgments of quality and a quantitative damages calculation by relying on three surveys of patent value having nothing to do with the Sun portfolio at issue. Each of these surveys1 conclude that the distribution of value among patents is highly skewed, with a handful of patents accounting for a large percentage of the value of all patents. Id. ¶ 405. Dr. Cockburn adopted the distributions of patent value described in these studies and concluded that the top 22 patents in Sun's portfolio are worth 67.9% to 91.9% of the overall value of the portfolio, with the six patents-in-suit being worth between 10.2% and 32.7% of the overall patent portfolio. Id. ¶¶ 408, 412.

The problem with this analysis is that there is no relationship between the Sun portfolio at issue in this case and the patents studied in the three surveys. As the Federal Circuit made clear in Uniloc, "[i]f the patentee fails to tie the theory to the facts of the case, the testimony must be excluded." 632 F.3d at 1315. Here, Dr. Cockburn's survey model is inapposite to the actual Sun

9


portfolio at issue here for the following reasons:

  • The Sun portfolio here is owned by a single company. None of the studies looked at single-owner portfolios at all. See, e.g., Zimmer Decl. Ex. A (Cockburn Dep.) at 94:15-95-7, 104:20-105-6. As Dr. Cockburn admitted about one of the studies, [REDACTED] Id. at 95:20-12.

  • The Sun portfolio at issue here is confined to a relatively narrow technology area: mobile smartphone platform software functionality. None of the studies had any subject-matter limitation; each of them evaluated every type of patent in every type of technology area under the sun. See, e.g., id. at 96:5-11 100:24-101:7.

  • The 569 patents in the Sun portfolio were selected deliberately for relevance to this case by Oracle engineers who also worked on pre-litigation infringement analyses of the patents-in-suit for Oracle. By contrast, the patents evaluated in the studies were randomly selected.

Each of these critical differences means that the patent populations evaluated in the three studies are likely to have different distributions of value than the Sun portfolio. Dr. Cockburn offers nothing to support his assumption that the value distribution of industry- and company- wide patents is the same as the value distribution of one software company's portfolio in a specific technology area.

Further, two of the surveys in Dr. Cockburn's report—the PatVal and Harhoff studies— looked only at European patents, and one of those two was confined to only German patents. All 569 Sun patents are U.S. patents. Id. at 95:8-12, 102:2-7. This is not a trivial difference. The German study recognized that one of its conclusions would be "surprising" to those familiar with U.S. patents. Zimmer Decl. Ex. C (Harhoff study) at 1355 ("Some of the mean values should be surprising to readers familiar with citation indicators from the US patent system. As reported in a previous paper (Harhoff et al., 1999), we find that the citation counts, both in the German and the European system, are spectacularly low by USPTO standards."). Dr. Cockburn again provides no support for his assumption that the value distribution of European patents is the same as that of U.S. patents, and the text of one of his source studies refutes that assumption.

The lone study in Dr. Cockburn's report that evaluated U.S. patents—the Barney study— is inapposite because it calculated patent value according to whether the patentee had paid the patent-renewal fees over the life of the patents. Zimmer Decl. Ex. A (Cockburn Dep.) at 105:14-

10


107:19. Dr. Cockburn did not attempt to tie this methodology to the Sun portfolio by examining whether Sun and Oracle had paid patent renewal fees for the 569 patents in the portfolio at issue. Assuming Sun and Oracle did so for all or nearly all 569 patents—and there is no evidence to the contrary—it would follow, using the logic of the Barney study, that each of those patents had an approximately equal value.

Despite the vast differences between the patents examined in these three studies and the 569 Sun patents at issue in this case, Dr. Cockburn in his deposition [REDACTED] Id. at 107:20-108:2. Instead, Dr. Cockburn simply [REDACTED] Id. at 99:11-18; see also id. at 100:20-23 [REDACTED] This type of ipse dixit link between external studies and the facts of this case is inadmissible under Daubert, as the Supreme Court made clear in General Electric Co. v. Joiner:

Trained experts commonly extrapolate from existing data. But nothing in either Daubert or the Federal Rules of Evidence requires a district court to admit opinion evidence that is connected to existing data only by the ipse dixit of the expert. A court may conclude that there is simply too great an analytical gap between the data and the opinion proffered.

522 U.S. 136, 146 (1997). In this case, the analytical gap between the three patent value studies and Dr. Cockburn's ultimate conclusions is cavernous. The Court should strike this analysis.

C. Dr. Cockburn failed to apportion the full value of the copyrights that would have
been included in the 2006 bundle.

In addition to using two unreliable patent apportionment methodologies, Dr. Cockburn failed to make any attempt to value all of the copyrights that would have been part of the 2006 intellectual property package. Indeed, Dr. Cockburn had no idea what Java-related copyrights Sun owned at the time of the hypothetical negotiation, let alone what they were worth.

The Court stated unequivocally in its January 9, 2012 Order that "[i]f the $100 million in

11


2006 is used as the starting point . . . then a fair apportionment of the $100 million as between the technology in suit and the remainder of the technology then offered must be made." Jan. 9, 2012 Order [Dkt. No. 685] at 8. In the patent context Dr. Cockburn at least made an attempt to follow the Court's instructions by apportioning the patents included in the 2006 bundle between the patents in suit and the remainder of Sun's Java-related patents. In the copyright context, however, Dr. Cockburn did not even attempt to undertake this analysis. Dr. Cockburn admitted in his deposition that he did not even know what Java-related copyrights Sun owned in 2006:

[REDACTED]

Zimmer Decl. Ex. A (Cockburn Dep.) at l53:17-154:3.

Without knowing what copyrighted material was at issue, it was of course impossible for Dr. Cockburn to apportion all of that copyrighted material between the 37 APIs at issue and the rest of the material. As Dr. Cockburn said in his deposition: [REDACTED] Id. at 161:9-12; see also id. at 161:22-162:2 [REDACTED] Yet there is no reason to think that there were not other copyrights being considered. Most obviously, the source code underlying Sun's implementation of the Java virtual machine is copyrighted, and at least would have been a basis of any new virtual machine jointly developed by Sun and Google. See id. at 158:7-159:11 [REDACTED] Because he made no effort to consider the scope of the copyrights at issue in the 2006

12


negotiations, Dr. Cockburn's analysis does not account for these or other copyrights that Google would have obtained through the hypothetical negotiation.

Dr. Cockburn also made no systematic effort to measure the value of the millions of lines of code in the API libraries that would have been part of the 2006 bundle. Unlike in the patent context, Dr. Cockburn never had anyone from Oracle examine the code libraries to determine their value in relation to the API specifications. Instead, he simply assumed that, whatever other copyrights were on the table in the Sun-Google negotiations, their value would have been subsumed in the operating and research-and-development expenses Sun projected it would incur as part of its partnership with Google. Accordingly, Dr. Cockburn avoids any specific valuation of those copyrighted materials at all. But Dr. Cockburn has no logical basis for using Sun's projected future R&D costs (in developing new intellectual property in a mobile smartphone platfonn partnership with Google) as a proxy for the value of Sun's then-existing intellectual property (the copyrighted class libraries and source code). This analysis falls apart for at least two reasons. First, Dr. Cockburn confuses cost with value. Consider the patents-in-suit—it would have cost Sun nothing in terms of R&D costs to license those patents to Google in 2006, because those inventions were already developed and patented. But Dr. Cockburn would contend that the patents have significant value to Google. Second, Dr. Cockburn is again mixing apples and oranges, by comparing two entirely distinct classes of intellectual property—Sun's existing copyrighted materials that were the subject of the licensing negotiation, on the one hand, and material that Sun might have developed during a partnership with Google, on the other.

Dr. Cockburn's conflation of projected cost with actual value, and his equal treatment of past copyrighted works and different, future copyrighted works are both efforts to cover up the fact that he has engaged in no rigorous evaluation of the individual values of the copyrights in the 2006 licensing bundle. Indeed, as already discussed, he cannot even identify the components of that bundle. His copyright apportionment analysis is unreliable and should be excluded.

D. Dr. Cockburn failed to conduct a claim-by-claim analysis, and hence failed to
attribute any value to the unasserted claims.

The court has repeatedly emphasized that Dr. Cockburn is required to calculate Oracle's

13


purported patent damages on a claim-by-claim basis. Jan. 9, 2012 Order [Dkt. N0. 685] at 9-10. Yet Dr. Cockburn has still failed to do so, asserting that [REDACTED] Cockburn Rep. ¶ 497. Thus, as the Court ordered before, Dr. Cockburn should be "precluded from apportioning an asserted patent's value among its claims at trial." Jan. 9, 2012 Order [Dkt. No. 685] at 9.

More problematic, however, is that Dr. Cockburn fails to attribute any value to any of the unasserted claims of the patents-in-suit. In his deposition, Dr. Cockburn [REDACTED] Zimmer Decl. Ex. A (Cockburn Dep.) at 90:13-18. This raises the distinct possibility that a portion of the value of some of the patents-in-suit may be located in claims that Google has not infringed. As the Court noted in its January 9 Order: "An 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. Indeed the idea that the unasserted claims had no value is belied by Oracle's actions in this case. At one time, Oracle claimed that Google infringed 132 claims from 7 patents. See May 3, 2011 Order [Dkt. No. 131] at 1. Oracle has now limited its patent case to 26 claims from 6 patents. But Oracle must have had a Rule 11 basis for asserting infringement of the now-abandoned claims by Google, so it cannot be heard to argue that those claims have no value to Google. Yet Dr. Cockburn has never tried to isolate the value of those claims—or other claims that Oracle never asserted—and deduct the value of the unasserted claims from the remaining value of the patents-in-suit. Yet again, he has violated the Court's express instructions and overstated Oracle's damages as a result.

E. Dr. Shugan's "conjoint" analysis is unreliable, methodologically flawed, and the
results defy common sense.

In addition to the flaws in his own analysis, Dr. Cockburn relies on a "conjoint" analysis

14


performed by another Oracle expert, Dr. Steven Shugan. Dr. Shugan's conjoint study purports to identify a limited number of smartphone features that consumers find important, then determine which of those features consumers value most. Then Dr. Shugan essentially converts the consumers' preference share into projected market shares—essentially, he concludes that, if 20% of consumers value application start time more than the other tested features, an increase in application start time on Android phones would mean a 20% drop in Android market share. This simplistic analysis may have some value as a marketing-research tool, but it is not an accepted method of calculating damages. As far as Google is aware, no court has ever approved the use of conjoint analyses in calculating damages.

Even setting aside the problems with conjoint analyses generally, the particular survey in this case had crippling methodological flaws, as is shown by the facially absurd responses it generated. Dr. Shugan's conjoint analysis is unreliable and would be unhelpful to the jury, and the Court should exclude it. Similarly, the parts of Dr. Cockburn's report relying on Dr. Shugan's analysis, primarily his "independent significance" and copyright apportionment analyses, see Cockburn Rep. ¶ 419; Zimmer Decl. Ex. A (Cockburn Dep.) at 135:12-15; 138:9-18; 140:10-142:3, also should be excluded.

1. Conjoint analysis is not an accepted basis for calculating damages.

Conjoint analysis is a marketing-research tool, not an accepted method of calculating damages in litigation. Calculation of reasonable-royalty damages in a patent case "requires sound economic and factual predicates." Riles v. Shell Exploration and Prod. Co., 298 F.3d 1302, 1311 (Fed. Cir. 2002). Whatever its value may be to marketing research, conjoint analyses like Dr. Shugan's simply lack the reliability required to prove damages with the "reasonable certainty" required in litigation. See id.; N.D. CAL. MODEL PATENT JURY INSTRUCTION No. 5.1.

Neither Dr. Shugan nor Dr. Cockburn cite to a single instance where a court in any jurisdiction has allowed the parties to rely on "conjoint" analyses to prove damages. Indeed, in one of the few reported decisions to discuss conjoint analyses, the Second Circuit reversed a trial court decision that relied on conjoint analysis to grant a class certification motion under FED. R. CIV. P. 23(b). McLaughlin v. Am. Tobacco Co., 522 F.3d 215, 234 (2d Cir. 2008), rev'ing sub

15


nom Schwab v. Philip Morris USA, Inc., 449 F. Supp. 2d 992, 1056 (E.D.N.Y. 2006). The plaintiffs in McLaughlin offered a conjoint survey to establish that they bought "light" cigarettes because of defendants' misrepresentations that those cigarettes were healthier than "full-flavored" cigarettes. Schwab, 449 F. Supp. 2d at 1167. According to plaintiffs' conjoint expert, 90.1% of class members considered "health concerns" as a factor in purchasing light cigarettes. Id. at 1048. The trial court cited plaintiffs' conjoint analysis in finding class-wide reliance, and the court also noted that the conjoint analysis would "probably be critical in estimation of damages." Id. at 1126, 1147, 1169. The trial court refused to exclude the conjoint analysis, and certified a nation-wide class of "light" smokers. Id. at 1170, 1278.

The Second Circuit reversed. The court found that the conjoint analysis was insufficient to establish reliance because it failed to take account of the myriad other factors that may have influenced class-members' decision to buy light cigarettes—e.g., taste or "personal style." Id. at 223. The court sharply criticized the use of conjoint analysis to prove reliance, stating:

Plaintiffs' expert, Dr. John R. Hauser, claimed that 90.1% of those who smoked Lights chose to do so because of Lights' alleged health benefits. But Dr. Hauser came to this conclusion on the basis of a method that determined whether, all things being equal, consumers prefer a safer cigarette to a less safe cigarette. And as plaintiffs conceded at oral argument, no one who understood this question would prefer a more dangerous product to a safer one.

Id. at 225 n.6 (internal citations omitted) (emphases added). Accordingly, the Second Circuit concluded that plaintiffs had failed to establish damages with "'sufficient precision to allow a jury award.'" Id. at 232 (quoting and rejecting the district court's conclusion as to damages).

McLaughlin illustrates the limitations of conjoint analysis. Conjoint analysis measures consumer preference for product features; it does not capture how consumers actually behave when purchasing a product. Consumers' stated preference for a given feature may be one of many factors a company considers in designing or launching a new product, but they are not a proxy for market share. The dearth of opinions discussing use of conjoint analysis in litigation further suggests that conjoint studies cannot provide the "reasonable certainty" needed to support a multi-million dollar damages claim. See July 22, 2011 Order [Dkt. No. 230] at 12 (striking Dr. Cockburn's reliance on the Nash bargaining solution in part because it "has never been approved

16


by a judge to calculate reasonable royalties in litigation, at least in the face of objection. This is despite the fact that for decades it has been lurking in the field of economics.").

2. The conjoint survey's methodology was flawed and unreliable.

Even assuming that conjoint analysis is reliable enough for damages calculations—and it is not—Dr. Shugan's survey in this case was methodologically flawed for at least two reasons. First, Dr. Shugan's selection of features was driven by the litigation, and the choice profiles he created do not approximate real-world purchase conditions. Second, as Dr. Shugan himself concedes, the survey violated the fundamental premise of conjoint analysis: that respondents are able to hold constant all product features other than those tested by the survey.

a. The design of Dr. Shugan's conjoint survey.

Dr. Shugan used a web-based survey to measure the relative importance to consumers of seven smartphone features: application multitasking, application startup time, availability of third-party applications, brand, price, screen size, and voice command capabilities. Shugan Rep. App. D at 5. The survey asked respondents to choose between side-by-side comparisons of different smartphone "profiles." Each profile was a written list of varying levels of functionality in each of the seven features—e.g., one phone might be described as (1) an Android phone with (2) a 4.5-inch screen that (3) can run five apps at once (4) with a startup time of 2 seconds, (5) 300,000 available apps, and (6) voice dialing and texting (7) available for a sale price of $200. Respondents were instructed to assume that every feature other than the seven listed features was the same for each profile. Shugan Rep. App. E at E12-E20.2.

Dr. Shugan used respondents' selections to rank and measure the relative importance of the seven features to consumers. He then plugged these ranked values—referred to in conjoint parlance as "partworths"—into a statistical software program in order to assess general consumer preference for an Android phone lacking the application volume, startup time, and multitasking A 25 capabilities allegedly provided by the patents- and copyrights-in-suit.

b. The conjoint survey's fatal methodological flaws.

First, Dr. Shugan's selection of allegedly important smartphone features for the conjoint survey was not driven, as one would expect, by which features are actually valuable to

17


consumers. Instead, Dr. Shugan selected features that were important to Oracle's needs in this litigation. He included just seven features in his smartphone "profiles," and five of those seven features—all features other than brand and price—were spoon-fed to him by either Dr. Cockburn or the Analysis Group, Dr. Cockburn's consulting group. Zimmer Decl. Ex. D (Shugan Dep.) at 29:22-30:11. Indeed, Dr. Shugan's own "market research" identifies 36 other features that real- world consumers actually said they would (or have) considered in purchasing a smartphone, including such obviously important features as the associated cellular network, battery life, and keyboard type and layout. See Shugan Report Ex. 1. Dr. Shugan made no effort, prior to administering his seven-feature survey, to determine whether any of the 36 other features he chose to omit were actually more important to consumers than the seven he actually tested. Instead, Dr. Shugan's survey ignores those 36 features, and asks consumers to assume that any feature not specifically listed among the seven tested features is the same for all phones. But all phones are not the same in the real world. Forcing consumers to operate under this assumption undermines the survey's ability to predict real-world behavior. As Judge Posner noted in a recent hearing in the Apple v. Motorola case, responding to a suggestion from counsel for Apple that Apple could measure the value of a patent through a consumer survey:

I'm not going to ask consumers how they like it. That is a totally — that is a totally fraudulent way of determining — I mean, look, you go out and start asking consumers, oh, this device had X and meant that [a certain feature was enabled,] would you pay more for that? Of course, they'll say yes, right, because you're focusing them on some feature.

Zimmer Decl. Ex. E (Transcript of Jan. 23, 2012 Hearing, Apple, Inc. v. Motorola, Inc., No. 11 C 8540 (N.D. Ill.)) at 22-23; see also id. at 23 ("I'm not going to have competing experts talking about their marketing surveys. I regard that evidence as totally worthless."). This flaw in conjoint analysis renders the results unfit for the demanding task of calculating actual damages in a multi-million dollar litigation.

Second, Dr. Shugan's study includes the same error this Court recognized in striking Dr. Cockburn's second report last month—it measures the value consumers place on certain phone features as a whole, rather than the incremental benefit to those features allegedly enabled by the

18


technology at issue. Courts have struck similar consumer surveys for this reason. See Fractus, S.A. v. Samsung Elecs. Co., No. 6:09-cv-0203, Dkt. No. 896 (E.D. Tex. Apr 29, 2011) (excluding consumer survey data, that "do[es] not measure the value of Plaintiff s technology, but merely measure[s] the perceived consumer value of cell phones with any internal antennas" and concluding that "[s]urvey evidence purportedly demonstrating the value of intemal antennas not tied directly to Plaintiff's technology confuses the issues and must be excluded").

Third, the results of the survey clearly show that respondents were not in fact holding constant all unnamed features. Dr. Shugan's failure to control for this fact is fatal, because the ability of conjoint analysis to predict consumer preference for a specified feature requires the consumer to ignore potential differences in any unspecified features. Dr. Shugan explained this fundamental presumption of conjoint analysis in his deposition:

[REDACTED]

Zimmer Decl. Ex. D (Shugan Dep.) at 38:2-25 (emphases added).

But the proof is in the pudding—and here, almost one quarter of all respondents claimed that they would prefer a smartphone costing $200 to a putatively identical smartphone costing $100. Zimmer Dec. Ex. F (Oct. 24, 2011 Leonard Rep.) at 114.2 No rational person, much less 24% of all rational people, would prefer to pay $200 for a phone they could have for $100. The explanation for this apparently nonsensical result is obvious—survey respondents were not holding non-specified features constant. Instead, as Dr. Shugan himself conceded in his reply report, his respondents did the opposite:

[REDACTED]

19


[REDACTED]

Shugan Reply Rep. at 17-18.

Dr. Shugan makes no effort to explain this result other than to say that we cannot assume that consumers behave rationally because the purpose of the conjoint analysis was to study consumer behavior. Id. at 17-18. This circular logic ignores the premise of Daubert and the Federal Rules of Evidence. The point is not that Dr. Shugan should have excluded a full quarter of his respondents as "irrational," but rather that the conclusion that 24% of survey respondents gave answers that made no sense proves that the survey design is fundamentally flawed. Courts routinely recognize that a "common sense" understanding of real-world consumer behavior is an important check against the reliability of surveys that produce preposterous results. See Johnson Elec. N Am. Inc. v. Mabuchi Motor Am. Corp., 103 F. Supp. 2d 268, 286 (S.D.N.Y. 2000) (striking a survey that, "despite its dazzling sheen of erudition and meticulous methodology, reaches a result which any average person could readily recognize as preposterous"). The results of Dr. Shugan's conjoint analysis are equally preposterous—not because consumers are irrational in the real world, but because of glaring flaws in the design of the survey.

Even if conjoint analysis was a recognized and appropriate methodology for calculating damages in litigation, which it is not and never has been, Dr. Shugan has failed to design a conjoint survey that approximates real-world purchase conditions, and further created a survey with such fundamental methodological flaws that it has returned nonsensical results. His survey is too unreliable to be the basis of a multi-million dollar damages claim. Dr. Shugan's report—and Dr. Cockburn's reliance upon that report—should be excluded under Daubert.

F. Dr. Cockburn's econometric analysis and calculation of market share are based on
inapplicable data and unrealistic assumptions.

Dr. Cockburn's econometric analysis purports to quantify consumer preferences for some smartphone features over others, but it draws its data not from the sales prices of mobile phone carriers who sell the vast majority of mobile phones in the United States, but rather from sales data for largely second-hand phones on the eBay auction website. This error is compounded by two unrealistic assumptions that Dr. Cockburn makes in converting his econometric analysis to

20


an estimate of market share in a world in which Google had released a slower version of Android, both of which inflate Oracle's damages. The Court should exclude this analysis.

1. Overview of Dr. Cockburn's econometric analysis and market share
calculations.

Dr. Cockburn began with data he acquired of auctions on eBay for mostly second—hand smartphones. This data included, for each auction, both (a) the maximum price each bidder would have been willing to pay for the phone, which he deemed to be the consumer's "willingness to pay," and (b) the price for which the phone actually sold. Cockburn Rep. App. C ¶ 13. Collecting consumers' willingness to pay is possible because eBay bidders are allowed to enter their maximum bid, and the computer will only bid the amount necessary to win the auction. Id. ¶ 5. Dr. Cockburn also collected data from other sources about the attributes of each phone being auctioned, such as battery life, storage space, and presence of a camera. Id. ¶ 25.

Based on this data, Dr. Cockburn conducted a regression analysis to attempt to predict the impact on a consumer's willingness to pay for a given phone from a change in the phone's features. Id. ¶¶ 29-30. The most important feature was the phone's "Linpack benchmark," which is one way of measuring a phone's operating speed. Id. ¶¶ 26-27. Oracle's engineers attempted to measure the impact the patents-in-suit had on smartphone performance by measuring the change in the Linpack benchmark when the patents were disabled. Id. ¶ 37.

Dr. Cockburn then attempted to convert his regression results into a calculation of smartphone market shares in the world in which Google released a slower version of Android. To make that calculation, Dr. Cockburn looked only at eBay bidders who placed bids on multiple smartphone models. Id. ¶ 39. For each of these bidders, Dr. Cockburn measured the bidder's so-called "consumer surplus" for each phone on which each bidder bid. Id. Consumer surplus is measured by calculating the difference between the consumer's willingness to pay for each phone (i.e. the consumer's maximum bid on eBay) and the price at which each phone sold. Id. n.25. Dr. Cockburn then used his regression results—specifically the coefficient on the Linpack benchmark that predicts the impact of a change in the Linpack benchmark on the consumer's willingness to pay—to estimate what each consumer's willingness to pay would have been for

21


the slower version of Android. Id. ¶ 41. Dr. Cockburn then measured each bidder's adjusted consumer surplus for the Android phones. Id.

Dr. Cockburn concluded that if the bidder's adjusted Android consumer surplus remained higher than the consumer surplus for non-Android phones, that bidder would still buy the Android phone. If the adjusted Android consumer surplus became lower for the Android phone than for the non-Android phone, then the bidder would buy the non-Android phone. And if all consumer surpluses were negative—i.e. if the consumer's willingness to pay for non-Android phones and adjusted Android willingness to pay are all lower than the price at which the phones eventually sold—then the consumer would not buy a smartphone at all. Id. Based on his determination of what each consumer would have done if Google had released a slower version of Android, Dr. Cockburn estimated what portion of consumers who in reality bought an Android phone would have switched to some other phone, or no phone at all, then estimated the effect of those lost sales on Google's Android revenue. See id. App. D.

2. Dr. Cockburn's econometric analysis is based on unrepresentative data.

Dr. Cockburn incorrectly and without any support assumes that the data for smartphone purchases on eBay is representative of the entire market for smartphones. Courts strike expert testimony when it is based on unrepresentative and unreliable data. See, e. g., Johnson Elec., 103 F. Supp. 2d at 283 ("[E]ven where an expert's methodology is reliable, if the analysis is not based upon relevant and reliable data, the expert's opinion will be inadmissible." (citing Joiner, 522 U.S. at 146)).

The vast majority of cell phones in the United States are sold new, as part of two-year agreements with cell phone carriers like Verizon or AT&T. Dr. Cockburn makes no effort to show that people buying unlocked, second-hand phones on eBay have the same consumer preferences as consumers buying new cell phones as part of a two-year service agreement. Dr. Cockburn never explains why he did not attempt to obtain data that measures how most Americans purchase cell phones, nor did he make any argument as to why the Court or the jury should assume that eBay customers' consumer preferences are representative of the general population. As the Johnson Electric court noted in striking an expert's report, although "[i]t can

22


be appropriate to utilize market reconstruction to prove damages, particularly in patent cases ... if [plaintiff] wishes to reconstruct the micro-motor market, that reconstruction must be grounded on the most relevant and reliable data available." Johnson Elec., 103 F. Supp. 2d at 286 (emphasis added). By contrast, "speculative economic analysis must be rejected." Id. Here, instead of using data from phone carriers, the way nearly all American consumers purchase cell phones, Dr. Cockburn took a shortcut by using non—representative eBay data that he makes no effort to connect to the actual market for smartphones. Daubert does not allow such shortcuts.

3. Dr. Cockburn's calculation of market share is based on unreasonable
assumptions about price and consumer choice.

At least as problematic as Dr. Cockburn's econometric analysis, if not more, is his effort to translate that econometric analysis into an estimate for what smartphone market share would have been had Google released a slower version of Android. Dr. Cockburn's translation into market share is based on two unrealistic assumptions that significantly inflate Oracle's damages. These calculations must be excluded as unreliable. See Boucher v. US. Suzuki Motor Corp., 73 F.3d 18, 21 (2d Cir. 1996) ("[A]n expert's testimony should be excluded as speculative if it is based on unrealistic assumptions."); Medical Instr. & Diagnostics Corp. v. Elekta AB, No. 397-CV-00271, 2001 WL 36151641 (SD. Cal. Dec. 12, 2001); see also July 22, 2011 Order [Dkt. No. 230] at 13 (Dr. Cockburn's use of the Nash bargaining solution "would invite a miscarriage of justice by clothing a fifty-percent assumption in an impenetrable facade of mathematics").

First, Dr. Cockburn's market share calculation is based on the incorrect assumption that although every consumer would be willing to pay less for a slower version of Android, the price of Android would remain constant. This assumption leads Dr. Cockburn to significantly overestimate the number of eBay users who would have chosen not to purchase an Android phone if Google had released a slower version of Android, which in tum leads Dr. Cockburn to overestimate Oracle's damages. As described above, Dr. Cockburn began with each bidder's actual consumer surplus, measured as the difference between the price that the consumer bid for the phone, and the price at which the phone sold. He then used his regression results to calculate the maximum price each consumer would have bid for the slower version of Android—a number

23


that is smaller than the maximum price the consumer actually bid. Dr. Cockburn then took this adjusted maximum bid and compared it to the unadjusted price at which the phone actually sold to get his adjusted consumer surplus.

In doing so, Dr. Cockburn assumed that if Google released this slow version of Android, the price for which it would sell on eBay would be exactly the same price as the current version of Android. Zimmer Decl. Ex. G (Oct. 17, 2011 Cockburn Dep.) at 106 [REDACTED] This assumption makes no sense. Plainly the eBay price of a slower Android would be lower because an eBay sales price is determined by the second highest bidder, whose willingness to pay would also decrease. But in measuring the adjusted consumer surplus of a given bidder, Dr. Cockburn adjusts the willingness to pay of that bidder while assuming that price—and hence all other bidders' willingness to pay—would be the same as under the current, much faster version of Android. If Dr. Cockburn accounted for the fact that other bidders' willingness to pay would have similarly decreased in the world in which Google released a slower Android, he would have found that far fewer people (if any) would have diverted from Android to other platforms.

Second, Dr. Cockburn assumed each bidder's choices are limited to the specific phones on which the bidder bid during a ten-day window. This unrealistic assumption again leads Dr. Cockburn to overestimate the number of people who would not have bought the slower Android phones. In Dr. Cockburn's artificial world, if a bidder bid on two phones—for example, an Android phone and an iPhone—-—Dr. Cockburn assumes that those are the only two phones that consumer would consider purchasing. If that bidder lost the iPhone auction, and also would have lost the Android auction based on his or her adjusted willingness to pay, Dr. Cockburn concludes that that bidder would not acquire a smartphone at all. Dr. Cockburn ignores the possibility that the bidder might have gone to a Verizon or AT&T store to buy a phone, or waited two weeks and then returned to eBay to try again. This incorrect assumption leads to bizarre results. According to Dr. Cockburn, out of all the people who bought Android phones in 2010, over 15% of all the people who bought Android phones in 2010 would not have acquired a smartphone at all had Google released Android without the patents in suit. Cockburn Rep. Ex. C-9 (purporting

24


to show that of the 55.4% of actual Android purchasers who would not have purchased an Android in the but-for world, 27.4% would have bought no smartphone).3 This unrealistic result significantly inflates Oracle's damages. According to Dr. Cockburn, Google's revenues decrease somewhat when consumers switch from Android to other platforms because Google still collects downstream revenue, but Goog1e's revenues from consumers who do not purchase smartphones at all go all the way to zero. Cockburn Rep. App. D ¶¶ 11-23. Under Dr. Cockburn's model, Google's revenues take a much bigger hit when a consumer decides not to buy a smartphone at all, as compared to when a consumer simply switches to another brand (as would logically be the case in the real world).

Both of these erroneous assumptions have an obvious inflationary effect on Oracle's damages numbers. When Dr. Cockburn's errors are corrected, the results fundamentally undermine his market share calculations and reveal them to be "unhelpful to the trier of fact and inadmissible at trial." Medical Instr., 2001 WL 36151641.

IV. CONCLUSION

For all the reasons set forth above, this Court should strike Dr. Cockburn's third damages report, and further exclude Dr. Cockburn's econometric analysis and Dr. Shugan's conjoint analysis under Daubert and its progeny.

Dated: February 17, 2012

KEKER & VAN NEST LLP

By: s/ Robert A. Van Nest
ROBERT A. VAN NEST
Attorneys for Defendant
GOOGLE, INC.

25


1 The three studies are A. Gambardella, P. Giuri & M. Mariani, The Value of European Patents - Evidence from a Survey of European Inventors, Final Report of the PatVal EU Project, January 2005, D. Harhoff, F. Scherer & K. Vopel, Citations, family size, opposition and the value of patent rights, 32 Research Policy 1343, (September 2003), and J. A. Barney, A Study of Patent Mortality Rates: Using Statistical Survival Analysis to Rate and Value Patent Assets, 30 AIPLA Quarterly Journal, Vol. 30, No. 3, Summer 2002.
2 Similarly, 26% of the respondents claimed to prefer a phone with an application startup time of 2 seconds to a phone with startup time of 0.2 seconds. Id. at 115.
3 Notably, Dr. Shugan's conjoint analysis does not even allow for the possibility that an Android user would have purchased no phone if Google released a slower version of Android.

  View Printable Version


Groklaw © Copyright 2003-2013 Pamela Jones.
All trademarks and copyrights on this page are owned by their respective owners.
Comments are owned by the individual posters.

PJ's articles are licensed under a Creative Commons License. ( Details )