There are a number of key points that Google drives home in the petition:

• The timing of the email: It is not an email from 2006 when the alleged infringement began; it is an email from the summer of 2010 following Oracle's threat against Google.
• The prejudice shown by Judge Alsup against privileged communication of in-house counsel:
  
  

  Going far beyond the D.C. Circuit’s holding [in In re Sealed Case], the district court extended this “clear showing” standard [i.e., that that the party claiming privilege carries the burden of showing that an in-house lawyer gave advice in a professional legal capacity] to all in-house lawyers because they are presumed to do more non-legal work than outside counsel. The district judge’s written opinion justified this approach by citing his long-held personal skepticism about privilege claims involving in-house counsel [emphasis added]. He also endorsed the Sealed Case standard and urged this Court to adopt it in this anticipated writ proceeding.

  Based on its unfounded assumptions about in-house lawyers, and wielding its enhanced “clear showing” standard against Google, the district court speculated that Lindholm “may well have been” discussing non-privileged business negotiations rather than Google’s internal legal investigation of Oracle’s claims. The district court then held (in the face of copious contrary evidence) that Google had failed to disprove that possibility by submitting evidence linking the email to Google’s legal investigation. Under its “clear showing” standard, the district court held, in effect, that the entire basis for asserting the privilege must be evident from the face of the document.

• Oracle exploited Google's inadvertent production of the eight "auto-saved" drafts of the email: This would appear to be a dangerous precedent in the face of modern email systems, and it is clear that Oracle knew exactly what it was doing:

  Twice on July 21, 2011, Oracle disclosed in court proceedings the contents of an inadvertently produced and incomplete draft of the Lindholm email, without providing prior notice to Google as the protective order requires. Google was taken by surprise because that version of the draft lacked Lee’s name or the work-product heading. Google immediately investigated the document and learned that it was an auto-saved draft of a privileged document listed twice on Google’s privilege log. Google clawed back the drafts the very next day.

• The actions of the magistrate and district court judge were inconsistent with the law and have exacerbated the problem.

Jump To Comments

1. The district court's order is clearly erroneous as a matter of law (incorrectly adopting In re Sealed Case as the standard for judging privileged communications and overriding Upjohn;
2. Google proved that the Lindholm email was privileged under Upjohn; and
3. If the district court's ruling is allowed to stand it would virtually eviscerate the attorney-client privilege as it relates to in-house counsel.

Of course, Oracle's take on the petition is much different. Not relying entirely on In re Sealed Case, Oracle emphasizes that the magistrate and district court judge did not commit error because Google failed to make a "clear showing" that the Lindholm email was privileged. Oracle asserts Google made no showing of privilege whatsoever other than the tagging of the email. Oracle also continues to argue that you can't "put the genie back in the bottle," i.e., having publicly disclosed the email Google cannot now un-disclose it. This argument belies Oracle's significant and almost exclusive role in the disclosure of the email.

More problematic for Google, if what Oracle states is true, is the following:

The Magistrate Judge made numerous, specific findings of fact adverse to Google, which were all affirmed by the District Court, and which Google does not directly challenge. Nonetheless, Google ignores those adverse findings, and instead advances a self-serving and incomplete recitation of the facts that was rejected on the record established below. At the same time, Google has submitted an appendix that omits one of its own declarations; includes only a few lines of the transcript of the hearing before the Magistrate Judge (A84-A90); includes a declaration Google submitted on a different motion after the Magistrate Judge had ruled (A33-A45); and omits all of the evidence that Oracle submitted and was considered in the proceedings below. The petition thus ignores the proper standard of review and fails to comply with Federal Rule of Appellate Procedure 21(a)(2)(C).

The follow-up letter from Google alerts the Federal Circuit to the fact that the district court denied Google's motion in limine pertaining to the Lindholm email, meaning that this petition for writ is now the only avenue open to Google.

The path to protection of the troubled Lindholm email is still very uncertain.

In other filings we see one final argument [PDF; Text] from Oracle pertaining to Google's motion in limine number 3, which Judge Alsup has now largely granted. There is nothing terribly revealing in the response; it is mostly a rehash of prior arguments.

The parties also continue to argue over the last of the claim construction language. Oracle argues [PDF; Text] that the language is clear and that Google's interpretation is all wrong. Google argues [PDF; Text], on the other hand, that Oracle is ignoring its own definitions/limitations on the language stated in the specification. The court should have heard enough on this topic to now settle on the final interpretation.

Finally, the parties again offer their opinions and proposed modifications to the final pretrial order. Google is largely okay [PDF; Text] with the order other than suggesting more time is needed for the patent and damages phases of the trial and that, with some limitations, that videotaped third-party testimony should be permitted.

Oracle still does not like [PDF; Text] the trifurcated approach to the trial [copyright, patent, damages] and the delay in getting to trial. The one thing Oracle request is a shortening of the first two phases and adding a phase 3A on injunctive relief prior to going to the issue of damages. However, given the interrelationship between injunctive relief and damages, it is hard to see how the two could be separate.

Here are all the documents:

The following three documents were made available through a January 11, 2012, article by Alison Frankel of Thomson Reuters entitled: "Google mandamus on Lindholm email: Key test of client privilege" We will make an html version of these documents available as soon as possible.

Google Petition for Writ of Mandamus

Oracle Response to Google Petition for Writ of Mandamus

Google Follow-up Letter re Petition for Writ of Mandamus

688 – Filed and Effective: 1/11/2012
RESPONSE
Document Text: Response ORACLE AMERICA INC.S RESPONSIVE SUPPLEMENTAL BRIEF IN OPPOSITION TO 549 re 494 GOOGLES MOTION IN LIMINE NO. 3 TO EXCLUDE PORTIONS OF COCKBURN REPORT ON DAMAGES byOracle America, Inc.. (Attachments: # 1 Exhibit Exhibits A, C-F, H, L to Dearborn Declaration)(Holtzman, Steven) (Filed on 1/11/2012) (Entered: 01/11/2012)

689 – Filed and Effective: 1/11/2012
Sealed Document
Document Text: DOCUMENT E-FILED UNDER SEAL re 687 Order on Administrative Motion to File Under Seal EXHIBITS J & M TO THE DECLARATION OF MEREDITH DEARBORN IN SUPPORT OF 688 ORACLE AMERICA INC.S RESPONSIVE SUPPLEMENTAL BRIEF IN OPPOSITION TO 549 re 494 GOOGLES MOTION IN LIMINE NO. 3 TO EXCLUDE PORTIONS OF COCKBURN REPORT ON DAMAGES by Oracle America, Inc.. (Holtzman, Steven) (Filed on 1/11/2012) (Entered: 01/11/2012)

690 – Filed and Effective: 1/11/2012
RESPONSE
Document Text: RESPONSE to Oracle's Response to Final Pretrial Order by Oracle America, Inc.. (Jacobs, Michael) (Filed on 1/11/2012) (Entered: 01/11/2012)

691 - Filed and Effective: 01/11/2012
RESPONSE
Document Text: RESPONSE to re 647 Objection Oracle's Responsive Brief Regarding Objections To Google's Proposed Claim Constructions by Oracle America, Inc.. (Jacobs, Michael) (Filed on 1/11/2012) (Entered: 01/11/2012)

692 - Filed and Effective: 01/11/2012
RESPONSE
Document Text: RESPONSE to re 675 Order and Objections to Final Pretrial Order by Google Inc.. (Van Nest, Robert) (Filed on 1/11/2012) (Entered: 01/11/2012)

693 - Filed and Effective: 01/11/2012 -
BRIEF
Document Text: Brief in Reply to Oracle's Proposed Claim Constructions filed by Google Inc.. (Attachments: # 1 Declaration, # 2 Exhibit A, # 3 Exhibit B, # 4 Exhibit C, # 5 Exhibit D)(Sabnis, Cheryl) (Filed on 1/11/2012) (Entered: 01/11/2012)

Here are the documents that we have done as text, beginning with Google's Petition for Writ of Mandamus, then Oracle's response, then Google's follow-up letter, and then the other miscellaneous documents listed above, with links so you can find what you want easily:

[ Google's Petition  ]  [ Oracle's response ]  [ Google's letter ]  [ 688 ] 
[ 690 ]  [ 691 ]  [ 692 ]  [ 693 ] 

Google's Petition:

United States Court of Appeals

for the

Federal Circuit

------------------

IN RE GOOGLE INC.,

Petitioner

___________________________________________
___________________________________________

Petition for Writ of Mandamus to the United States District Court,
Northern District of California, San Francisco Division
in 3:10-CV-03561-WHA

____________________________________________
____________________________________________

PETITION FOR WRIT OF MANDAMUS

Robert A. Van Nest
Christa M. Anderson
Steven A. Hirsch
KEKER & VAN NEST LLP
[address, phone]

Scott T. Weingaertner
Bruce W. Baber
King & Spalding LLP
[address, phone]

Ian C. Ballon
Greenberg Traurig, LLP
[address, phone]

Attorneys for Petitioner Google Inc.

NOVEMBER 4, 2011

CERTIFICATE OF INTEREST

Counsel for petitioner Google Inc. certifies the following:

1. The full name of every party we represent (which are the real parties in interest) are:

Google Inc.

2. All parent corporations and any publicly held companies that own 10 percent of more of the stock of the parties we represent are:

None.

3. The name of all law firms and the partners or associates that appeared in the trial court for the party we now represent or are appearing in this Court are:

KEKER & VAN NEST LLP

Robert A. Van Nest, Christa M. Anderson, Steven A. Hirsch,
Michael S. Kwun, Daniel E. Purcell, Eugene M. Paige,
Matthias A. Kamber, Reid P. Mullen

KING & SPALDING LLP

Donald F. Zimmer, Jr., Cheryl A. Sabnis, Scott T. Weingaertner,
Brian C. Banner, Bruce W. Baber, Christopher C. Carnaval,
Geoffrey M. Ezgar, Mark H. Francis, Robert F. Perry,
Steven T. Snyder, Truman H. Fenton

i

GREENBERG TRAURIG, LLP

Ian C. Ballon, Heather J. Meeker, Dana K. Powers, Joseph R. Wetzel
Luis Villa, IV, Valerie W. Ho, Wendy M. Mantell

GOOGLE INC.

Renny F. Hwang

Dated: November 4, 2011

KEKER & VAN NEST LLP By:[signature]
ROBERT A. VAN NEST
Attorneys for Defendant
GOOGLE INC.

ii

TABLE OF CONTENTS

Page

CERTIFICATE OF INTEREST .........................i

TABLE OF AUTHORITIES ......................... v

I. Petition for Writ of Mandamus and Relief Sought....................... 1

II. Introduction............................................... 1

III. Principal Issues Presented . ................. 6

IV. Statement of Facts.......................................................... 6

A. Soon after Oracle charged Google with patent infringement,
Google began an Upjohn-type internal investigation of
Oracle's claims. ............................... 6

B. Mr. Lindholm's August 6, 2010 email reported some results
of Google's Upjohn-type investigation of Oracle's claims. ................. 7

C. Oracle exploited Google's inadvertent production of eight
"auto-saves" of the Lindholm email to attack the email's
privileged status. ................................. 9

D. The district court erroneously compelled production of
the Lindholm email and drafts. ................................. 10

V. Reasons why the writ should issue ....................... 12

A. The district court's order is clearly erroneous as a matter of
law and raises new and important problems of first
impression for mandamus review. ............................ 13

1. The district court erroneously held that a corporation
must make a heightened showing that a
communication from an employee to an in-house
lawyer had a primarily legal purpose--even when
both parties undisputedly were engaged in an Upjohn-
type internal investigation. ........................... 14

iii

a. Google proved that the Lindholm email was
privileged under Upjohn and under Ninth
Circuit precedents that treat in-house and
outside counsel the same for privilege purposes. ........... 14

b. The district court adopted a distorted version of
the D.C. Ciruit's Sealed Case standard, which
allowed it to rule based on unwarranted
assumptions about in-house counsel instead of
the undisputed evidence.................................. 19

2. The district court's expanded Sealed Case standard
led it into additional legal errors. ............................... 25

3. The district court erroneously denied work-product
protection to the Lindholm email..................... 27

4. The district court clearly erred by deferring to the
magistrate's erroneous legal conclusions, in violation
of Rule 72(a). .............. 27

B. The remaining mandamus factors are satisfied as well................ 29

1. Google has no other means, such as an appeal, to
obtain the desired relief. .............................. 29

2. Google will be damaged or prejudiced in ways not
correctable on appeal. ............................ 29

VI. Conclusion .................................................. 30

iv

TABLE OF AUTHORITIES

Page(s)

Cases

Admiral Ins. Co. v. U.S. Dist. Ct. for Dist. of Ariz.,
881 F.2d 1486 (9th Cir. 1989) ................................. 13, 16, 22, 29

Barthelemy v. Air Lines Pilots Ass'n,
897 F.2d 999 (9th Cir. 1990) ...............................................26

Barton v. U.S. Dist. Ct.,
410 F.3d 1104 (9th Cir. 2005) .............................26

Bauman v. U.S. Dist. Ct,
557 F.2d 650 (9th Cir.1977) ........................................ passim

Boca Investerings P'ship v. U.S.,
31 F. Supp. 2d 9 (D.D.C. 1998) .....................................22

Brownlow v. Gen. Motors Corp.,
CIV. A. 3:05CV-414-R, 2007 WL 2712925 (W.D. Ky. Sept. 13, 2007) ............28

Burlington N. & Santa Fe Ry. Co. v. U.S. Dist. Ct.,
408 F.3d 1142 (9th Cir. 2005) ..........................2

Computer Econ., Inc. v. Gartner Group, Inc.,
50 F. Supp. 2d 980 (S.D. Cal. 1999) .................................28

Gomez v. Vernon,
255 F.3d 1118 (9th Cir. 2001) .........11

Hemstreet v. Spiegel, Inc.,
851 F.2d 348 (Fed. Cir. 1988) .............18

Hercules, Inc. v. Exxon Corp.,
434 F. Supp. 136, 196 U.S.P.Q. 401 (D. Del. 1977) .......................25

Hernandez v. Tanninen,
604 F.3d 1095 (9th Cir. 2010) .................................13

Holly Farms Corp. v. N.L.R.B.,
517 U.S. 392 (1996)................28

v

In re Brand Names Prescription Drugs Litig.,
No. 94 C 897, 1995 WL 557412 (N.D. Ill. Sept. 19, 1995)..................18

In re Columbia/HCA Healthcare Billing Practices Litig.,
293 F.3d 289 (6th Cir. 2002) ...............................27

In re Kaypro,
218 F.3d 1070 (9th Cir. 2000) .................................26

In re Nintendo Co., Ltd.,
589 F.3d 1194, 1197, 93 U.S.P.Q.2d 1152 (Fed. Cir. 2009).............12

In re OM Secs. Litig.,
226 F.R.D. 579 (N.D. Ohio 2005) .........................18

In re Sealed Case,
737 F.2d 94 (D.C. Cir. 1984) ............................. passim

Jacobs v. Nintendo of Am., Inc.,
370 F.3d 1097 (Fed. Cir. 2004) ..........................................18

Lenz v. Universal Music Corp.,
C 07-3783 JF (RS), 2009 WL 3573990 (N.D. Cal. Oct. 30, 2009) ..............17

Med. Imaging Ctrs. of Am., Inc. v. Lichtenstein,
917 F. Supp. 717 (S.D. Cal. 1996) ...............................29

Milwaukee Carpenter's Dist. Council Health Fund v. Philip Morris, Inc.,
70 F. Supp. 2d 888 (E.D. Wis. 1999) ..................29

Mohawk Indus., Inc. v. Carpenter,
130 S. Ct. 599 (2009) ..................... 11, 12, 30

Motley v. Marathon Oil Co.,
71 F.3d 1547 (10th Cir. 1995) ...................26

Navajo Nation v. U.S. Forest Serv.,
535 F.3d 1058 (9th Cir. 2008) ............................28

Perry v. Schwarzenegger,
591 F.3d 1147 (9th Cir. 2010) ..................................... 12, 14, 28

vi

Potter v. United States,
No. 02-CV-0632-H (POR), 2002 WL 31409613 (S.D. Cal. July 26, 2002) ...............................18

PowerShare, Inc. v. Syntel, Inc.,
597 F.3d 10 (1st Cir. 2010) .................................28

Segerstrom v. United States,
No. C 00-0833 SI, 2001 WL 283805 (N.D. Cal. Feb. 6, 2001) ...........................18

Therasense, Inc. v. Becton, Dickinson & Co.,
C 04-02123 WHA, 2008 WL 2323856 (N.D. Cal. May 22, 2008)......................19

Transamerica Computer Co. v. IBM Corp.,
573 F.2d 646 (9th Cir. 1978) ................................11

U.S. v. Ritchey,
632 F.3d 559 (9th Cir. 2011) .......................................19

United States v. Bergonzi,
216 F.R.D. 487 (N.D. Cal. 2003) .......................................27

United States v. Chen,
99 F.3d 1495 (9th Cir. 1996) ..................................... 16, 17, 19

United States v. ChevronTexaco Corp.,
241 F. Supp. 2d 1065 (N.D. Cal. 2002) ..................................20

United States v. KPMG LLP,
237 F. Supp. 2d 35 (D.D.C. 2002) ...................................22

United States v. Rowe,
96 F.3d 1294 (9th Cir. 1996) ..........................................16

United States v. Ruehle,
583 F.3d 600 (9th Cir. 2009) ...........................................29

Upjohn Co. v. United States,
449 U.S. 383 (1981) ...................................... passim

Wessel v. City of Albuquerque,
No. MISC 0000532 ESH/AK, 2000 WL 1803818 (D.D.C. Nov. 30, 2000) ...............22

vii

Rules

Federal Rule of Civil Procedure 72(a) .................................... passim

Other Authorities

Grace M. Giesel, The Legal Advice Requirement of the
Attorney-Client Privilege: A Special Problem for in-House
Counsel and Outside Attorneys Representing Corporations,
48 Mercer L. Rev. 1169 (1997) ................................... 17, 23, 24

12 Wright & Miller, Federal Practice & Procedure § 3069, at
350 (2d ed. 1997) ...................................................29

viii

I. Petition for Writ of Mandamus and Relief Sought

Google Inc. ("Google"), defendant in the above-captioned case, petitions this Court to issue a writ of mandamus directing the United States District judge presiding over the case to enter an order (1) confirming that the August 6, 2010 email (and all drafts thereof) authored by Tim Lindholm are privileged; (2) granting all of the relief detailed in the Proposed Order submitted below (A1);1 (3) sealing its October 20, 2011 order denying Rule 72(a) relief (A4); and (4) vacating its November 2, 2011 order stripping the Lindholm email and drafts of their confidentiality designations (Dkt. 596).

II. Introduction

"[T]he [attorney-client] privilege exists to protect not only the giving of professional advice to those who can act on it but also the giving of information to the lawyer to enable him to give sound and informed legal advice."

Upjohn Co. v. United States, 449 U.S. 383, 390 (1981).

"Adding the name of a lawyer to a list of business recipients is exceedingly easy and is very often done without any intention that it be used to frame legal advice."

A14 at n.4.

Thirty years ago, the Supreme Court issued its landmark ruling in Upjohn Co. v. United States,² which recognized the importance of maintaining confidentiality in corporate internal investigations. Upjohn held that the

1

effectiveness of such investigations depends upon a broad attorney-client privilege that protects communications between in-house lawyers and lower-level employees, who often possess the knowledge that lawyers need if they are to render sound legal advice.

Now--in a ruling that undermines Upjohn and reflects an unwarranted distrust of in-house lawyers--the district court ³ has compelled production of a confidential email between Google engineer Tim Lindholm and Google in-house counsel Ben Lee (among others). The email concerned an investigation that Mr. Lindholm conducted at the direction of Google's lawyers as part of the legal department's effort to evaluate a major lawsuit for the company's senior management. In that lawsuit, real party in interest Oracle America, Inc. ("Oracle") asserts that Google's Android operating system--now the world's leading "smart phone" platform--infringes patents and copyrights that Oracle acquired in 2010 when it purchased Sun Microsystems. Oracle has claimed billions in damages.

The Lindholm email presents a classic Upjohn scenario--a corporation asking its counsel to communicate with a lower-level employee to investigate a threatened legal claim against the corporation. Google's uncontradicted evidence showed that:

• Oracle lawyers met with Google lawyers on July 20, 2010 to present

2

Oracle's claims that parts of Google's Android platform infringed on several Oracle patents. Oracle threatened to sue Google over those patents. Three weeks later, Oracle filed this lawsuit.
• After the July 20, 2010 meeting, Google Senior Counsel Ben Lee asked Google engineer Tim Lindholm to gather information related to Oracle's infringement claims to assist Google's legal analysis of those claims.

• On July 30, 2010, Google General Counsel Kent Walker convened a meeting to formulate a response to Oracle's infringement claims. Attorney Lee and Mr. Lindholm attended that meeting, as did Google's top management. At the meeting, Attorney Walker instructed Mr. Lindholm and Google engineer Dan Grove to continue to work under Attorney Lee's supervision to gather information about the technology underlying Oracle's patent-infringement claims.

• On August 6, 2010, Lindholm sent Attorney Lee and others an email reporting on some of the results of his investigation stemming from Oracle's litigation threat and the July 30, 2010 meeting involving Google legal counsel and top management. The email concerned the very technology that Oracle--only days before--had told Google was infringing Oracle's patents (and also mentioned related technology potentially affected by Oracle's claims). The email was not intended to convey general business advice about Android.

• Lindholm's email said "Attorney Work Product" and "Google Confidential" at the top. During the five minutes that it took Lindholm to draft the email, Google's computer system "auto-saved" nine drafts.

• The final version of the Lindholm email was listed, twice, on Google's privilege log. Eight drafts were produced inadvertently and then "clawed back" pursuant to a protective order, triggering this discovery battle.

Despite this evidence, the district court compelled production of the final Lindholm email and of the auto-saved drafts (collectively, "the Lindholm email"). The district court held that Google had failed to establish that the Lindholm email

3

"constitutes a communication related to the purpose of obtaining legal advice from a legal advisor in his capacity as such."

The district court reached that result by adopting and expanding upon a D.C. Circuit precedent, In re Sealed Case,⁴ which requires a heightened "clear showing" of the communication's legal purpose where the lawyer in question is an in-house lawyer who is also a corporate officer with significant non-legal responsibilities. Going far beyond the D.C. Circuit's holding, the district court extended this "clear showing" standard to all in-house lawyers because they are presumed to do more non-legal work than outside counsel. The district judge's written opinion justified this approach by citing his long-held personal skepticism about privilege claims involving in-house counsel. He also endorsed the Sealed Case standard and urged this Court to adopt it in this anticipated writ proceeding.⁵

Based on its unfounded assumptions about in-house lawyers, and wielding its enhanced "clear showing" standard against Google, the district court speculated that Lindholm "may well have been" discussing non-privileged business negotiations rather than Google's internal legal investigation of Oracle's claims. The district court then held (in the face of copious contrary evidence) that Google had failed to disprove that possibility by submitting evidence linking the email to Google's legal investigation. Under its "clear showing" standard, the district court

4

held, in effect, that the entire basis for asserting the privilege must be evident from

That approach to privilege claims is contrary to Upjohn's teachings and intent, because it effectively strips privilege from communications by lower- and mid-level employees to in-house counsel unless those employees are sufficiently savvy and detail-oriented to include specific references to the legal investigation in each of those communications. Under the district court's approach, the employee's failure to include such references in each communication cannot be remedied later by declarations explaining the connection between the communication and the legal investigation.

That is not how courts review privilege claims. Rather, courts read and evaluate extrinsic evidence that furnishes a context for, and explains, the privileged communication. Here, by contrast, the district court went so far as to discount the entire declaration of Mr. Lee--the in-house lawyer who received, described, and explained the context of Lindholm's email--because it didn't contain a superfluous affirmation that he "read" the email. This ultra-hyper-technical approach reflected the district court's extreme distrust of in-house lawyers and its consequent adoption (and overextension) of the Sealed Case standard.

For these and other reasons detailed below, this Court should issue a writ of mandamus ordering the district court to grant the relief set forth in Part I, above.

5

III. Principal Issues Presented

1. Did the district court commit clear error as a matter of law, warranting mandamus relief, when it held that a corporation must make a heightened “clear showing” that an email from an employee to an in-house lawyer had a primarily legal purpose—even when both parties undisputedly were engaged in the type of internal corporate investigation discussed in Upjohn Co. v. United States?

2. Did the district court violate Federal Rule of Civil Procedure 72(a), and thus commit clear error as a matter of law, warranting mandamus relief, when it deferred to the legal conclusions of the magistrate judge?

IV. Statement of Facts

A. Soon after Oracle charged Google with patent infringement, Google
began an Upjohn-type internal investigation of Oracle’s claims.

The following undisputed facts were supported by Google’s in camera submission of the final Lindholm email,⁶ supplemented by five declarations of current and former Google employees.⁷ The declarations are attached.

On July 20, 2010, Google Senior Counsel Ben Lee attended a meeting at Google with attorneys representing Oracle.⁸ At that meeting, Oracle asserted that Google infringed several Oracle patents.⁹ Oracle threatened to sue Google over

6

those patents.¹⁰ Just over three weeks later, Oracle made good on its threat, filing this lawsuit on August 12, 2010.¹¹

After the July 20, 2010 meeting, Google Senior Counsel Ben Lee asked Google engineer Tim Lindholm to gather information related to Oracle's infringement claims to assist Google's legal analysis of those claims.¹²

On July 30, 2010, Google General Counsel Kent Walker convened a meeting to formulate a response to Oracle's infringement claims.¹³ Attorney Lee and Mr. Lindholm attended that meeting, as did Google's top management.¹⁴ At the meeting, Attorney Walker instructed Mr. Lindholm and Google engineer Dan Grove to continue to work under Attorney Lee's supervision to gather information about the technology underlying Oracle's patent-infringement claims.¹⁵

B. Mr. Lindholm's August 6, 2010 email reported some results of Google's
Upjohn-type investigation of Oracle's claims.

On August 6, 2010, Mr. Lindholm sent an email to Attorney Lee and to Andy Rubin (then a Google Vice President in charge of Android).¹⁶ The final version of the email, sent at 11:05 a.m., contained the phrases "Attorney Work

7

Product" and "Google Confidential" at the top, and later was listed twice on Google's privilege log.¹⁷ Mr. Lindholm's declaration states that his email "report[ed] investigations and analyses that Mr. Grove and I conducted at the request of Google General Counsel Kent Walker, under the supervision of Mr. Lee, and in anticipation of Oracle's threatened lawsuit."¹⁸ Likewise, Mr. Lee's declaration states: "On or about August 6, 2010, I received an email from Mr. Lindholm regarding the investigation Mr. Walker and I had asked him to conduct. On information and belief, I understand that two copies of this email were listed on Google's privilege log as entries 2551 and 5513 and that one copy of it has been submitted in camera to the Court."¹⁹

Mr. Lindholm also declares that his email did not contain "general business advice" about Android.²⁰ Rather, he was following up on the July 30, 2010 meeting by reporting on the results of his continued investigation of Oracle's infringement claims so that Attorneys Lee and Walker could develop legal advice and convey that advice to Google's top management.²¹

Six days after Mr. Lindholm sent his email to Attorney Lee, Oracle filed its

8

complaint in this action.²² The Lindholm email addressed some of the technology that, according to Oracle, infringed Oracle's patents (and the email also mentioned related technology potentially affected by Oracle's claims).²³

C. Oracle exploited Google's inadvertent production of eight "auto-saves"
of the Lindholm email to attack the email's privileged status.

Google inadvertently produced to Oracle eight drafts of the Lindholm email.²⁴ As the district court found, the drafts are "auto-save" snapshots of the email that Mr. Lindholm ultimately sent, taken at eight different stages of preparation during the five-minute period in which he drafted the email.²⁵ The drafts slipped through Google's privilege-review system because they depicted the email before Lindholm had added the attorney's name or any privilege legend.²⁶

This privilege battle arose because Oracle willfully violated the protective order to which the parties had agreed. Indeed, one disturbing aspect of this dispute is that the district court's ultimate order compelling disclosure rewarded Oracle's misconduct. Twice on July 21, 2011, Oracle disclosed in court proceedings the contents of an inadvertently produced and incomplete draft of the Lindholm email, without providing prior notice to Google as the protective order requires.²⁷ Google

9

was taken by surprise because that version of the draft lacked Lee's name or the work-product heading. Google immediately investigated the document and learned that it was an auto-saved draft of a privileged document listed twice on Google's privilege log. Google clawed back the drafts the very next day.²⁸

D. The district court erroneously compelled production of the Lindholm
email and drafts.

Oracle then moved the magistrate to compel production of the Lindholm email and all drafts thereof.²⁹ The magistrate did not question the honesty or integrity of Google's declarants.³⁰ Yet she stated that Google had failed to "connec[t] the dots" by specifically linking the Lindholm email to the pre-litigation investigation described in the declarations.31The magistrate applied a variant of the D.C. Circuit's Sealed Case "clear showing" standard for communications with in-house counsel and concluded that, under that standard, Google had failed to establish that the Lindholm email was "a communication related to the purpose of obtaining legal advice from a legal advisor in his capacity as such."³² Instead, the magistrate speculated that Lindholm "may well have been communicating with Lee about other non-privileged matters, including the business of negotiating for a

10

Java license."³³ The magistrate speculated that this was "a simple and reasonable explanation for the Email that Google ma[de] no effort to foreclose."³⁴

Google then produced the documents--prominently stamped with privilege legends--under a written reservation of the right to claw them back (again) if a court later confirmed their privileged status.³⁵

The district court later denied Google's Rule 72(a) motion for relief from the magistrate's order. The district court noted that the "clear showing" standard applied by the magistrate originated in the D.C. Circuit's Sealed Case decision and acknowledged that the Ninth Circuit had "yet to address" that standard.³⁶ The district court also expressed skepticism about the role and intentions of in-house counsel, praised the Sealed Case standard, and effectively urged this Court to adopt that standard:

In his prior career, the undersigned judge practiced in a large civil litigation firm for 25 years and had considerable experience with and exposure to the practice of company officers and employees routinely copying

11

internal (and even external) counsel on all manner of business communications as an attempt to cloak a business message in privilege. Adding the name of a lawyer to a list of business recipients is exceedingly easy and is very often done without any intention that it be used to frame legal advice. For that reason, the rule adopted by now-Justice Ginsburg in In re Sealed Case makes considerable sense and addresses real-world practices. This experience is added for the benefit of the record and for any appellate review. It has not colored the outcome of this order save and except to reinforce the wisdom of the holding in In re Sealed Case.³⁷

V. Reasons why the writ should issue

"[L]itigants confronted with a particularly injurious or novel privilege ruling" may obtain a writ of mandamus "in extraordinary circumstances--i.e., when a disclosure order `amount[s] to a judicial usurpation of power or a clear abuse of discretion,' or otherwise works a manifest injustice." Mohawk Indus., Inc. v. Carpenter, 130 S. Ct. 599, 607-08 (2009) (citation omitted). Federal appellate courts "rely on mandamus to resolve `new [privilege] questions that otherwise might elude appellate review' or `to protect important or clear claims of privilege.'" Perry v. Schwarzenegger, 591 F.3d 1147, 1157 (9th Cir. 2010) (citation omitted).

The Ninth Circuit³⁸ applies five criteria, known as the "Bauman factors,"³⁹

12

to determine whether mandamus is appropriate in a given case: (1) whether the petitioner has no other means, such as an appeal, to obtain the desired relief; (2) whether the petitioner will be damaged or prejudiced in any way not correctable on appeal; (3) whether the district court order is clearly erroneous as a matter of law; (4) whether the district court's order is an oft-repeated error or manifests a persistent disregard of the federal rules; (5) whether the district court's order raises new and important problems or issues of first impression. See Admiral Ins. Co. v. U.S. Dist. Ct. for Dist. of Ariz., 881 F.2d 1486, 1490-91 (9th Cir. 1989). The factors serve as "guidelines"--"a point of departure" for the Court's analysis of the propriety of mandamus relief. Id. at 1491. The petitioner need not satisfy all five factors--indeed, "the fourth and fifth Bauman factors are rarely, if ever, present at the same time." Id.; see Hernandez v. Tanninen, 604 F.3d 1095, 1101-02 (9th Cir. 2010) (granting writ where only first three Bauman factors were satisfied and challenged privilege ruling was "particularly injurious").

Because the third and fifth elements--clear error as a matter of law raising new and important problems--are critical to this petition, we discuss them first.

A. The district court's order is clearly erroneous as a matter of law and
raises new and important problems of first impression for mandamus
review.

The third and fifth Bauman factors are met here because the district court's ruling not only misinterprets and misapplies the law, but throws a wrench into corporate internal investigations by undermining the confidentiality that the

13

Upjohn ruling was intended to protect. It does so by adopting a radically broadened version of the D.C. Circuit's Sealed Case standard, which the Ninth Circuit and this Court have yet to address.

Under the third Bauman factor, when the Court of Appeals is "firmly convinced that a district court has erred in deciding a question of law, [it] may hold that the district court's ruling is 'clearly erroneous as a matter of law as that term is used in mandamus analysis.'" Perry, 591 F.3d at 1158. For all the reasons set forth below, the district court's privilege rulings are clearly erroneous.

1. The district court erroneously held that a corporation must make
a heightened showing that a communication from an employee to
an in-house lawyer had a primarily legal purpose—even when
both parties undisputedly were engaged in an Upjohn-type
internal investigation.

The district court relied on an erroneous legal standard to hold that Google had failed to establish “the first three prongs of the attorney-client privilege test, namely that the Lindholm Email constitutes a communication related to the purpose of obtaining legal advice from a legal advisor in his capacity as such.”⁴⁰ This erroneous standard infected its entire analysis and spawned additional legal errors, discussed below.

a. Google proved that the Lindholm email was privileged
under Upjohn and under Ninth Circuit precedents that
treat in-house and outside counsel the same for privilege
purposes.

The lodestar for Google’s privilege claim is the Supreme Court’s landmark

14

Upjohn decision, which established the importance of preserving the confidentiality of communications made between employees and in-house counsel during internal corporate investigations.

Upjohn involved communications generated in the course of a company's internal investigation of possible illegal payments to foreign governments. The company's chairman asked its general counsel to conduct the investigation; and the general counsel (after consulting with outside counsel) distributed a questionnaire to certain managers about the payments. See 449 U.S. at 386-87. The general counsel and outside counsel also interviewed those managers and other company officers and employees. See id. The IRS later began an investigation to determine the tax consequences of the questionable payments. The IRS demanded production of the completed questionnaires, memoranda, and interviews generated by the company's internal investigation. See id. at 387-88. The district court and court of appeals ordered disclosure. See id. at 388.

The Supreme Court reversed in a decision that recognized the importance of allowing both in-house counsel and outside counsel to communicate confidentially with low- and mid-level employees when investigating possible legal claims against the corporation. The Court held that the effectiveness of such investigations depends upon a broad attorney-client privilege, extending beyond the corporation's "control group" of top officers to embrace communications with

15

lower-level employees, who often possess the knowledge that an attorney needs in order to render sound legal advice. See id. at 388-97. As the Upjohn court observed, the attorney-client privilege "recognizes that sound legal advice or advocacy serves public ends and that such advice or advocacy depends upon the lawyer's being fully informed by the client." 449 U.S. at 383. "The first step in the resolution of any legal problem is ascertaining the factual background and sifting through the facts with an eye to the legally relevant." Id. at 390-91. The attorney-client privilege therefore "protect[s] not only the giving of professional advice to those who can act on it but also the giving of information to the lawyer to enable him to give sound and informed advice." Id. at 390; see also United States v. Chen, 99 F.3d 1495, 1501 (9th Cir. 1996); Admiral Ins., 881 F.2d at 1492-93.

Significantly, Upjohn involved communications with both in-house and outside counsel--and the decision drew no distinction between them for purposes of establishing privilege. In keeping with Upjohn's teachings, the Ninth Circuit adheres to a general principle that, "[i]n determining the existence of a privilege, no attempt is made to distinguish between 'inside' and 'outside' counsel." United States v. Rowe, 96 F.3d 1294, 1296 (9th Cir. 1996) (citation, brackets, and internal quotation marks omitted). "This position accords with the traditional view in the United States that no difference exists for professional or ethical purposes between an in-house counsel and an outside counsel." Grace M. Giesel, The Legal Advice

16

Requirement of the Attorney-Client Privilege: A Special Problem for in-House Counsel and Outside Attorneys Representing Corporations, 48 Mercer L. Rev. 1169, 1207 (1997) [hereinafter, "In-House Counsel"].

The Ninth Circuit has rejected the contention--endorsed by the district court in this case--that in-house counsel are different from other lawyers for privilege purposes because they sometimes become involved in the corporation's "business," as opposed to its legal issues.⁴¹ Rather, the controlling question is whether the lawyer was employed to give legal advice based on his "knowledge and discretion in the law," regardless of whether the subject of that advice is "criminal or civil, business, tort, domestic relations, or anything else." Chen, 99 F.3d at 1501-02. When deciding this question, "[c]alling the lawyer's advice 'legal' or 'business' advice does not help in reaching a conclusion; it is the conclusion. That the lawyers were 'involved in business decision-making' . . . is irrelevant." Id. at 1502 (emphasis in original).⁴² "A client is entitled to hire a lawyer, and have his secrets kept, for legal advice regarding the client's business

17

affairs." Id. at 1501.⁴³

Under these principles, Google clearly made out its privilege claim. This was a classic Upjohn scenario in which corporate counsel (Ben Lee and Kent Walker) enlisted the help of lower-level employees (Tim Lindholm and Dan Grove) to investigate potential legal claims against the corporation. The fact that the Lindholm email refers to licensing as an alternative to litigation in no way suggests that Attorney Lee was not being consulted in his legal capacity. Indeed, no bright line can be drawn between IP litigation and license negotiations because licenses are a typical component of settlements involving an intellectual-property dispute. See, e.g., Jacobs v. Nintendo of Am., Inc., 370 F.3d 1097, 1098-99 (Fed. Cir. 2004).⁴⁴

This case bears no resemblance to those in which the privilege has been held inapplicable because the lawyer was employed for "business" purposes, "without

18

'reference to his knowledge and discretion in the law.'" Chen, 99 F.3d at 1502. That conclusion may be justified, for example, "where a counterfeiter hired a man who was a lawyer to buy printing equipment for him," or where the clients hired the attorney to find investment opportunities, see id. at 1501, or to provide valuation services. See U.S. v. Ritchey, 632 F.3d 559, 566-67 (9th Cir. 2011). Here, by contrast, the uncontradicted evidence submitted by Google shows that Attorney Lee was tasked by Google General Counsel Kent Walker with overseeing an investigation of facts relating to Oracle's infringement claims and threat of litigation--a quintessentially legal task calling upon Lee's "'knowledge and discretion in the law.'" Id. at 1502.

Accordingly, the Lindholm email was privileged, and only the district court's clear errors of law--discussed below--allowed it hold otherwise.

b. The district court adopted a distorted version of the D.C.
Ciruit's Sealed Case standard, which allowed it to rule
based on unwarranted assumptions about in-house counsel
instead of the undisputed evidence.

The district court cast the governing legal principles aside in holding that Google had failed to prove that the Lindholm email constitutes "a communication related to the purpose of obtaining legal advice from a legal advisor in his capacity as such."⁴⁵ Instead, the district court adopted an erroneous and overbroad variant of the "clear showing" standard announced 27 years ago by the D.C. Circuit in In

19

re Sealed Case, 737 F.2d 94. As applied by the district court, that standard erects a nearly insuperable obstacle to establishing that Upjohn-type communications involving in-house counsel are privileged.

In Sealed Case, a corporation asserted the right to instruct a grand-jury witness who had been the corporation's vice president, general counsel, and sole in-house attorney not to answer questions concerning some putatively privileged communications. See 737 F.2d at 96-97. The district court disagreed and granted most of the government's requests to compel the attorney to testify. See id. at 97.

On appeal, the D.C. Circuit affirmed as to some rulings and reversed as to others. The court observed that, because the attorney had been the company's vice president with "responsibilities outside the lawyer's sphere," the company had to make a "clear showing" that he had given his advice "in a professional legal capacity." Id. at 99. Notably, the D.C. Circuit stated that the mere fact that the attorney had served as an in-house counsel did not "alone . . . dilute the privilege."

Id. In this case, the district court went further by holding, in effect, that "the mere fact" that Lee served as an in-house counsel did "dilute the privilege," even though there was no evidence that Lee had any "responsibilities outside the lawyer's sphere." Citing a single district-court case that relied on Sealed Case,⁴⁶

20

the magistrate opined that "Lee's role as in-house counsel warrants heightened scrutiny."⁴⁷ She held that Google therefore was required to make "a 'clear showing'" that the "'primary purpose' of the [Lindholm email] was securing legal advice'";⁴⁸ and she then concluded that "Google ha[d] made no such showing."⁴⁹ The district judge (improperly)⁵⁰ deferred to the magistrate's legal conclusions and applied the same test.⁵¹

The district court's interpretation and use of the Sealed Case doctrine was prejudicial error, for three reasons.

First, the Ninth Circuit has never adopted Sealed Case. As discussed above, the Ninth Circuit takes a much more nuanced, case-specific approach to privilege claims involving in-house counsel. Under Ninth Circuit precedents, a court has to consider the actual evidence about what the lawyer was doing, rather than relying on unfounded assumptions about in-house counsel generally.

Second, the district court expanded the "clear showing" standard far beyond its original context (concerning communications with in-house counsel who are also corporate officers with non-legal duties) to cover communications with all in-house counsel. That was clear error. Even district courts in the D.C. Circuit

21

(where Sealed Case governs)⁵² recognize that a different rule applies to ordinary in-house counsel: "There is a [rebuttable] presumption that a lawyer in the legal department or working for the general counsel is most often giving legal advice, while the opposite presumption applies to a lawyer . . . who works for . . . some other seemingly management or business side of the house." Boca Investerings P'ship v. United States, 31 F. Supp. 2d 9, 12 (D.D.C. 1998) (emphasis added).⁵³ Had that presumption been applied here, the Lindholm email should have been found privileged, because it is undisputed that Mr. Lee works for Google's legal department and was directed by Google's general counsel to conduct an Upjohn- type investigation of Oracle's claims;⁵⁴ and there was no evidence that Lee had

22

"responsibilities outside the lawyer's sphere." Sealed Case, 737 F.2d at 99.

Third, the district court's expanded version of Sealed Case substitutes speculation and bias for evidence. A commentator points out that Sealed Case and similar district-court holdings are founded upon the "unsubstantiated and odd" assumptions that in-house counsel "abuse the privilege" and "do not render legal assistance." In-House Counsel at 1208. But a survey of New York executives, law firm partners, in-house counsel, and judges revealed that outside attorneys give "business" advice at about the same frequency as do in-house attorneys who have no official nonlegal responsibilities. Id. at 1211. By relying on mere assumptions about the conduct and role of in-house lawyers, this expanded and misinterpreted Sealed Case doctrine excuses courts from carefully considering the evidence in the cases before them. That is both inappropriate and deeply corrosive of the privilege:

[C]ourts should not assume that abuse of the privilege is occurring in particular cases and should not assume that communications involving attorneys relate to or do not relate to legal advice, service, or assistance. Such assumptions are, at best, based on speculation and guessed probabilities. Courts should require all types of privilege claimants to prove that the privilege applies. Because of the nature of corporations, they may find doing so more difficult. A different standard of proof, however, seems inappropriate absent specific evidence of abuse or specific evidence of inapplicability of the privilege.

23

In-House Counsel at 1216. The district court's approach to privilege law creates new and important problems of first impression (thereby satisfying the fifth Bauman mandamus factor). Extending Sealed Case to all in-house lawyers introduces substantial uncertainty as to whether the privilege will apply to any given communication generated by an Upjohn-type internal corporate investigation. The Upjohn court sought to facilitate corporate internal investigations by guaranteeing that the extensive attorney-client communications they generate will remain confidential. As the Upjohn court observed, "[a]n uncertain privilege, or one which purports to be certain but results in widely varying applications by the courts, is little better than no privilege at all." 449 U.S. at 393.

The district court's approach returns corporate privilege law to its uncertain, pre-Upjohn state. Corporate counsel and employees will have to worry again about whether some future reviewing court will view each and every communication generated by the investigation as having the "primary purpose" of obtaining legal advice. The resulting uncertainty as to the scope or applicability of the attorney-client privilege will chill internal investigations by making in-house lawyers reluctant to reach out to lower-level employees for fear that one of those employees will respond with a privilege-destroying email or memo. The quality of legal advice to corporate clients will decline; for it is "only natural that [low- and mid-level] employees would have the relevant information needed by corporate

24

counsel if he is adequately to advise the [corporate] client . . . ." Id. at 391.

1. The district court's expanded Sealed Case standard led it into
additional legal errors.

The district court's misinterpretation of the Sealed Case standard--which substitutes assumptions about in-house counsel for case-specific facts--led the court into two additional legal errors, which we briefly address.

First, the district court held, in effect, that unless the face of the allegedly privileged document furnishes a "clear showing" that its "primary purpose" was to secure legal advice, the privilege claim fails--even if the privilege proponent submits credible and undisputed extrinsic evidence that the document relates to an Upjohn investigation, and even if the document itself is not facially inconsistent with that evidence.

That approach was clearly erroneous as a matter of law. There is no requirement that a document expressly request or refer to legal advice in order to qualify as privileged. "Client communications intended to keep the attorney apprised of continuing business developments, with an implied request for legal advice based thereon . . . may also be protected." Hercules, Inc. v. Exxon Corp., 434 F. Supp. 136, 144, 196 U.S.P.Q. 401 (D. Del. 1977). And courts do not confine themselves to the face of the allegedly privileged document when making privilege determinations; they also consider extrinsic evidence of the document's context. For example, in Barton v. United States District Court, 410 F.3d 1104

25

(9th Cir. 2005), the Ninth Circuit granted a writ of mandamus to prevent disclosure of interview questionnaires that potential plaintiffs had submitted to class-action lawyers. In so doing, the court weighed such extrinsic evidence as "the context of supplying information to lawyers who apparently were bringing a Paxil class action," "the ultimate representation of these four plaintiffs," and the law firm's statements about its purpose for distributing the questionnaires. Id. at 1110. Likewise, the Tenth Circuit upheld a claim of privilege based on an in-house lawyer's declaration that he "did not render business advice" when he prepared a facially business-oriented document (a list of employees slated for termination). See Motley v. Marathon Oil Co., 71 F.3d 1547, 1550-51 (10th Cir. 1995).

Second, apparently extending its harsh "clear showing" standard to matters of evidentiary foundation, the district court dismissed Attorney Lee's declaration in its entirety because he did not expressly state: "I read the Lindholm email." The district court therefore concluded that he lacked personal knowledge of what the email said or how it related to Google's Upjohn-type investigation. But the Ninth Circuit has held that personal knowledge and competence to testify may be "reasonably inferred" from an affiant's "positio[n] and the nature of [his] participation in the matters to which [he] swore." Barthelemy v. Air Lines Pilots Ass'n, 897 F.2d 999, 1018 (9th Cir. 1990); see also In re Kaypro, 218 F.3d 1070, 1075 (9th Cir. 2000). Lee declared under oath that he was one of the recipients of the Lindholm email; that he supervised Mr. Lindholm's work on the investigation

26

discussed in the email; and that that he could competently testify to the circumstances under which the email was created.⁵⁵ It was simply wrong to hold that his declaration could not be considered because he failed to say the magic words, "I read it." This requirement, too, apparently flows from the "clear showing" standard.

2. The district court erroneously denied work-product protection to
the Lindholm email.

“The work product doctrine extends beyond confidential communications between the attorney and client to ‘any document prepared in anticipation of litigation by or for the attorney.’” United States v. Bergonzi, 216 F.R.D. 487, 494- 95 (N.D. Cal. 2003) (emphasis added) (quoting In re Columbia/HCA Healthcare Billing Practices Litig., 293 F.3d 289, 304 (6th Cir. 2002)). The Lindholm email is exactly that—a document prepared in anticipation of Oracle’s lawsuit for Google attorney Ben Lee, among others. But the district court denied the Lindholm email work-product protection for the same reasons it denied attorney-client privilege protection.⁵⁶ That ruling, too, was clearly erroneous as a matter of law.

3. The district court clearly erred by deferring to the magistrate’s
erroneous legal conclusions, in violation of Rule 72(a).

The district court clearly erred by deferring to the magistrate’s erroneous legal conclusions, thereby violating the de novo standard of review mandated by

27

Federal Rule of Civil Procedure 72(a). Rule 72(a) governs objections to a magistrate's nondispositive pretrial order and states that "[t]he district judge in the case must consider timely objections and modify or set aside any part of the order that is clearly erroneous or is contrary to law."⁵⁷

Because the disjunctive term "or" connects the term "clearly erroneous" with the term "contrary to law," those terms must have different meanings. See Holly Farms Corp. v. N.L.R.B., 517 U.S. 392, 413 (1996). And they do. "Clearly erroneous" echoes the general Rule 52(a) standard for reviewing findings of fact made after bench trials, while "contrary to law" echoes the de novo standard for reviewing legal conclusions reached after bench trials. See generally Navajo Nation v. United States Forest Serv., 535 F.3d 1058, 1067 (9th Cir. 2008) (en banc). Accordingly, courts overwhelmingly hold that that, "[w]hen . . . review of a non-dispositive motion by a district judge turns on a pure question of law, that review is plenary under the 'contrary to law' branch of the Rule 72(a) standard." PowerShare, Inc. v. Syntel, Inc., 597 F.3d 10, 14-15 (1st Cir. 2010) (emphasis added). Accordingly, "[t]he Magistrate's legal conclusions are reviewed de novo to determine whether they are contrary to law." Perry v. Schwarzenegger, 268 F.R.D. 344, 348 (N.D. Cal. 2010).⁵⁸

28

Here, however, the district court erased any distinction between Rule 72(a)'s "clearly erroneous" and "contrary to law" standards by deferring not only to the magistrate's factfinding but also to her legal conclusions. Ninth Circuit law, by contrast, holds generally that the existence and scope of the attorney-client privilege is reviewed de novo. See United States v. Ruehle, 583 F.3d 600, 606 (9th Cir. 2009) (citation omitted). The district court's deference on matters of law was prejudicial to Google, because it effectively sheltered the magistrate's erroneous legal conclusions from Rule 72 review.

B. The remaining mandamus factors are satisfied as well.

1. Google has no other means, such as an appeal, to
obtain the desired relief.

The first Bauman factor is satisfied where, as here, the petitioner seeks review of a district court's order compelling production of an allegedly privileged communication. See Admiral Ins. Co., 881 F.2d at 1491.

2. Google will be damaged or prejudiced in ways not correctable on
appeal.

The second Bauman factor also is satisfied. Mandamus is proper to prevent the irreparable harm likely to result from the erroneously compelled disclosure of privileged communications. Admiral Ins., 881 F.2d at 1491.

29

Oracle may argue that Google cannot satisfy this factor because it already has complied with the district court's order to produce the documents (albeit stamped with large privilege legends,⁵⁹ and subject to a written reservation of clawback rights). That argument fails. If the Lindholm email is stripped of privilege, it may be shown to the jury. In that event, Oracle intends to use the document to suggest that Mr. Lindholm conducted an infringement analysis and concluded that Google was in the wrong.⁶⁰ Although that is untrue,⁶¹ the email cannot be properly explained without disclosing the nature of Google's internal investigation and of Mr. Lindholm's role in it. Google therefore will be confronted with a choice of failing to explain a facially prejudicial document, or explaining it and thereby risking a broad subject-matter waiver that never can be undone--even if the case is retried after an appeal.⁶²

VI. Conclusion

This Court should direct the district court to grant the relief requested in Part I, above.

30

Dated: November 4, 2011

Respectfully submitted,

KEKER & VAN NEST LLP

By: [signature]
ROBERT A. VAN NEST

Attorneys for Petitioner
GOOGLE INC.

_________________
¹ "A" refers to the Appendix hereto.

² 449 U.S. 383 (1981).

³ Unless otherwise indicated, actions and statements attributed to "the district court" include those of the magistrate judge. See Burlington N. & Santa Fe Ry. Co. v. United States Dist. Ct., 408 F.3d 1142, 1146 n.1(9th Cir. 2005).

⁴ 737 F.2d 94 (D.C. Cir. 1984).

⁵ A14 at n.4.

⁶ Google does not submit those documents in camera here, because the district court’s unsealed, publicly filed order quotes the final Lindhom email in full.

⁷ See A19; A22; A26; A30; A33.

⁸ A20at¶5.

⁹ A20at¶5.

¹⁰ A20at¶5.

¹¹ Complaint [Dkt. No. 1].

¹² A20at¶6;A23at¶5.

¹³ A20at¶7.

¹⁴ A20at¶7;A23at¶6.

¹⁵ A20at¶8;A23at¶7;A31-A32at¶7-8.

¹⁶ A21at¶9;A24at¶8.

¹⁷ A21at¶9;A24at¶9.

¹⁸ A25 at ¶ 14.

¹⁹ A21at¶9.

²⁰ A31-A32 at ¶ 7.

²¹ A31-A32 at ¶¶ 7-8.

²² Dkt. No. 1.

²³ A20-A21 at ¶¶7-10; ¶¶ A23-A24 7-8.

²⁴ A ninth draft of the email was listed on Google’s privilege log as entry 5512.

²⁵ A5; A37; see also A24 at ¶ 11.

²⁶ A29 at ¶13.

²⁷ A52:1-4.

²⁸ A71.

²⁹ A73.

³⁰ A87:15-21; A88:23–A89:19.

³¹ A89:9-12. ³² A39:18-21.

³³ A40.

³⁴ A40.

³⁵ Producing documents under compulsion does not waive privilege if reasonable efforts are taken to protect the privilege. See Gomez v. Vernon, 255 F.3d 1118, 1131-32 (9th Cir. 2001); Transamerica Computer Co. v. IBM Corp., 573 F.2d 646, 650-51 (9th Cir. 1978). Here, Google has pursued every means available to protect the privilege, including filing this petition. The Supreme Court recognizes that seeking mandamus is a viable alternative to the strategy of defying a discovery order and later appealing from a resulting order imposing sanctions or finding contempt. See Mohawk Indus., Inc. v. Carpenter, 130 S. Ct. 599, 607-08 (2009).

³⁶ A14.

³⁷ A14 at n.4.

³⁸ Because this petition does not involve substantive patent-law issues, this Court applies the laws of the circuit in which the district court sits. See In re Nintendo Co., Ltd., 589 F.3d 1194, 1197, 93 U.S.P.Q.2d 1152 (Fed. Cir. 2009).

³⁹ See Bauman v. United States Dist. Ct, 557 F.2d 650 (9th Cir.1977).

⁴⁰ A39:18-21.

⁴¹ A42:9-18; A14:4-8.

⁴² Google does not rely on Chen’s additional holding that, “[i]f a person hires a lawyer for advice, there is a rebuttable presumption that the lawyer is hired ‘as such’ to give ‘legal advice[.]’” Chen, 99 F.3d at 1501. Some district-court cases interpret the Chen presumption as limited to the retention of outside counsel. See, e.g., Lenz v. Universal Music Corp., C 07-3783 JF (RS), 2009 WL 3573990, at *2 (N.D. Cal. Oct. 30, 2009). The presumption aside, Chen bars reliance on an artificial “legal/business” distinction when determining whether the lawyer was acting “as such” when communicating with a client.

⁴³ Accordingly, documents setting forth “economic or business data” may be privileged if they were created “for the purpose of receiving legal advice” and were “intended to be confidential.” In re Brand Names Prescription Drugs Litig., No. 94 C 897, 1995 WL 557412, at *2 (N.D. Ill. Sept. 19, 1995). It is wrong to suppose that “economic or business information sent to an attorney can never fall under the protection of the attorney-client privilege.” Id. “In the context of the attorney-client privilege, documents prepared for the purpose of obtaining or rendering legal advice are protected even though the documents also reflect or include business issues.” In re OM Secs. Litig., 226 F.R.D. 579, 587 (N.D. Ohio 2005); see also Potter v. United States, No. 02-CV-0632-H (POR), 2002 WL 31409613, at *4 (S.D. Cal. July 26, 2002); Segerstrom v. United States, No. C 00- 0833 SI, 2001 WL 283805, at **2-5 (N.D. Cal. Feb. 6, 2001).

⁴⁴ See also Hemstreet v. Spiegel, Inc., 851 F.2d 348, 349 (Fed. Cir. 1988); Therasense, Inc. v. Becton, Dickinson & Co., C 04-02123 WHA, 2008 WL 2323856 (N.D. Cal. May 22, 2008) (Alsup, J).

⁴⁵ A39:18-21.

⁴⁶ United States v. ChevronTexaco Corp., 241 F. Supp. 2d 1065 (N.D. Cal. 2002).

⁴⁷ A42.

⁴⁸ A42 (emphases in original; citations and brackets omitted).

⁴⁹ A42.

⁵⁰ See Part V.A.4., below.

⁵¹ A15:2-3.

⁵² Outside the D.C. Circuit, the Sealed Case standard has languished. No other federal appellate court has adopted it, and only a handful of district courts outside the D.C. Circuit have applied it.

⁵³ See also United States v. KPMG LLP, 237 F. Supp. 2d 35, 47 (D.D.C. 2002) (applying Boca presumption); Wessel v. City of Albuquerque, No. MISC 00–00532 ESH/AK, 2000 WL 1803818, at *5 (D.D.C. Nov. 30, 2000) (same).

⁵⁴ The district court made much of the fact that the Lindholm email indicates that Lindholm was performing some tasks requested by Google’s founders, Larry Page and Sergey Brin, rather than by its general counsel. The court appears to have believed that a legal investigation can be ordered and supervised either by top management or by the general counsel—but never by both. That assumption is incorrect and contradicted by the undisputed evidence of record. Leading cases on this subject involve facts where both top management and corporate counsel order and supervise the investigation—e.g., where “senior management” instructed managers to give statements to the corporation’s counsel (see Admiral Ins., 881 F.2d at 1493) or a board chairman asked general counsel to investigate corporate conduct by interviewing lower-level employees (see Upjohn, 449 U.S. at 387). And it stands to reason that top management would get involved in an investigation of multibillion-dollar claims by a large corporate adversary. Indeed, Mr. Lindholm’s declaration confirms that Google’s “top management” attended the July 30, 2010 meeting where Oracle's infringement claims were discussed. A23 at ¶ 6.

⁵⁵ A20-A21 at ¶¶ 4, 8-10.

⁵⁶ A43:15-:21; A17:4-8.

⁵⁷ Emphases added.

⁵⁸ See also Brownlow v. Gen. Motors Corp., CIV. A. 3:05CV-414-R, 2007 WL 2712925, at *4 (W.D. Ky. Sept. 13, 2007); Computer Econ., Inc. v. Gartner Group, Inc., 50 F. Supp. 2d 980, 983 (S.D. Cal. 1999); Milwaukee Carpenter’s Dist. Council Health Fund v. Philip Morris, Inc., 70 F. Supp. 2d 888, 892 (E.D. Wis. 1999); Med. Imaging Ctrs. of Am., Inc. v. Lichtenstein, 917 F. Supp. 717, 719 (S.D. Cal. 1996); 12 Wright & Miller, Federal Practice & Procedure § 3069, at 350 (2d ed. 1997) (observing that Rule 72(a)’s phrase “‘contrary to law’ indicates plenary review as to matters of law”).

⁵⁹ The district court now has ordered the legends removed. See Dkt. 596.

⁶⁰ Oracle has filed a “motion in limine,” based on the Lindholm email, to preclude Google from disputing that Lindholm investigated Java and all alternatives to Java and that Google needed a license for Java and the patents-in-suit. A91.

⁶¹ See A34 at ¶4.

⁶² Adding to the prejudice, the district court has refused to grant Google’s requests to seal transcripts and documents that reference the Lindholm email. Indeed, the court quoted the entire final email in its order and then refused to seal the order. The Supreme Court has noted that “protective orders are available to limit the spillover effects of disclosing sensitive information.” Mohawk Indus., 130 S. Ct. at 608.

31

CERTIFICATE OF SERVICE
[see PDF]

Oracle's response:



Miscellaneous No. 106

____________________

United States Court of Appeals

for the

Federal Circuit

------------------

IN RE GOOGLE INC.,

Petitioner

___________________________________________
___________________________________________

On Petition for Writ of Mandamus to the United States District Court,
Northern District of California in case no. 10-CV-3561,
Judge William H. Alsup

____________________________________________
____________________________________________

RESPONSE IN OPPOSITION TO PETITION
FOR MANDAMUS

David Boies
BOIES, SCHILLER & FLEXNER LLP
[address, phone]

Steve C. Holtzman
Fred Norton
BOIES, SCHILLER & FLEXNER LLP
[address, phone]

Michael A. Jacobs
Kenneth A. Kuwayti
Marc D. Peters
Daniel P. Muino
MORRISON & FOERSTER LLP
[address, phone]

Counsel for Respondent

NOVEMBER 28, 2011

---



Miscellaneous No. 106

____________________

United States Court of Appeals

for the

Federal Circuit

------------------

IN RE GOOGLE INC.,

Petitioner

___________________________________________
___________________________________________

On Petition for Writ of Mandamus to the United States District Court,
Northern District of California in case no. 10-CV-3561,
Judge William H. Alsup

____________________________________________
____________________________________________

RESPONSE IN OPPOSITION TO PETITION
FOR MANDAMUS

David Boies
BOIES, SCHILLER & FLEXNER LLP
[address, phone]

Steve C. Holtzman
Fred Norton
BOIES, SCHILLER & FLEXNER LLP
[address, phone]

Michael A. Jacobs
Kenneth A. Kuwayti
Marc D. Peters
Daniel P. Muino
MORRISON & FOERSTER LLP
[address, phone]

Counsel for Respondent

NOVEMBER 28, 2011

---

Form 9

FORM 9. Certificate of Interest

UNITED STATES COURT OF APPEALS FOR THE FEDERAL CIRCUIT

In re Google Inc. v.____________

Misc. No. 106

CERTIFICATE OF INTEREST

Counsel for the (petitioner)(appellant)(respondent)(appellee)(amicus)(name of party) Oracle America, Inc. certifies the following (use "None" if applicable; use extra sheets if necessary):

1. The full name of every pary or amicus represented by me is:

Oracle America

2. The name of the real party in interest (if the party named in the caption is not the real party of interest) represented by me is:

Oracle America, Inc.

3. All parent corporations and any publicly held companies that own 10 percent or more of the stock of the party or amicus curiae represented by me are:

Oracle Corporation

4. X The names of all law firms and the partners or associates that appeared for the party or amicus now represented by me in the trial court or agency or are expected to appear in this court are:

Please see Attachment A

11-28-2011
Date

[signature]
Signature of Counsel

Fred Norton
Printed name of counsel

Please Note: All questions must be answered
cc:__________________

---

In re Google Inc., Petitioner
Misc. No. 106
Attachment A to Form 9 - Certificate of Interest

4. The names of all law firms and the partners or associates that appeared for the party or amicus now represented by me in the trial court or agency or are expected to appear in this court are:

BOIES, SCHILLER AND FLEXNER LLP

David Boies, W. Fred Norton, Jr., Steven C. Holtzmann, Alanna Rutherford, Beko O. R. Reblitz-Richardson, Meredith R. Dearborn

MORRISON & FOERSTER LLP

Michael A. Jacobs, Marc D. Peters, Daniel P. Muino, Kenneth A. Kuwayti, Mark E. Ungerman, Rudolph Kim, Yuka Teraguchi, Roman A. Swoopes, Benjamin A. Petersen, Ruchika Agrawal, Richard S. Ballinger [terminated]

ORACLE CORPORATION

Dorian E. Daley, Matthew M. Sarboraria, Deborah K. Miller

Attachment A to Form 9 - Certificate of Interest

ii



TABLE OF CONTENTS

Page

Certificate of Interest .....................................................i

Table of Authorities .................................................... v

I. Introduction.......................................... 1

II. Statement of Facts................................ 5

A. The text of the Lindholm email proves it is a business
communication, not a request for legal advice ..................... 6

B. Google did not rebut the facts established by the text of the
email .............................. 9

C. Google's own lawyers understood that the email concerned
business negotiations and an investigation requested by
business executives -- and said so to the Court ................... 12

D. The Magistrate Judge and District Court properly compelled
production of the email after a thorough review of the evidence ....... 13

III. Argument ..................................... 15

A. Google cannot establish the required "clear and indisputable
right to relief" because the orders of the Magistrate Judge and
District Court are correct under Supreme Court and Ninth
Circuit precedent ............................ 16

1. Google did not prove that the Lindholm email is
privileged under Upjohn or Ninth Circuit law ........... 16

2. The Magistrate Judge's and District Court's reliance on
the Sealed Case's "clear showing" rule was entirely
proper under Supreme Court, Ninth Circuit, and this
Court's precedents....................... 20

3. The Magistrate Judge properly considered all of the
evidence in making her factual findings and committed
no error in her evidentiary ruling ................... 24

iii



4. The Magistrate Judge and the District Court properly
rejected Google's claim of work-product protection ................ 26

B. The District Court correctly applied Ninth Circuit law and
Federal Rule of Civil Procedure 72(a) in rejecting Google's
objections........................... 27

C. Google cannot establish a "clear and indisputable right" to the
writ because the Magistrate Judge and the District Court would
not have reached a different result if a different legal standard
applied ............................. 28

D. Google has failed to show that it cannot obtain the same relief
by other means, or that the alleged errors cannot be corrected on
direct appeal ........................... 28

E. This Court cannot grant Google the relief it seeks because
Google has failed to submit the complete evidentiary record on
the privilege issue and the courts below have not decided the
waiver issue ........................... 30

IV. Conclusion ........................... 30

iv



TABLE OF AUTHORITIES

Page(s)

Cases

Acumen Re Mgmt. Corp. v. Gen. Sec. Nat. Ins. Co.,
2010 WL 3260166 (S.D.N.Y. July 30, 2010)...............................23

Adams v. Gateway, Inc.,
2003 WL 23787856 (D. Utah Dec. 30, 2003) ....................................20

Admiral Insurance v. U.S. Dist. Ct.,
881 F.2d 1486 (9th Cir. 1989) .............................................19

Ali v. Douglas Cable Comm'cns, Ltd. P'ship,
890 F. Supp. 993 (D. Kan. 1995) .................................20

Allied Chem. Corp. v. Daiflon, Inc.,
449 U.S. 33 (1980)......................................15

Ames v. Black Entm't Television,
1998 WL 812051 (S.D.N.Y. Nov. 18, 1998) .................................23

Amway Corp. v. Procter & Gamble Co.,
2001 WL 1818698 (W.D. Mich. Apr. 3, 2001) ...................................... 22, 23

Argenyi v. Creighton Univ.,
2011 WL 3497489 (D. Neb. Aug. 10, 2011) ..................................23

Boca Investerings P'ship v. United States,
31 F. Supp. 2d 9 (D.D.C. 1998)...............................22

Borase v. M/A COM, Inc.,
171 F.R.D. 10 (D. Mass. 1997) ...............................23

Daubert v Merrell Dow Pharmaceuticals,
509 U.S. 570 (1993)........................................12

Griffith v. Davis,
161 F.R.D. 687 (C.D. Cal. 1995).......................... 17, 20

Grimes v. City and Cty. of San Francisco,
951 F.2d 236 (9th Cir. 1991) ....................... 14, 27

v



Hakim v. Cannon Avent Group, PLC,
479 F.3d 1313 (Fed. Cir. 2007) .......................26

Hardy v. New York News,
114 F.R.D. 633 (S.D.N.Y. 1987) .........................20

In re Cordis Corp.,
769 F.2d 733 (Fed. Cir. 1985) ..........................15

In re CV Therapeutics, Inc. Sec. Litig.,
2006 WL 2585038 (N.D. Cal. Aug. 30, 2006) ..............................18

In re Diagnostics Systems Corp.,
328 F. App'x 621 (Fed. Cir. 2008) .......................... 3, 24

In re EchoStar Comm'cns Corp.,
448 F.3d 1294 (Fed. Cir. 2009) .......................30

In re Grand Jury Subpoena (Torf),
357 F.3d 900 (9th Cir. 2004) ........................26

In re Rospatch Sec. Litig.,
1991 WL 574963 (W.D. Mich. Mar. 14, 1991) ....................................23

In re Sealed Case,
737 F.2d 94 (D.C. Cir. 1984) .......................... passim

In re Shared Memory Graphics LLC,
___F.3d ___, 2011 WL 4390020 (Fed. Cir. Sept. 22, 2011).........................15

In re Van Dusen,
654 F.3d 838 (9th Cir. 2011) .............................28

Lewis v. Pacific Maritime Ass'n,
2007 WL 2429554 (N.D. Cal. Aug. 24, 2007) ..................................25

Lindley v. Life Investors Ins. Co. of Am.,
267 F.R.D. 382 (N.D. Okla. 2010) ............................23

Marten v. Yellow Freight System, Inc.,
1998 WL 13244 (D. Kan. Jan. 6, 1998) ...........................23

vi

McDonough Power Equip., Inc. v. Greenwood,
464 U.S. 548 (1984)................................28

Mohawk Industries v. Carpenter,
130 S.Ct. 599 (2009)..................................30

Navigant Consulting, Inc. v. Wilkinson,
220 F.R.D. 467 (N.D. Tex. 2004) ........................20

Neuberger Berman Real Estate Income Fund, Inc. v. Lola Brown Trust No. 1B,
230 F.R.D. 398 (D. Md. 2005) .......................23

O'Shea v. Epson Am., Inc.,
2010 WL 2305863 (C.D. Cal. June 4, 2010) ....................................23

Peat, Marwick, Mitchell & Co. v. W.,
748 F.2d 540 (10th Cir. 1984), cert. denied, 469 U.S. 1199 (1985) .........................................20

Pizza Mgmt., Inc. v. Pizza Hut, Inc.,
1989 WL 9334 (D. Kan. Jan. 10, 1989) .....................................23

SCM Corp. v. Xerox Corp.,
70 F.R.D. 508 (D. Conn. 1976) ........................19

Teltron, Inc. v. Alexander,
132 F.R.D. 394 (E.D. Pa. 1990) .........................23

Truckstop.net, LLC v. Spring Corp.,
547 F.3d 1065 (9th Cir. 2008) ............................29

U.S. ex rel. Parikh v. Premera Blue Cross,
2006 WL 3733783 (W.D. Wash. Dec. 15, 2006) ............................23

United States v. Chen,
99 F.3d 1495 (9th Cir. 1996) .......................... 18, 19, 21

United States v. Chevron Corp.,
1996 WL 264769 (N.D. Cal. Mar. 13, 1996) ................................. 22, 23

United States v. ChevronTexaco Corp.,
241 F. Supp. 2d 1065 (N.D. Cal. 2002) ........................ 22, 23

vii



United States v. Richey,
632 F.3d 559 (9th Cir. 2011) ...................................17

United States v. Rowe,
96 F.3d 1294 (9th Cir. 1996) ......................... 17, 21

United States v. Ruehle,
583 F.3d 600 (9th Cir. 2009) ...........................23

Upjohn Co. v. United States,
449 U.S. 383 (1980).............................. passim

Ward v. First Federal Savings Bank,
173 F.3d 611 (7th Cir. 1999) ............................25

Statutes

Fed. R. App. P. 21(a)(2)(C) ...............................5

Fed. R. Civ. P. 72(a)........................................ passim

Fed. R. Evid. 602 ....................................................25

Other Authorities

Kenneth S. Broun et al., 1 MCCORMICK ON EVIDENCE § 10 .......................25

Paul R. Rice, 1 ATTORNEY-CLIENT PRIVILEGE IN THE UNITED STATES,
§ 7:1 (2d ed. 1999) ................................ 20

viii



I. Introduction

Google's petition for mandamus arises from its repeated, unsuccessful efforts to conceal an email that Google engineer Tim Lindholm wrote to Google business executive (and non-lawyer) Andy Rubin in August 2010, shortly before this lawsuit was filed.

In that email, Mr. Lindholm told Mr. Rubin that he had been asked by Google founders (and non-lawyers) Larry Page and Sergey Brin to "investigate what technical alternatives exist to Java" for Google's Android and Chrome. Mr. Lindholm reported to Mr. Rubin that the "technical alternatives" to Java "all suck" and that consequently "we need to negotiate a license for Java under the terms we need." Mr. Lindholm also recommended that, as a negotiation strategy, Google should make the "threat" of using the "most credible alternative" in order to obtain a Java license from Oracle at "better terms and price," while asking Mr. Rubin whether this strategy was a "nonstarter for negotiation purposes."

After careful review of the email itself, as well as all of the other evidence submitted by the parties, the Magistrate Judge below found that "the contents of the email itself severely undermine" Google's claim of privilege. As the Magistrate Judge also found, nowhere does the email text mention lawyers or legal advice, nor does it seek input from lawyers or legal advice. Rather, the facts established by the email indicate that it is "a business discussion," not a "proffering

1

of research for an attorney preparing legal advice." The Magistrate Judge further found that "[n]othing in the content of the Email indicates that Lindholm prepared it in anticipation of litigation or to further the provision of legal advice." The Magistrate Judge gave Google leave to file as many declarations as it wanted, but she found, and the District Court affirmed, that the declarations Google offered suffered from numerous gaps that failed to rebut the plain meaning of the email itself. Accordingly, the Magistrate Judge found, and the District Court affirmed, that Google failed to prove that the engineer's email to the business executive was a privileged communication or attorney work product.

Google now seeks to evade those factual findings through this petition for mandamus. The petition should be denied for at least five reasons.

First, the Magistrate Judge's and District Court's orders do not "undermine" Upjohn Co. v. United States, as Google now contends. Rather, the Magistrate Judge reviewed all of the evidence, including the email itself, and made the factual finding that Google "failed to meet its burden of showing that the Lindholm email was generated in a privileged scenario, whether of the type described in Upjohn or otherwise." Although the Magistrate Judge specifically found that the email's contents "severely undermine" the claim of privilege, Google never addressed those contents in any of its briefs or declarations, and still has not done so now.

2

 Second, Google has failed to identify any error, much less the "clear error" it must establish to win mandamus relief. There was no error in requiring Google to make a "clear showing" that the Lindholm email was a privileged communication. That is the rule of now-Justice Ginsburg's opinion for the D.C. Circuit in In re Sealed Case, 737 F.2d 94 (D.C. Cir. 1984) -- a rule that has never been criticized, much less rejected, by any federal Court of Appeals; that has been applied repeatedly by district courts in nine circuits, including the Ninth Circuit; and that has been acknowledged by this Court in rejecting a mandamus petition arising out of the Ninth Circuit. In re Diagnostics Systems Corp., 328 F. App'x 621, 622-23 (Fed. Cir. 2008). That rule does not rest on an "unwarranted distrust" of in-house counsel, as Google protests; it rests on the practical reality that in-house lawyers are frequently included on business communications that do not seek or provide legal advice. To defy such a rule would be clear error; to adhere to it is not.

There also was no error in the District Court's review of the Magistrate Judge's order. The District Court repeatedly quoted and applied the Rule 72(a) standard -- "clearly erroneous or contrary to law" -- throughout its analysis. Although Google's statement of issues asserts that the District Court "deferred to the legal conclusions of the magistrate judge," nowhere in its brief does Google identify any instance where the Court actually did so. In fact, the District Court

3

expressly noted that a "non-dispositive order entered by a magistrate must be deferred to unless it is 'clearly erroneous or contrary to law.'" (emphasis added)

Third, even if there were some error -- and there is none -- Google has not established, as it must, that it is "clear and indisputable" that mandamus should issue. Specifically, there is no basis to conclude that, under some standard other than the "clear showing" rule, the Magistrate Judge or the District Court would have been compelled to find the Lindholm email privileged on the full evidentiary record presented below. To the contrary, the District Court found that even if the "clear showing" rule did not apply, and Google were entitled to a presumption that in-house counsel are most often giving legal advice, the evidence cited by the Magistrate Judge would rebut that presumption. Google still would lose.

Fourth, mandamus is not appropriate when disclosure has already occurred. The Magistrate Judge ordered Google to produce the document, and Google complied. The document's full text has been published in a public order, available on Westlaw, on the internet, and in the press. Google's petition would not prevent disclosure; it would only prevent Oracle from using the document as one of many pieces of evidence at trial. That is not a proper basis for mandamus relief.

Fifth, Google did not give this Court the full record necessary to resolve the privilege issue, nor has any court yet reached Oracle's waiver argument. Thus, this

4

Court cannot grant Google the relief it seeks: to hold that the Lindholm email is privileged, compel its return, and seal all court documents "referencing" it.

Google's petition for a writ of mandamus should be denied.

II. Statement of Facts

Google's petition challenges a non-dispositive order of a Magistrate Judge, which it concedes is subject to review under the "clearly erroneous or contrary to law" standard of Federal Rule of Civil Procedure 72(a). (Petition, 28) The Magistrate Judge made numerous, specific findings of fact adverse to Google, which were all affirmed by the District Court, and which Google does not directly challenge. Nonetheless, Google ignores those adverse findings, and instead advances a self-serving and incomplete recitation of the facts that was rejected on the record established below. At the same time, Google has submitted an appendix that omits one of its own declarations; includes only a few lines of the transcript of the hearing before the Magistrate Judge (A84-A90); includes a declaration Google submitted on a different motion after the Magistrate Judge had ruled (A33-A45); and omits all of the evidence that Oracle submitted and was considered in the proceedings below. The petition thus ignores the proper standard of review and fails to comply with Federal Rule of Appellate Procedure 21(a)(2)(C). Without assuming Google's burden to provide this Court with the necessary record, Oracle submits the following statement of facts and attached supplemental appendix.

5

A. The text of the Lindholm email proves it is a business communication,
not a request for legal advice.

On July 20, 2010, Oracle lawyers made a presentation to Google, asserting that Google's Android smartphone platform infringed specific patents related to Java. No one mentioned Chrome. (A173, ¶¶3-5) About two weeks later, on August 6, 2010, Google engineer Tim Lindholm sent an email to Google Vice President Andy Rubin, a non-lawyer in charge of Android. Mr. Lindholm had been a Sun Microsystems¹ engineer working on Java until July 2005, when he joined Google. At Google, Mr. Lindholm immediately began advising Mr. Rubin about Android, negotiating unsuccessfully with his former Sun colleagues for a Java license, and evaluating alternatives to Java for use in Android. (A132, ¶ 11; A144-A162) On behalf of himself and another Google engineer, Dan Grove, this is what Mr. Lindholm wrote in his August 6, 2010 email:

Hi Andy,

This is a short pre-read for the call at 12:30. In Dan's earlier email we didn't give you a lot of context, looking for the visceral reaction that we got.

What we've actually been asked to do (by Larry and Sergei) is to investigate what technical alternatives exist to Java for Android and Chrome. We've been over a bunch of these, and think they all suck. We conclude that we need to negotiate a license for Java under the terms we need.

6

 That said, Alan Eustace said that the threat of moving off Java hit Safra Katz hard. We think there is value in the negotiation to put forward our most credible alternative, the goal being to get better terms and price for Java.

It looks to us that Obj-C provides the most credible alternative in this context, which should not be confused with us thinking we should make the change. What we're looking for from you is the reasons why you hate this idea, whether you think it's a nonstarter for negotiation purposes, and whether you think there's anything we've missed in our understanding of the option.

-- Tim and Dan

(A6:1-15) After he wrote the entire email, at literally the last moment, Mr. Lindholm added the words "Google Confidential" and "Attorney Work Product" at the top of the page, and added Ben Lee, a Google in-house lawyer, to the email's addressees. (A17-18; A197-A207)

As the Magistrate Judge found, the text of the email makes no reference, express or implied, to litigation, infringement, lawyers, or legal advice. (A41:3-5) Three of the four paragraphs refer to negotiation, and in particular, Google's business strategy in negotiating for a Java license. The facts considered by Mr. Lindholm are in no way legal or even inputs to providing legal advice, as the Magistrate Judge found. (A40:21-22; A207)

Indeed, as the Magistrate Judge noted, every substantive sentence of the email refers to a business or technical consideration, never a legal one. (A37:8-17; A40:21-A41:14) Google founders "Larry and Sergei," not any lawyer, gave Mr. Lindholm the task described in the email, which was to "to investigate what

7

technical alternatives exist to Java for Android and Chrome," not evaluate claims of infringement or assist with legal analysis. His conclusion that the technical alternatives "all suck" was that of an engineer assessing business options; his conclusion that "we need to negotiate a license for Java under the terms we need" recognized the business consequences of having no viable alternatives to Java. Mr. Lindholm's repetition of a statement by Alan Eustace (a non-lawyer Google executive) that "the threat of moving off Java hit [Oracle President] Safra Katz hard" focused on the business executives involved in the negotiation, not any lawyer, and the leverage of threatening to walk away from a business negotiation. Mr. Lindholm recommended that Google put forward its "most credible alternative" to "get better terms and price for Java," not to obtain any litigation advantage. He then proposed that the "most credible alternative" to threaten was a move to Objective-C instead of Java. Finally, he asked Mr. Rubin to comment on that strategy, including whether Mr. Rubin believed that the tactic was "a nonstarter for negotiation purposes."

The Magistrate Judge expressly relied on what Mr. Lindholm had written in finding that "the Email appears to be a strategy discussion intended to address business negotiations regarding a Java license." (A41:12-13) "Nothing in the content of the Email indicates that Lindholm prepared it in anticipation of litigation or to further the provision of legal advice." (A40:21-22) The email text appears to

8

be "a business discussion," not a "proffering of research for an attorney preparing legal advice." (A41:3-7) Accordingly, the Magistrate Judge found that that "the contents of the email itself severely undermine the claim that Lindholm generated this particular email as part of an attorney-directed effort to provide legal advice or prepare for litigation." (A40:5-7 (emphasis added)) The District Court in turn affirmed these factual determinations as "true," and rejected Google's assertions that they were clearly erroneous. (A10:15-26)

B. Google did not rebut the facts established by the text of the email.

Just as it did in the proceedings below, Google ignores every single one of the statements Mr. Lindholm wrote in his email.

In support of its privilege claim, Google submitted declarations from Mr. Lindholm and former Google in-house lawyer Ben Lee, none of which makes any reference to the words, facts, or persons described in the email. The declarations from Mr. Lindholm and Mr. Lee assert, in conclusory fashion, that Mr. Lindholm was reporting on investigations that he and Dan Grove (another Google engineer) conducted "at the request of Google General Counsel Kent Walker, under the supervision of Mr. Lee, and in anticipation of Oracle's threatened lawsuit." (A25

9

at ¶ 14) Google incorrectly suggests that the Magistrate Judge found those declarations credible.² (Petition, 10)

In its briefing below and before this Court, Google limits its discussion of the evidence to events around the time of the email, ignoring the specifics of the document itself. Google emphasizes that Oracle had threatened suit over Android's infringement of Java patents, that Mr. Lindholm was gathering "information for Google's lawyers and management to consider in evaluating technology issues related to Oracle's infringement claims" (A23, at ¶7), and that the email concerns Java and Android, the technology at issue in those claims. Google says these facts establish that it was engaged in "an Upjohn-type internal investigation." But Google did not persuade the Magistrate Judge or the District Court that the email Mr. Lindholm actually wrote on August 6, 2010, was in furtherance of such an investigation, as opposed to what the words of the email reveal it to be: "a strategy discussion intended to address business negotiations regarding a Java license." (A41:12-13)

10

 In particular, the Magistrate Judge found that "there are many basic gaps in the factual record that Google failed to fill, despite having had ample opportunity to do so." (A40:7-8) For example, the Magistrate Judge made the following findings, each of which was expressly affirmed by the District Court:

• Google had "no rejoinder" to the "central facts" that the email is directed to Mr. Rubin, not Mr. Walker or Mr. Lee, or any other lawyer; and that it expressly states that Mr. Page and Mr. Brin, not a lawyer, instructed Lindholm to investigate technical alternatives to Java. (A40:21-41:2; affirmed, A9:17-10:9)

• The Lindholm and Lee declarations failed to foreclose "the reasonable explanation" that in August 2010 they were also communicating about non- privileged matters, "including the business of negotiating for a Java license." (A40:11-16; affirmed, A12:1-13:4)

• Neither Mr. Lee nor Mr. Lindholm asserted that Mr. Rubin, Mr. Page or Mr. Brin was involved in the supposed efforts to formulate legal advice, or explained "why these individuals feature so prominently in the email." (A40:22-25; affirmed, A13:5-14:2)

• The declaration of Mr. Lee, a former Google employee, "did not indicate that he reviewed the Email and could competently represent it was connected to work he requested from Lindholm as part of the provision of legal advice." (A40:8-11; affirmed, A11:9-28)

• The email refers to investigation of alternatives to Java for Google's Chrome product, not just Android, despite the fact that Chrome was not mentioned at the July 20, 2010, meeting and has played no role in this litigation. (A41:7-10; affirmed, A10:25-28; A9:12-15)

In an effort to refute the plain meaning of the email, Google asserts that "Mr. Lindholm also declares that his email did not contain 'general business advice about Android.'"³ Even assuming Mr. Lindholm could credibly claim his email

11

contained no business advice ≠ and he cannot ≠ his conclusory statement does not preclude the finding that he gave Mr. Rubin specific advice about "technical alternatives to Java" and business strategies for negotiating for a Java license. The District Court specifically considered, and rejected, Google's contention that Mr. Lindholm had foreclosed any non-privileged purpose for his email. (A12:24-13:2)

C. Google's own lawyers understood that the email concerned business
negotiations and an investigation requested by business executives– and
said so to the Court.

Before Google asserted privilege, the substance of the email was discussed at two hearings in this case held on July 21, 2011: first, at a telephonic discovery hearing before the Magistrate Judge in which Oracle sought to compel Mr. Lindholm's deposition, and then at a Daubert hearing before the District Court later that same day. At the discovery hearing, Google was represented by four outside lawyers and one in-house counsel. (A135, ∂ 18) At the Daubert hearing, Google again was represented by three of those same outside lawyers, and two in- house lawyers, including the in-house lawyer who had appeared at the discovery hearing. (A137, ¶ 27).

At each of the hearings, before discussing the document's contents, Oracle's counsel indicated that Google had designated the document as "Attorneys' Eyes Only." (A105:5-13; A112:19-A113:18; A135-A138, ¶¶ 17-37) Google claims that by using the documents, Oracle "willfully violated the protective order." (Petition, 9) That claim is false. Both the District Court and the Magistrate Judge rejected it

12

each time Google raised it. No Google lawyer at the hearing objected to the email's disclosure; no Google lawyer asserted that the document was privileged. To the contrary, Google's trial counsel argued the substance of the email. Google's counsel argued that the email did not show willful infringement, but concerned "negotiations" and that Mr. Lindholm was responding to a "question from the CEO." (A121:18-A122:10; A138:5-18)⁴ The in-court discussion of the email was widely reported in the press, and the District Court published much of the email in a July 22 order, before Google claimed privilege. (A141, ¶¶ 46-47)

D. The Magistrate Judge and District Court properly compelled
production of the email after a thorough review of the evidence.

The day after the hearing, Google clawed back all versions of the Lindholm email. Oracle complied with the clawback demand, and promptly moved to compel the document's production on the grounds that the email was not privileged, and, alternatively, that Google had waived any otherwise applicable privilege by deliberately producing the Lindholm email, failing to object to its use at two separate court hearings, and arguing the email's substance at those two hearings. (A:73-A78; A130-A143) The Magistrate Judge directed Google to submit the email for in camera review, and allowed both parties to submit as many declarations as they wished, as well as additional declarations in response. (A126)

13

 After argument and consideration of all the evidence, the Magistrate Judge held that the email was not privileged and was not subject to the work product doctrine, making the waiver issue moot. (A36-A44). Google was ordered to produce the email back to Oracle, and it immediately did so.

Google then objected to the Magistrate Judge's order, and the District Court allowed full briefing and yet another oral argument. In overruling Google's objections, the District Court cited Ninth Circuit law that a "non-dispositive order entered by a magistrate must be deferred to unless it is 'clearly erroneous or contrary to law.'" (A8:4-5 (quoting Grimes v. City and Cty. of San Francisco, 951 F.2d 236, 241 (9th Cir. 1991) (emphasis added)). The Court then carefully considered each of Google's objections, and affirmed the order in its entirety.

The District Court specifically rejected Google's argument that the "clear showing" rule Sealed Case is contrary to the law of the Ninth Circuit, holding that no Ninth Circuit decision had addressed the issue, but that "the rule adopted by now Justice Ginsburg . . . makes considerable sense and addresses real world practices." (A14:9-A15:3 & n.4) The District Court then considered Google's argument that, under the "clear showing" rule, there is a "rebuttable presumption that in-house counsel are most often giving legal advice" when the attorney is in the legal department or general counsel's office. The District Court found that such a presumption would not make any difference, as the Magistrate Judge's order "cited sufficient record evidence to rebut any such presumption." (A15:4-7)

14

 The District Court also considered and rejected Google's argument that the Magistrate Judge's order was contrary to Upjohn. The Court acknowledged the need for corporate counsel to communicate freely with employees to provide legal advice, but noted that "the privilege protecting such communications must be shown to apply before it can be invoked as to a particular document." (A16:12-15) In this case, Google "failed to meet its burden of showing that the Lindholm email was generated in a privileged scenario, whether of the type described in Upjohn or otherwise." (A16:16-17 (emphasis in original))

III. Argument

Google's burden is a heavy one. Mandamus is a "drastic" remedy, "to be invoked only in extraordinary circumstances." Allied Chem. Corp. v. Daiflon, Inc., 449 U.S. 33, 34 (1980). Google must show that there is no other way to obtain the relief it seeks and that its right to issuance of a writ is "clear and indisputable." Id. at 35. Google cannot meet that burden if a "rational and substantial legal argument can be made in support of the rule in question." In re Cordis Corp., 769 F.2d 733, 737 (Fed. Cir. 1985).⁵ This burden is particularly heavy where, as here, the challenged order rests on factual determinations that may be set aside – even on direct appeal – only on a showing of clear error. See FED. R. CIV. P. 72(a).

15

A. Google cannot establish the required "clear and indisputable right to
relief" because the orders of the Magistrate Judge and District Court
are correct under Supreme Court and Ninth Circuit precedent.

Google argues that the Magistrate Judge's and the District Court's orders are "clearly erroneous as a matter of law" because their application of the clear showing rule of Sealed Case would undermine Upjohn Co. v. United States, 449 U.S. 383 (1980) and Ninth Circuit precedent. Google is wrong.

1. Google did not prove that the Lindholm email is privileged under
Upjohn or Ninth Circuit law.

Google's argument centers on the holding of Upjohn, and as a result, Google misses the point.⁶ The Magistrate Judge, the District Court, and Oracle all agree that a communication from an employee to an in-house lawyer for the purpose of obtaining legal advice from the lawyer is subject to the attorney-client privilege. As the District Court found, however, Google "failed to meet its burden of showing that the Lindholm email was generated in a privileged scenario, whether of the type described in Upjohn or otherwise." (A16:16-18). Google's own cited case reinforces the point:

Prior to Upjohn, in claiming the protection of the attorney-client privilege the corporation had the burden of showing that the communication was made for the purpose of securing legal advice. Where the attorney was asked for business (as opposed to legal) counsel, no privilege attached. Upjohn did not eliminate this distinction.

16

United States v. Rowe, 96 F.3d 1294, 1297 (9th Cir. 1996) (internal citations and alterations omitted; emphasis in original). Google cannot meet its burden of establishing attorney-client privilege by simply claiming that there was an "Upjohn-type internal investigation." It must show that the claimed privileged communication was made primarily for the purpose of obtaining legal advice ≠ a showing that it simply could not make for this email, for the reasons detailed in the Magistrate Judge's and the District Court's orders. (A8:12-A10:12; A16:9-20; A39:15-44:2) See United States v. Richey, 632 F.3d 559, 566 (9th Cir. 2011) (setting out required elements of privilege claim); Griffith v. Davis, 161 F.R.D. 687, 697 (C.D. Cal. 1995) (party asserting attorney-client privilege must prove that all of the communications it seeks to protect were made "primarily for the purpose of generating legal advice") (emphasis in original).

The cases Google cites do not help it – indeed, they only demonstrate further that Mr. Lindholm's email did not seek legal advice as courts apply that requirement. Google argues that the "controlling question is whether the lawyer was employed to give legal advice based on his `knowledge and discretion in the law,'" (Petition, 17) yet Google has never offered any evidence or argument that the actual information contained in Mr. Lindholm's email called upon Mr. Lee's

17

"knowledge and discretion in the law" in any way. Indeed, Mr. Lindholm's email asks for Mr. Rubin's thoughts, but does not ask Mr. Lee to do anything at all.⁷

In each of the cases Google cites, lawyers were giving "legal advice regarding the client's business affairs." United States v. Chen, 99 F.3d 1495, 1501 (9th Cir. 1996) (emphasis added). For example, in Upjohn, company lawyers used questionnaires to gather information from employees about payments to foreign government officials, which Upjohn suspected were illegal. Upjohn, 449 U.S. at 386-87. The magistrate judge in Upjohn had already found that the purpose of that information was to allow company lawyers to "determine the nature and extent of the questionable payments and to be in a position to give legal advice to the company with respect to the payments." Id. at 394 (emphasis in original).

In Chen, the defendants' lawyers filed disclosures with U.S. Customs that purported to correct prior customs disclosures. The government claimed that the amended disclosures were actually part of a tax evasion scheme, and sought to compel the testimony of the lawyers. 99 F.3d at 1498. The Ninth Circuit rejected the government's conclusory assertion that the lawyers were engaged in non- privileged "business decision-making" when they filed the disclosures. Rather, the

18

lawyers had been employed to use their "knowledge and discretion in the law" to "bring their clients into compliance with the law in the least burdensome way possible," such that their communications were privileged. Id. at 1501-02. In Admiral Insurance v. U.S. Dist. Ct., the company had been sued for securities fraud and other torts related to financial transactions with another entity, JNC. 881 F.2d 1486, 1488-89 (9th Cir. 1989). Company lawyers interviewed two company employees in order to "render legal advice to Admiral regarding its potential interests and liabilities arising from the JNC transactions." Id. at 1489. The Ninth Circuit held that those interviews were privileged. Id. at 1493.

Mr. Lindholm's email about "technical alternatives" and tactics to get a Java license for Android and Chrome, at "better terms and price," is nothing like the Upjohn questionnaires to assess the extent and legality of payments to foreign governments, the Chen lawyers' advice about tax and customs compliance, or the Admiral Insurance lawyers' fact gathering to assess potential liability for fraud.

Google's argument that license negotiations can be a component of litigation settlements adds nothing. "Licensing decisions may contain a legal component, but are not inherently dependent on legal advice; they are essentially business decisions." SCM Corp. v. Xerox Corp., 70 F.R.D. 508, 517 (D. Conn. 1976) (Newman, J.). Business reasons for or against a course of action "are like any other business evaluations and motivations and do not enjoy any protection

19

because they were alluded to by conscientious counsel. To protect the business components in the decisional process would be a distortion of the privilege." Id.⁸

2. The Magistrate Judge's and District Court's reliance on the
Sealed Case's "clear showing" rule was entirely proper under
Supreme Court, Ninth Circuit, and this Court's precedents.

Countless courts and commentators have made the common-sense observation that in-house counsel are often copied on communications that do not concern legal advice, or are asked to provide advice that is not legal at all. See, e.g., Paul R. Rice, 1 ATTORNEY-CLIENT PRIVILEGE IN THE UNITED STATES, § 7:1 (2d ed. 1999) (collecting cases). In recognition of that fact, now-Justice Ginsburg, writing for the D.C. Circuit, held that the party asserting privilege must make "a clear showing" that an in-house lawyer gave advice "in a professional legal capacity." Sealed Case, 737 F.2d at 99. As Google concedes (Petition, 20), the clear showing rule does not "dilute the privilege"; it concerns the standard of proof for establishing the existence of the privilege.⁹ Sealed Case is not inconsistent

20

with Upjohn; in fact, it cites Upjohn in the very paragraph that sets out the clear showing rule. Id.

Google argues that it was clear error to apply that rule to Mr. Lindholm's email. First, Google contends that the Ninth Circuit has "rejected" the premise of Sealed Case. (Petition, 16-18) The two cases Google cites say no such thing. United States v. Rowe rejected an argument that in-house lawyers advising their employer had no "client" for purposes of the attorney-client privilege, holding that inside and outside counsel were no different for that purpose. 96 F.3d 1294, 1296 (9th Cir. 1996). As the District Court held here, the Rowe court never purported to address the showing necessary to establish privilege for in-house counsel communications, and no case has ever cited Rowe as addressing that issue. (A14:17-A15:3) Chen, which is discussed above, involved communications with outside counsel, and never addressed what showing would be required for communications with inside counsel. 99 F.3d at 1498.¹¹

Google next argues that the District Court improperly expanded the clear showing rule to apply to all in-house counsel, not just in-house counsel with

21

business responsibilities. (Petition, 21-22) Recognizing that titles may not be informative and certainly are not dispositive, other courts have applied the clear showing rule to in-house counsel generally.¹¹ Google itself has successfully argued in this very case that Oracle lawyers with responsibility for litigation should not be allowed to see highly confidential discovery material because those lawyers also have responsibility for making business decisions.¹²

Google complains that it should have been given the benefit of a "rebuttable presumption that a lawyer in the legal department or working for the general counsel is most often giving legal advice." (Petition, 22). There are at least three flaws in this argument, each fatal. First, the District Court specifically held that even if this presumption applied, Google would still lose: the Magistrate Judge's order "cited sufficient record evidence to rebut any such presumption in this instance." (A15:4-9) Second, Google – the party with the burden of proof on all issues – provided no evidence whatsoever about Mr. Lee's role at Google. (A19- 21; A180 at 21:4-14) Compare Boca Investerings P'ship v. United States, 31 F. Supp. 2d 9, 12 (D.D.C. 1998) (cited by Google) (observing that in-house lawyer's place on organizational chart is an important, though not dispositive, factor in

22

assessing the lawyer's role, and quoting inside lawyer's detailed description of his responsibilities). Third, the Ninth Circuit has expressly rejected the claim that "[w]here an attorney-client relationship exists, communications made in the context of that relationship are prima facie subject to the privilege." United States v. Ruehle, 583 F.3d 600, 608 n.8 (9th Cir. 2009) (finding that this argument "directly conflicts with our case law").

Finally, Google argues that Sealed Case is an outlier that has "languished" outside the D.C. Circuit. (Petition, 22 n.52) But no Circuit Court has ever rejected, much less criticized, that holding. District courts in nine different circuits, including the Ninth, have followed it repeatedly.¹³ And this Court has refused to grant a mandamus petition arising from the Ninth Circuit, where the

23

district court rejected a privilege claim because that petitioner – like Google – "failed to make a clear showing that the primary purpose of the communication was securing legal advice[.]" See In re Diagnostics Sys. Corp., 328 F. App'x at 622-23.

3. The Magistrate Judge properly considered all of the evidence in
making her factual findings and committed no error in her
evidentiary ruling.

Google contends that the Magistrate Judge committed two additional "legal errors" as a result of her reliance on Sealed Case. Both assertions are meritless.

First, Google argues that the Magistrate Judge held "in effect" that unless the document on its face establishes the privilege, the privilege claim fails, even if there is "undisputed evidence" that the document "relates to an Upjohn investigation" and "even if the document itself is not facially inconsistent with that evidence." (Petition, 25) This nonsensical contention is directly at odds with the detailed factual findings by the Magistrate Judge. As described above, the Magistrate Judge considered all of the evidence submitted by the parties, and carefully analyzed Google's declarations. (A39:21-A42:22) As the District Court found, the Magistrate Judge's order "set forth a detailed review of the email itself and the extrinsic evidence bearing on the question of privilege" and reached a conclusion "based on holistic consideration of all the intrinsic and extrinsic evidence." (A8:13-17) In the course of that review, the Magistrate Judge

24

explicitly rejected the idea that the email was "not facially inconsistent" with the privilege claim: "the contents of the email itself severely undermine the claim that Lindholm generated this particular email as part of an attorney-directed effort to provide legal advice or prepare for litigation." (A40:5-7) (emphasis added)

Second, Google argues that the Magistrate Judge "apparently" extended the clear showing rule to the evidentiary requirement of foundation and dismissed Mr. Lee's "declaration in its entirety." (Petition, 26) But nothing in the Magistrate Judge's or the District Court's orders suggests any connection between the clear showing rule and the Court's proper insistence that Federal Rule of Evidence 602 be satisfied.

A witness may not testify to a matter absent sufficient evidence to show that he has personal knowledge of the matter. FED. R. EVID. 602. As the party offering the testimony, Google had the burden under Rule 602 of laying a foundation demonstrating that Mr. Lee "had an adequate opportunity to observe and presently recalls the observation." Kenneth S. Broun et al., 1 MCCORMICK ON EVIDENCE § 10. In establishing that predicate, "the source of the knowledge must be disclosed; it is not sufficient for a witness merely to say that he or she is aware of a fact." Lewis v. Pacific Maritime Ass'n, 2007 WL 2429554, at *5 (N.D. Cal. Aug. 24, 2007) (citing Ward v. First Federal Savings Bank, 173 F.3d 611, 617-18 (7th Cir. 1999)). The Magistrate Judge credited parts of the Lee declaration, but correctly

25

found that Mr. Lee "did not indicate that he reviewed the Email and could competently represent that it was connected to work that he requested from Lindholm as part of the provision of legal advice he describes in his declaration"; consequently Google failed to provide a sufficient foundation to permit Mr. Lee to testify about it. (A11:16-28; A40:9-11). Google has not shown that this evidentiary ruling was clearly erroneous. See Hakim v. Cannon Avent Group, PLC, 479 F.3d 1313, 1320 (Fed. Cir. 2007).

4. The Magistrate Judge and the District Court properly rejected
Google's claim of work-product protection.

Under Ninth Circuit law, to establish work-product protection, Google must show that the "document was created because of anticipated litigation, and would not have been created in substantially similar form but for the prospect of that litigation." In re Grand Jury Subpoena (Torf), 357 F.3d 900, 908 (9th Cir. 2004). Google failed to make that showing. In its argument to the Magistrate Judge, Google devoted only two sentences to its work product claim. The Magistrate Judge found that Google waived any dual-purpose argument, which Google did not challenge. (A43:15-21 & n.6) Thus, Google can establish work product protection only if the email was created exclusively in preparation for litigation, a showing that Google has never tried to make, and which is foreclosed by the content of the email itself. Indeed, the investigation that Mr. Lindholm describes in his email is

26

much like the Android-related work he did in 2005 and 2006. (A132-34 ¶¶ 10-11; A144-A162)

B. The District Court correctly applied Ninth Circuit law and Federal Rule
of Civil Procedure 72(a) in rejecting Google's objections.

Google claims the District Court committed reversible error by "deferring" to the Magistrate Judge's "erroneous legal conclusions." (Petition 27-29) The Court did no such thing. It began by stating the rule that Google itself advocates: a "non-dispositive order entered by a magistrate must be deferred to unless it is 'clearly erroneous or contrary to law.'" (A8:4-5 (quoting Grimes, 951 F.2d at 241 (emphasis added)). The District Court then quoted Grimes's holding that orders of a magistrate judge on non-dispositive matters "are not subject to de novo determination." Grimes, 951 F.2d at 241.

The District Court then correctly cited and applied the "clearly erroneous or contrary to law" standard on nearly every page of its analysis. (A8:5, A8:12-13, A8:25, A9:15-16, A9:19-20; A10:11-12, A11:19-20, A11:7-8, A11:16-17, A11:27- 28, A12:8-9, A13:3-4, A13:17-18, A14:2, A14:9, A15:2-3, A15:7-9, A15:13-14, A15:27, A17:4-5) Google does not identify any actual instance in which the District Court "deferred" to the Magistrate Judge's conclusions of law. Indeed, the only "error of law" that Google objected to in the Magistrate Judge's order was the application of Sealed Case, which the District Court clearly agreed with. (A15 n.4

27

(holding that Sealed Case "makes considerable sense and addresses real world practices" and endorsing its "wisdom"))

C. Google cannot establish a "clear and indisputable right" to the writ
because the Magistrate Judge and the District Court would not have
reached a different result if a different legal standard applied.

Not only has Google failed to establish "clear error,"¹⁴ it has failed to establish that it would have obtained a different result absent the supposed errors. The Magistrate Judge plainly did not see the privilege question as a close call, and nothing in her order suggests that the result would have been different if she had applied a lower standard of proof. To the contrary, the District Court expressly held that even if it presumed that the Lindholm email was for the purpose of obtaining legal advice, the evidence cited by the Magistrate Judge would rebut that presumption, and Google would lose. (A15:4-7) It is a general principle of appellate review that harmless errors are not reversible. McDonough Power Equip., Inc. v. Greenwood, 464 U.S. 548, 553 (1984). If there would be no error that would merit correction on a direct appeal, there can be no error that warrants a writ of mandamus.

D. Google has failed to show that it cannot obtain the same relief by other
means, or that the alleged errors cannot be corrected on direct appeal.

As Google concedes, mandamus exists to correct errors that cannot be corrected by other means. Even if Google's claims of error had merit – and they

28

do not – the only practical relief that Google seeks at this late stage of the proceedings is to prevent Oracle from using the email at trial.¹⁵ As discussed above, the email has been produced and is publicly available, so preventing disclosure (or "spillover effects" of disclosure (Petition, 30 n.62)) is impossible.

Trying to prevent Oracle from using the document at trial is not a proper basis for mandamus relief. The Ninth Circuit and other courts consistently have held that, while interlocutory appeals were available to prevent disclosure in the first place, they serve no purpose once disclosure has already occurred. Thus, in Truckstop.net, LLC v. Spring Corp., the Ninth Circuit held that there could be no interlocutory review of a discovery order that had already mandated release of an entire document, as the fact of disclosure would make "the issue of privilege effectively moot." 547 F.3d 1065, 1068-70 & 1069 n.2 (9th Cir. 2008). The Court held that although the inadvertent disclosure of the allegedly privileged email "may be unfortunate, the chicken has already flown the coop ≠ the alleged harm from disclosure had already occurred." Id. at 1070. Any other harm would be

29

effectively reviewable on appeal from a final judgment. Id. See also In re EchoStar Comm'cns Corp., 448 F.3d 1294, 1298 (Fed. Cir. 2009).¹⁶

E. This Court cannot grant Google the relief it seeks because Google has
failed to submit the complete evidentiary record on the privilege issue
and the courts below have not decided the waiver issue.

Finally, this Court cannot grant Google the relief it seeks: to hold that the Lindholm email is privileged, compel its return by Oracle, and seal all court documents "referencing" it. (A1-A3). First, Google has provided this Court with only a portion of the evidentiary record below, making it impossible for this Court to determine that the Lindholm email is privileged. Second, Oracle argued below that Google waived any otherwise applicable privilege, a question that the courts below did not need to answer. Consequently, even if there were error – and there is not – and even if that error warranted mandamus – and it would not – the only relief that this Court could grant would be remand to the District Court to decide the privilege and waiver issues applying some other legal standard.

IV. Conclusion

There is no error in the rulings below, and any harm resulting from disclosure has already occurred. The petition should be denied.

30

Dated: November 28, 2011

Respectfully submitted,

BOIES, SCHILLER & FLEXNER LLP

By: [signature]
Fred Norton

Attorneys for Respondent
Oracle America Inc.

________________
¹ In 2010, Oracle acquired Sun Microsystems, Inc. and along with it, the Java patents and copyrights at issue.

² The Magistrate Judge merely observed – at the hearing – that “I don’t need to reach the question of whether somebody is not credible or is lying, or anything like that.” (A180, at 24:17-19 (emphasis added)). Her actual written finding in the Order – that the “contents of the email itself severely undermine the claim” of privilege – must be credited over any contention that Google’s declarations credibly established that the email was for the purpose of seeking legal advice, or prepared in anticipation of litigation.

³ Petition, 8, citing A31-32 at ¶ 7. Mr. Lindholm’s actual, somewhat fuzzier claim was that he “was not intending to give general business advice to anyone in connection with Android’s ongoing business operations.” A31-32 at ¶ 7.

⁴ Google has never disclaimed these representations to the Court. Instead, it has argued in its briefs that its trial counsel were “taken by surprise” by the use of the Lindholm email at the hearing. (Petition, 10) The District Court properly rejected that claim as lacking any evidentiary basis. (A15:18-19)

⁵ Google argues that the Ninth Circuit’s five-factor Baumann test applies here. It does not. “A request for mandamus relief is determined under Federal Circuit law, except to the extent that underlying procedural issues may be governed by the law of the regional circuit . . . .” In re Shared Memory Graphics LLC, ___F.3d ___, 2011 WL 4390020, at *3 (Fed. Cir. Sept. 22, 2011).

⁶ Google never cited Upjohn to the Magistrate Judge, in its briefs or at oral argument. (A16:7-8; A78-83; A175-182)

⁷ Further, the fact that non-legal executives are included on the communication and non-legal executives are asked to comment on the strategy – as was the case here – indicates that the communication is not primarily for the purpose of obtaining legal advice. See, e.g., In re CV Therapeutics, Inc. Sec. Litig., 2006 WL 2585038, at *3 (N.D. Cal. Aug. 30, 2006) (finding no privilege where “documents on their face” sought comments from non-attorneys).

⁸ See also Griffith, 161 F.R.D. at 697 (“no privilege can attach to any communication . . . that would have been made because of a business purpose”); Adams v. Gateway, Inc., 2003 WL 23787856, at *15 (D. Utah Dec. 30, 2003) (“Documents showing business or negotiation strategy are not matters of litigation strategy protected by the work product privilege.”); Hardy v. New York News, 114 F.R.D. 633, 644-45 (S.D.N.Y. 1987) (“When the ultimate corporate decision is based on both a business policy and a legal evaluation, the business aspects of the decision are not protected simply because legal considerations are also involved.”).

⁹ Although Google complains that “clear showing” is a “heightened standard” for in-house lawyers, many courts apply a clear showing rule to any privilege claim. See, e.g., Peat, Marwick, Mitchell & Co. v. W., 748 F.2d 540, 542 (10th Cir. 1984) (per curiam), cert. denied, 469 U.S. 1199 (1985); Navigant Consulting, Inc. v. Wilkinson, 220 F.R.D. 467, 473-74 (N.D. Tex. 2004); Ali v. Douglas Cable Comm’cns, Ltd. P’ship, 890 F. Supp. 993, 994 (D. Kan. 1995).

¹⁰ Google attempts to bolster its argument by asserting that the Supreme Court in Upjohn made no distinction between inside and outside counsel. (Petition, 16) But there was no need to do so, as the magistrate judge in that case had already found that the purpose of the communications with employees was “to give legal advice to the company[.]” Upjohn, 449 U.S. at 394; see also id. at 386 (“We decline to lay down a broad rule or series of rules to govern all conceivable future questions in this area, even if we were able to do so.”).

¹¹ See, e.g., United States v. ChevronTexaco Corp., 241 F. Supp. 2d 1065, 1076 (N.D. Cal. 2002); Amway Corp. v. Procter & Gamble Co., 2001 WL 1818698, at *5 (W.D. Mich. Apr. 3, 2001); United States v. Chevron Corp., 1996 WL 264769 at *4 (N.D. Cal. Mar. 13, 1996).

¹² A208-A217.

¹³In addition to courts in the D.C. Circuit, see Borase v. M/A COM, Inc., 171 F.R.D. 10, 14 (D. Mass. 1997); Acumen Re Mgmt. Corp. v. Gen. Sec. Nat. Ins. Co., 2010 WL 3260166, at *2 (S.D.N.Y. July 30, 2010); Ames v. Black Entm't Television, 1998 WL 812051, at *8 (S.D.N.Y. Nov. 18, 1998); Teltron, Inc. v. Alexander, 132 F.R.D. 394, 396 (E.D. Pa. 1990); Neuberger Berman Real Estate Income Fund, Inc. v. Lola Brown Trust No. 1B, 230 F.R.D. 398, 411 n.20 (D. Md. 2005); Amway Corp. v. Procter & Gamble Co., 2001 WL 1818698, at *5 (W.D. Mich. Apr. 3, 2001); In re Rospatch Sec. Litig., 1991 WL 574963, at *8 (W.D. Mich. Mar. 14, 1991); Argenyi v. Creighton Univ., 2011 WL 3497489, at *4 (D. Neb. Aug. 10, 2011); O’Shea v. Epson Am., Inc., 2010 WL 2305863, at *4 (C.D. Cal. June 4, 2010); U.S. ex rel. Parikh v. Premera Blue Cross, 2006 WL 3733783, at *5 (W.D. Wash. Dec. 15, 2006); United States v. ChevronTexaco Corp., 241 F. Supp. 2d 1065, 1076 (N.D. Cal. 2002); United States v. Chevron Corp., 1996 WL 264769, at *4 (N.D. Cal. Mar. 13, 1996); Lindley v. Life Investors Ins. Co. of Am., 267 F.R.D. 382, 390 (N.D. Okla. 2010); Marten v. Yellow Freight System, Inc., 1998 WL 13244, at *6 (D. Kan. Jan. 6, 1998); Pizza Mgmt., Inc. v. Pizza Hut, Inc., 1989 WL 9334 at *4 (D. Kan. Jan. 10, 1989).

¹⁴ Under Ninth Circuit law, which Google relies on, failure to show clear error alone necessarily defeats a petition for mandamus. In re Van Dusen, 654 F.3d 838, 841 (9th Cir. 2011).

¹⁵ Google failed to inform this Court that it has a pending motion in limine with the District Court, seeking to exclude the Lindholm email on Rule 403 grounds – the very issue it argues to this Court on page 30 of the petition. (A183) Thus, Google cannot claim it has no other means of obtaining the only practical relief it seeks from this Court. Even if that motion is denied – as it should be – any claimed error in the use of the document at trial can and should be reviewed on direct appeal.

¹⁶ The Supreme Court’s decision in Mohawk Industries v. Carpenter, 130 S.Ct. 599 (2009) precludes appeals of privilege questions under the collateral order doctrine. Nonetheless, the logic of these decisions remains sound: courts of appeal should not intervene, by interlocutory appeal or mandamus, if the allegedly irreparable harm has already occurred and any other injury can be reviewed on direct appeal.

31

CERTIFICATE OF SERVICE
[see PDF]

Google's follow-up letter:

[KEKER & VAN NEST letterhead]

January 5, 2012

Jan Horbaly, Circuit Executive and
Clerk of the Court
United States Court of Appeals
for the Federal Circuit
[address]

Re: In re Google Inc., Misc. No. 106

Dear Mr. Horbaly:

I represent petitioner Google Inc. in the above-referenced matter, in which Google requests that the Court issue a writ of mandamus that (among other things) would direct the district court to treat as privileged certain attorney-client communications collectively known as "the Lindholm email." I request that this letter be provided to the panel members currently considering this matter.

In opposing Google's petition, Oracle America, Inc. argued that the Court should deny the petition because Google could obtain the requested relief by means of its then-pending motion in limine to exclude the Lindholm email under Federal Rule of Evidence 403. See Oracle Opp. at p. 29 n.15.

I now write to inform the Court that the district court yesterday denied Google's motion in limine in its entirety. See ECF 676, pp. 1-2. Accordingly, that motion cannot provide the relief sought in the petition pending before this Court.

Although the district court has now ruled on motions in limine, trial is not imminent. Yesterday, the district court also issued a Final Pretrial Order stating that trial will not commence before March 19, 2012. See ECF 675, p. 1.

Respectfully submitted,

[signature]
ROBERT A. VAN NEST

UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
SAN FRANCISCO DIVISION

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

Case No. CV 10-03561 WHA

ORACLE AMERICA INC.’S RESPONSIVE
SUPPLEMENTAL BRIEF IN OPPOSITION
TO GOOGLE’S MOTION IN LIMINE NO. 3
TO EXCLUDE PORTIONS OF
COCKBURN REPORT ON DAMAGES

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

1

TABLE OF CONTENTS

INTRODUCTION ...... 1

ARGUMENT ...... 3

A. Prof. Cockburn’s Apportionment Analysis Is Legally and Factually
Correct ...... 3

B. Prof. Cockburn’s Upward Adjustment Based On Sun’s Loss Of
Compatibility And Control Is Warranted Under the Facts And The Law ...... 8

C. Cockburn’s Calculation Of A Hypothetical License Damages Award For
Google’s Copyright Infringement Is Valid And Supported ...... 14

D. Cockburn’s Calculation Of Patent Damages By Features And Calculation
Of Copyright Damages Without Any Individual Copyright Breakdown Is
Appropriate ...... 15

E. There Is No Reason To Bar Dr. Cockburn From Testifying About Other
Relevant Licenses And Settlements, Including The Sun v. Microsoft
Agreements ....... 17

CONCLUSION ...... 19

i

TABLE OF AUTHORITIES

CASES

Cf. Cornell Univ. v. Hewlett-Packard Co.,
609 F. Supp. 2d 279 (N.D.N.Y. 2009) ...... 17

Finjan Inc. v. Secure Computing Corp.,
626 F.3d 1197 (Fed. Cir. 2010) ...... 5

Function Media, L.L.C. v. Google Inc.,
No. 2:07-CV-279-CE, 2010 WL 272409 (E.D. Tex. Jan. 15, 2010) ...... 5, 16

Garretson v. Clark,
111 U.S. 120 (1884) ...... 4

Georgia-Pacific Corp. v. U.S. Plywood Corp.,
318 F. Supp. 1116 (S.D.N.Y. 1970) ....... 4, 9

Global-Tech Appliances, Inc. v. SEB S.A.,
131 S. Ct. 2060 (2011) ...... 10

ID Sec. Sys. Canada, Inc. v. Checkpoint Sys., Inc.,
249 F. Supp. 2d 622, amended, 268 F. Supp. 2d 448 (E.D. Pa. 2003) ...... 11

Interactive Pictures Corp. v. Infinite Pictures, Inc,
274 F.3d 1371 (Fed. Cir. 2009) ....... 9, 10, 13

Lucent Techs, Inc. v. Gateway, Inc.,
580 F.3d 1301 (Fed. Cir. 2009) ...... 10, 18

Medtronic, Inc. v. Boston Scientific Corp.,
Civ. No. 99-1035, 2002 WL 34447587 (D. Minn. Aug. 8, 2002) ...... 5, 6, 7

Metallic Rubber Tire Co. v. Hartford Rubber Works Co.,
275 F. 315 (2d Cir. 1921) ...... 16

Oracle USA, Inc. v. SAP AG,
No. C07-1658 PJH, at Dkt. No. 1088 ...... 14, 15

Panduit Corp. v. Stahlin Bros. Fibre Works,
575 F.3d 1152 (6th Cir. 1978) ...... 2, 9, 11

Polar Bear Prods. v. Timex Corp.,
384 F.3d 700 (9th Cir. 2004) ...... 15

ResQNet.com v. Lansa, Inc.,
594 F.3d 860 (Fed Cir. 2010)...... 4, 18

ii

SEB S.A. v. Montgomery Ward & Co., Inc.,
594 F.3d 1360 (Fed. Cir. 2010) ...... 10

Studiengesellschaft Kohle, m.b.H. v. Dart Indus., Inc.,
862 F.2d 1564 (Fed. Cir. 1988) ...... 7

TAS Distrib. Co. v. Cummins Engine Co., Inc.,
491 F.3d 625 (7th Cir. 2007) ...... 11

TK-7 Corp. v. Estate of Barbouti,
993 F.2d 722 (10th Cir. 1993) ...... 11

Trademark Research Corp. v. Maxwell Online, Inc.,
995 F.2d 326, 332 (2d Cir. 1993) ...... 11

VS Techs., LLC v. Twitter, Inc.,
2:11CV43, 2011 WL 4744572 (E.D. Va. Oct. 5, 2011) ...... 16

Wordtech Sys., Inc. v. Integrated Networks Solutions, Inc.,
609 F.3d 1308 (Fed. Cir. 2010 ...... 10

Zenith Electronics Corp. v. WH-TV Broad. Corp.,
395 F.3d 416 (7th Cir. 2005) ...... 11

RULES

Fed. Rules of Evid. 403 ...... 3

Fed. Rules of Evid. 701 ...... 11

iii

Pursuant to the Court’s September 27, 2011 order (Dkt. No. 470), Oracle submits this supplemental brief in opposition to Google’s supplemental brief (Dkt. No. 549) regarding Google’s motion in limine to exclude the testimony of Oracle’s damages expert, Prof. Iain M. Cockburn.

INTRODUCTION

Google’s supplemental brief does not raise any additional arguments that warrant excluding or limiting Prof. Cockburn’s testimony. Although the Court directed that this round of briefing be “based on Dr. Cockburn’s reply report and deposition” (Dkt. No. 470), nothing from Prof. Cockburn’s reports replying to the opinions of Google’s damages experts, Drs. Cox and Leonard, nor in Prof. Cockburn’s October 17, 2011 deposition testimony provides any new basis for Google’s Daubert challenge. Google simply restates and repackages it original arguments, cutting and pasting carefully selected snippets from Prof. Cockburn’s deposition that, taken out of context and ignoring the rest of the testimony, it believes supports its original arguments. The deposition taken as a whole confirms that Prof. Cockburn’s analysis is based on a wealth of evidence, detailed economic analysis, and the methodology suggested by this Court, and therefore passes muster under Daubert.

First, Google argues that Prof. Cockburn’s reasonably royalty calculations—for both patents and copyrights—should be excluded because Prof. Cockburn does not value all the intellectual property that is not in suit. But Google cites no case that requires such an endeavor, particularly where the expert—as Prof. Cockburn does—provides a detailed, multifaceted analysis, (including econometric analysis, conjoint analysis, and consideration of numerous contemporaneous Google documents) of the demonstrated value of both the full package of intellectual property over which the parties were negotiating in 2006 and the specific intellectual property at issue. Rather than address these issues in a Daubert motion, Google acknowledges Prof. Cockburn’s exhaustive methods of valuing the intellectual property at issue in this case, but relegates to a footnote the weak rejoinder that it intends to challenge the scientific basis for these methods “at trial.” (Supp. Br. at 5 n. 1.) Google neither attacks the reliability or scientific basis of Prof. Cockburn’s valuation of the specific patents and copyrights at issue, nor provides any legal basis for its assertion that more is required. In all respects, Prof. Cockburn’s valuation analysis withstands Google’s purported Daubert challenge.

1

Second, Google again argues that Prof. Cockburn’s analysis is flawed because he adjusts the hypothetical license to account for Sun’s contemporaneous expectations for convoyed sales in the jointly controlled Android ecosystem without fulfilling the four-factor Panduit test for recovery under a lost profits measure. This argument is inconsistent with both the real-world considerations of the negotiating parties and the law governing patent and copyright damages. Unlike recovery of damages using a lost-profits measure, which is what the Panduit factors relate to, a reasonable royalty must be based on the parties’ expectations at the time of the negotiation. Numerous Federal Circuit cases approve of using internal projections to calculate a reasonable royalty, and it is commonplace to use the patentee’s expectation of convoyed sales as part of the reasonable royalty analysis as a whole, as Prof. Cockburn does. Google completely ignores the law governing recovery of a reasonable royalty based on a hypothetical negotiation, instead relying entirely on cases addressing the recovery of lost profits in contract, civil conspiracy, and antitrust cases. Those cases are all governed by different standards that are not, as the hypothetical license must be, based on the parties’ reasonable expectations. Google’s argument is baseless, and Prof. Cockburn’s analysis is sound.

Google also again argues that Prof. Cockburn’s upward adjustment to the “starting point” reasonable royalty should be stricken because it is supposedly based on a single, “unreliable” Sun business planning document. Google conveniently ignores that, with regard to the same upward adjustment, Google’s own damages experts rely on a single document of their own (a license between Sun and a third party, Danger)—indeed, a distortion of that document. Google also ignores that Prof. Cockburn explained at deposition and in his report that the adjustment is based on far more than the one Sun document. But even putting those facts aside, Google fails to explain how or why reliance on a contemporaneous revenue projection, especially where it declined to take the author’s deposition to even try to establish that those projections are unreliable, is problematic under Daubert. It is not.

Third, Google again argues that Prof. Cockburn’s copyright hypothetical license calculations should be stricken because Google contends there are no comparable real-world licenses. Even though Oracle pointed out in its opposition that the only legal authority for Google’s argument had been revised in a matter of weeks by its own author, Google persists in its faulty legal argument. Google’s argument also fails as a matter of fact. Prof. Cockburn calculates the hypothetical lost copyright license

2

fee with reference only to real-world contemporaneous documents, not speculation or “subjective” evidence of the sort Google says is insufficient. (By contrast, Google’s experts rely heavily on subjective interview evidence that reinvents history and contradicts documents written by those same interviewees in Google’s own contemporaneous documents.) Prof. Cockburn quantifies those aspects of the hypothetical license that are objectively grounded in the parties’ actual negotiations. There is neither a factual nor a legal basis for Google’s attack.

Fourth, Google repeats its argument that Prof. Cockburn is required to assign value to each individual asserted claim of each patent and each portion of the copyright claim. Google failed to cite a single case that would establish this hyper-granular apportionment requirement in its opening brief, and it fails to do so again. Prof. Cockburn carefully evaluates each of the individual patented features at issue in this case. Google has not suggested that the opinions would be any different on a claim-byclaim basis, and provides no basis on which to exclude Prof. Cockburn’s opinions.

Fifth, Google again argues that Prof. Cockburn should be barred from discussing certain other technology licenses, including the license entered into between Sun and Microsoft. Prof. Cockburn’s treatment of the Sun-Microsoft dispute is correct under applicable legal principles and correct under the facts. Google does not make a Daubert argument so much as an argument under Rule 403, but the relevance of the Sun-Microsoft settlement far outweighs its prejudicial force. The jury should be able to decide what weight to give that agreement. Moreover, Google had the opportunity to file a motion in limine to exclude those agreements from evidence, and it declined. Google should not be able to shoehorn a second in limine motion into its Daubert motion.

Oracle respectfully requests that the Court deny Google’s motion.

ARGUMENT

A. Prof. Cockburn’s Apportionment Analysis Is Legally and Factually Correct

Google now focuses its Daubert challenge on its assertion that Prof. Cockburn’s entire reasonable royalty analysis, for both patents and copyrights, should be stricken because Prof. Cockburn’s opinion is limited to the value of the intellectual property in suit and he did not assign a value to intellectual property that is not in suit. (Supp. Br. at 1–6 (“Having no sense of the whole, he has no basis to opine on the value of any of the parts.”).) This argument is meritless on its face. Prof.

3

Cockburn has a detailed, multifaceted “basis to opine on the value” of precisely the parts that are at issue in this case—an analysis that, as the Court directed, starts from “the whole” and then specifically apportions, based on econometric analysis and multiple contemporaneous data points, to value the relevant parts: the patents and copyrights in suit. Google’s complaint that Prof. Cockburn somehow committed an error by not assigning value to technology that is not claimed here has no basis in law or reason.

The underlying legal principles are not new. “When a patent is for an improvement, and not for an entirely new machine or contrivance, the patentee must show in what particulars his improvement has added to the usefulness of the machine or contrivance.” Garretson v. Clark, 111 U.S. 120, 121 (1884); see also ResQNet.com v. Lansa, Inc., 594 F.3d 860, 869 (Fed Cir. 2010) (“the trial court must carefully tie proof of damages to the claimed invention’s footprint in the market place.”) (citing cases). The calculation of damages should consider the “portion of the realizable profit that should be credited to the invention as distinguished from non-patented elements, the manufacturing process, business risks, or significant features or improvements added by the infringer.” Georgia-Pacific Corp. v. U.S. Plywood Corp., 318 F. Supp. 1116, 1120 (S.D.N.Y. 1970).

Professor Cockburn’s analyses are consistent with these principles and the Court’s order. He evaluates how the claimed intellectual property “added to the usefulness” of the infringing Android devices and what portion of the starting point should be “credited to the invention.” (See Dkt. No. 230 (July 22, 2011 Daubert order) at 9 (“we must ultimately isolate the infringing features’ contribution”).) He does this by considering (a) contemporaneous Google documents discussing the importance of the functionality provided by the patents and copyrights in suit, (b) benchmarking analyses that measure the precise performance enhancements contributed by the patents in suit, (c) an econometric analysis measuring consumers’ willingness to pay for the patented enhancements and the value of the large number of applications facilitated by the use of the copyrighted Java APIs, and (d) a conjoint marketing analysis, based on survey data, measuring the value that consumers place on the particular performance enhancements and number of applications.¹ Having valued both the whole and the relevant

__________________________________

¹ Google contends in a footnote, without providing any basis for its contention, that all of these methodologies must have been rigged because Oracle engineers conducted some of the benchmarking analyses (using established, standard methods) underlying them. (Supp. Br. at 5 n.1.) They were not. In fact, Google’s expert’s profound misunderstanding of one of the methodologies that Prof. Cockburn considered—the conjoint—is detailed in Oracle’s Daubert motion to exclude Dr. Leonard’s testimony. (See Dkt. Nos. 558–560.) As noted above, it is telling that Google has not raised any actual Daubert challenge to the scientific basis for Prof. Cockburn’s apportionment methodologies, and instead just indicated that it will do so at trial.

4

components, Prof. Cockburn is not obligated to separately value any non-infringing components. See, e.g., Function Media, L.L.C. v. Google Inc., No. 2:07-CV-279-CE, 2010 WL 272409, at *4 (E.D. Tex. Jan. 15, 2010); Finjan Inc. v. Secure Computing Corp., 626 F.3d 1197, 1211 (Fed. Cir. 2010) (finding that, based on internal documents calling the patented features “important,” the jury could infer that a substantial fraction of the accused products’ profits stemmed from the patented method). Google cites only one case for its argument—an unpublished opinion from the District of Minnesota, Medtronic, Inc. v. Boston Scientific Corp., Civ. No. 99-1035, 2002 WL 34447587 (D. Minn. Aug. 8, 2002). That case provides no basis for excluding or limiting Prof. Cockburn’s testimony. In Medtronic, the court excluded testimony by the plaintiff’s expert because the expert had claimed that the patents-in-suit were worth “virtually all” of the value of the purchase price of the portfolio that defendants had paid to acquire them. Id. at *7–12. The expert opined that the patents-in-suit accounted for “all or substantially all” of the value of the portfolio based on his “gut feeling,” his “speculation” that the patents-in-suit were “broad and cover a lot of applications,” and his vague recollection of a separate deposition in which an outside law firm had referred to the patents as “very valuable.” Id. at *6–7. The expert relied on one piece of documentary evidence to back up his assertions: an internal memorandum from the defendants that stated that the three patents-in-suit were the “primary value” of the portfolio. Id. at *12. The court found that “primary value” did not necessarily translate into or support the expert’s opinion that the patents supplied “virtually all of the value.” Id. Accordingly, the court concluded that there was not a reliable factual basis for the expert’s opinion regarding the value of the patents-in-suit and excluded his testimony. Id.

This case bears no resemblance to Medtronic. As Google concedes, Prof. Cockburn’s apportionment analysis is not based on a “gut feeling.” (Supp. Br. at 5 (Cockburn’s analysis based on “three performance-based studies”).) First, Prof. Cockburn’s evaluation of the contribution of the

5

intellectual property rests on multiple economic analyses based on two separate and independent measures of consumer preferences, input from the technical experts, and the review of many contemporaneous documents that establish Google’s appraisal of the value of the patented functionality and the importance of adopting copyrighted Java API to attract application developers. Andy Rubin, the head of Android, has acknowledged that this is the whole purpose of an open platform. (Declaration of Meredith Dearborn (“Dearborn Decl.”) Ex. A (Rubin Dep. Tr. 4/5/2011 at 91:19-23 (“There is no purpose of building an open platform other than to attract third-party developers to it.”)).) This is more than enough “reliable factual basis.” Medtronic, 2002 WL 34447587, at *12. Second, rather than looking at just one document, as the expert did in Medtronic, Prof. Cockburn takes into account numerous Google documents that establish that an incremental increase in speed (provided by several of the patents in suit) is vitally important:

• “Application switching on a mobile device is extremely critical; we target significantly less than 1 second to launch a new application . . . A noticeable wait in such situations will quickly make users hate you.” (Dearborn Decl. Ex. B (GOOGLE-03169165).)
• “Speed remains one of the most important factors in adoption and retention. This includes … delays that are introduced before an action executes.” (Dearborn Decl. Ex. C (GOOGLE- 03403114).)
• “Every test and every survey indicates that speed is one of the most important ‘features’ of a product; even milliseconds in response time seem to matter.” (Dearborn Decl. Ex. D (GOOGLE-10-00045447).)
• “[Larry Page] Wants all screens to load in

Third, Prof. Cockburn does not assert that 100% of the starting point should be apportioned to the intellectual property, nor does he claim that the patents and copyrights account for “virtually all of the value” of that starting-point license, as did the expert in Medtronic. His opinion is that the proper apportionment percentage is at least 30% for the patents and 15% for the copyrights. Medtronic provides no basis to exclude any of Prof. Cockburn’s testimony.

Google’s attack on Prof. Cockburn also relies on a false factual proposition regarding the components of the starting point license. Google’s contends that Prof. Cockburn should have accounted for the value of “Sun’s JAVA trademark” and Google’s “access to Sun’s engineers” (Supp. Br. at 2), and presumably reduced his apportionment percentages by some amount as a result. But the draft

6

contract sent by Sun to Google in March 2006 specifies what was covered by Google’s $100 million payment to Sun: “In consideration for Sun’s agreement to provide Google access to the Sun Technology and Sun’s agreement to release the Sun Technology as part of the Open Source Stack . . . Google agrees to pay Sun” an amount referred to in that contract as “Under Consideration.” (Dearborn Decl. Ex. F (GOOGLE-01-00062072 at 080) (emphasis added).) The payment clause did not refer to any trademark or access to any engineers. In fact, Sun and Google both stood to benefit in other ways from that agreement, but none of those other benefits were tied to the payment term, which was for a license to “Sun Technology” and the release of those technologies under an open source license. Sun defined “Sun Technology” to include its patents and copyrights (not trademarks) pertaining to the new platform. But in any event, Prof. Cockburn values the whole agreement under consideration by the parties in 2006 and then apportions from there based on his analysis of the portion of the value of Android attributable to the intellectual property in suit. It does not matter—at least not in this case— what specifically is included in the remainder of the valuation of the unconcluded 2006 agreement.²

If Google truly believed it was important to value the intellectual property not in suit, it could have asked its experts to conduct that analysis, and then presented that analysis at trial. Google did not. Instead, Google decided to ask Prof. Cockburn a series of questions at his deposition in an effort to tie this case to the unpublished Medtronic decision, and to attack Prof. Cockburn for not doing something

_________________________________

² It is true that the fact that Sun and now Oracle ordinarily license Java patents in a bundle should be relevant to the hypothetical license here. Studiengesellschaft Kohle, m.b.H. v. Dart Indus., Inc., 862 F.2d 1564, 1568 (Fed. Cir. 1988) (“[T]he patentee’s usual licensing approach should be considered in assessing a reasonable royalty.”). Sun, and now Oracle, are not in the custom of licensing particular patents; instead, they license the bundle of intellectual property that comprises the Java platform as a whole. Both parties’ experts agree that there are no comparable benchmark licenses on which to base a reasonable royalty for these particular isolated patents, nor is there a clean economic way to apportion all the way down to the patent level. In fact, under similar circumstances in a different case, Google’s own expert, Dr. Leonard, employed the entire market value rule: “The problem is, how much is that component worth versus the whole product? It makes it difficult. . . . So like let’s take a car. And let's say you had a patent on the engine. You could base a royalty for that using a patented technology on the price of the whole car. That’s really what I'm talking about here. Or you could somehow try to figure out what the value of the engine is. But the problem is that the engine isn’t really sold separately. It’s really hard to figure out what that value is. And you’re going to get into disputes between the two sides, the person who is paying the royalty is going to say, oh, the engine isn’t worth much. And the person receiving the royalty is going to say, oh, the engine is worth everything. So to avoid those kind of disputes you just set the royalty base as being the whole car.” (Dearborn Decl. Ex. G (Trial Tr., Day 7, November 15, 2007, 1342:9-11, 1342:23-1343:10, Bard Peripheral Vascular v. W.L. Gore & Associates, No. CV-03-597-PHX-MEM).)

7

that he had no obligation to do. Google’s arguments are baseless, and the motion should be denied.

B. Prof. Cockburn’s Upward Adjustment Based On Sun’s Loss Of Compatibility And
Control Is Warranted Under the Facts And The Law

Google’s second attack on Prof. Cockburn’s opinions repeats an argument raised in its original motion in limine: that Prof. Cockburn should not be permitted to consider and account for the substantial revenues that Sun expected from commercial licensing in connection with a compatible Android, despite the fact that the entire agreement with Google was based on the assumption that Sun would obtain those revenues. (Supp. Br. at 6–11.) Google adds nothing new in substance to its prior attack, and there is no basis to force the experts and the jury to ignore those substantial expected revenues in calculating damages in this case. Google’s argument mischaracterizes Prof. Cockburn’s report and testimony, misunderstands the facts, and ignores governing law.

First, Google contends that Prof. Cockburn “admitted the sole source” of this adjustment is “profits Sun allegedly lost the chance to earn because Google deployed the Android position.” (Supp. Br. at 6–7.) Google mischaracterizes Prof. Cockburn’s testimony. Although Google’s counsel designed its questions to try to get Prof. Cockburn to adopt the “lost profits” terminology, Prof. Cockburn testified that it was incorrect to characterize the calculation as “being lost profits” and that they instead reflect—as described in Prof. Cockburn’s reports—“a method of capturing the value to Sun of compatibility and control.” (Dearborn Decl. Ex. H (Cockburn Dep. at 135:17–136:2).) Prof. Cockburn later elaborated:

I’m using this not to – not to assess lost profits, but rather to understand the magnitude of the value that [Sun] placed upon the kind of compatibility and control which would allow them to realize those revenues. . . . These provide, in my view, some concrete evidence as to a minimum amount of money that Sun would have required from the hypothetical license in order to compensate them for loss of compatibility and control.

(Id. at 192:11–194:20.) Prof. Cockburn properly applies the difference between a lost profits measure of damages and the appropriate components of a reasonable royalty. As a measure of copyright damages, he calculates lost profits from Sun’s Java ME licensing revenues and its own plan for a full mobile software stack, Project Acadia, as an alternative measure of actual damages for Google’s copyright infringement in this case. (Cockburn Report ¶¶ 439–98.) The upward adjustment to the

8

hypothetical license is not actual lost profits—it reflects Sun’s expectations in 2006 for convoyed sales to third parties as a result of a license agreement with Google. Notably, Prof. Cockburn does not include the loss of such convoyed sales—which would have been realized only had there been a license, not in the but-for world (no infringement) that informs lost profits analysis—in his lost-profits calculation. The expected convoyed sales are only, and appropriately, relevant to valuation of the patents and copyrights using a hypothetical license measure that seeks to measure (a) the value of a compatible, jointly controlled Android, plus (b) the value lost if compatibility is changed to incompatibility.

Second, Google’s legal challenge to Prof Cockburn’s adjustment, which Google repeats from its original motion in limine, is still meritless. As Oracle described in its original opposition brief (Dkt. No. 494-1 at 12), Google’s argument about the relationship between lost profits and reasonable royalty analysis is foreclosed by Panduit and Georgia-Pacific Factor 6, which specifically permits the expert to consider convoyed and ancillary sales in the hypothetical license analysis. In fact, Google now admits that “a patentee’s likely lost profits is [sic] one factor that can be used to justify a reasonable royalty calculation.” (Supp. Br. at 7 (emphasis added).) This legal standard makes sense, given that negotiating parties of course consider expected gains and losses when negotiating a license. Yet Google now claims, with no case citation, that the established Georgia-Pacific rule permitting consideration of any reasonably expected losses must not apply to this case. Google is incorrect.

Sales projections are routinely used to calculate damages based on the parties’ reasonable expectations at the time of the hypothetical negotiation. The Federal Circuit’s opinion in Interactive Pictures Corp. v. Infinite Pictures, Inc, 274 F.3d 1371, 1385 (Fed. Cir. 2009), a case that Oracle cited in its opposition to Google’s original motion (Dkt. No. 494-1 at 12) and which Google still ignores in its supplemental brief, is fatal to Google’s position. In Interactive Pictures, the plaintiff based its reasonable-royalty base on the infringer’s business plan and projections for future sales—a document that was prepared two months before infringement began. 274 F.3d at 1385. The defendant argued that this royalty base was outdated, and contained overly optimistic assumptions of future revenue growth that never actually materialized. The Federal Circuit disagreed with the defendant. The court first observed that “[w]e have previously upheld awards of damages premised on a lump sum royalty

9

payment based on an infringer’s expected sales.” Id. The court also rejected the defendant’s argument that the projections were too speculative, even though the projections were not ultimately realized. “The fact that Infinite did not subsequently meet those projections is irrelevant to Infinite’s state of mind at the time of the hypothetical negotiation. Nor does Infinite’s subsequent failure to meet its projections imply that they were grossly excessive or based only on speculation and guesswork. Instead, Infinite’s subsequent failure to meet its projections may simply illustrate the ‘element of approximation and uncertainty’ inherent in future projections.” Id. In fact, the court said that the opposite rule—requiring only actual, realized sales revenue to be considered, rather than expected sales revenue—would “essentially eviscerate the rule that recognizes sales expectations at the time when infringement begins as a bases for a royalty base[.]” Id. Finally, the court concluded that convoyed sales were properly included in the royalty base. Id. (citing cases calling such an approach “eminently reasonable”).) The court thus refused to disturb the jury’s damages award.

Because of the hypothetical license’s focus on expectations, many cases apply this same standard with regard to both infringer’s and the patentee’s profit projections. See Wordtech Sys., Inc. v. Integrated Networks Solutions, Inc., 609 F.3d 1308, 1319 (Fed. Cir. 2010) (reasonable royalty can be calculated from the “infringers’ profit projections for infringing sales”); Lucent Techs, Inc. v. Gateway, Inc., 580 F.3d 1301, 1327 (Fed. Cir. 2009) (citing Interactive Pictures for the proposition that reasonable royalty can be calculated based on “the patentee’s business plan and its projections for future sales prepared two months before infringement began”) (internal quotation marks omitted); SEB S.A. v. Montgomery Ward & Co., Inc., 594 F.3d 1360, 1380 (Fed. Cir. 2010), aff'd sub nom. Global- Tech Appliances, Inc. v. SEB S.A., 131 S. Ct. 2060 (2011) (refusing to reverse hypothetical license damages award based on plaintiff’s argument that it expected to obtain a 40 to 45 percent profit margin, because the “hypothetical construct seeks the percentage of sales or profit likely to have induced the hypothetical negotiators to license use of the invention” and “this court sees no reason to preclude a jury from hearing attorney argument based on an expectation of success.”) (citations omitted). Google has offered no response to any of these cases.

Rather than addressing the governing law, Google relies on five irrelevant, nonbinding cases for the proposition that projections cannot form the basis for an adjustment to a hypothetical license.

10

(Supp. Br. at 10-11.) None of these cases is a patent or a copyright case, and none concerns a hypothetical license. The hypothetical license, of course, is a damages measure unique to intellectual property law, and focuses (unlike lost profits analysis) on the parties’ reasonable expectations at the time of the hypothetical negotiation. Google’s foray into unrelated areas of law—lost-profit damages for breach of contract,³ civil conspiracy,⁴ and antitrust⁵—where such expectations are not part of the damages calculation confirms the weakness of its position. Patent and copyright law provide no support for Google’s argument.

Google’s attempt to knock out the upward adjustment would, as Oracle has previously described, allow Google to pay less than it could have paid if it had just taken the license it needed in 2006. But:

The setting of a reasonable royalty after infringement cannot be treated, as it was here, as the equivalent of ordinary royalty negotiations among truly ‘willing’ patent owners and licensees. That view would constitute a pretense that the infringement never happened. It would also make an election to infringe a handy means for competitors to impose a ‘compulsory license’ policy upon every patent owner.

Panduit Corp. v. Stahlin Bros. Fibre Works, 575 F.3d 1152, 1158 (6th Cir. 1978). It is for this precise reason that the Panduit court, in considering a reasonable royalty, remanded to the district court with instructions to consider “the future business and attendant profit Panduit would expect to lose.” Id. at 1164. Google’s arguments are wrong on the law.

Third, as a matter of fact, Professor Cockburn does not, as Google incorrectly suggests, consider Sun’s convoyed sales projections in a vacuum or rely only on a single document to support the upward adjustment. Instead, he considers the reasonableness of the adjustment in light of Sun’s business model, Sun’s experience in the industry, Sun’s other documents, and Google’s contemporaneous expectation that Sun would have generated significant convoyed sales had a compatible license with

____________________________________

³ Zenith Electronics Corp. v. WH-TV Broad. Corp., 395 F.3d 416, 420 (7th Cir. 2005) (contract lost profits, concerning lay opinion testimony under Rule 701); TAS Distrib. Co. v. Cummins Engine Co., Inc., 491 F.3d 625, 633 (7th Cir. 2007) (contract lost profits, construing Illinois state law); Trademark Research Corp. v. Maxwell Online, Inc., 995 F.2d 326, 332 (2d Cir. 1993) (breach of contract, construing New York state law).

⁴ TK-7 Corp. v. Estate of Barbouti, 993 F.2d 722, 723 (10th Cir. 1993) (civil conspiracy lost profits)

⁵ ID Sec. Sys. Canada, Inc. v. Checkpoint Sys., Inc., 249 F. Supp. 2d 622, amended, 268 F. Supp. 2d 448 (E.D. Pa. 2003) (lost profits under antitrust and state law claims)

11

shared control been concluded. (Cockburn Report ¶ 286–307; see also Agrawal 10/7 Decl. (Dkt. No. 509) at Ex. 3-6, 3-7, 3-8 (documents confirming that Google knew Sun would have a commercial implementation).) Whereas Google provides the Court snippets of Prof. Cockburn’s deposition testimony relating specifically to the Sun business model, it omits that Prof. Cockburn also clearly testified that he (1) assessed the model “in the general context of the record and the other evidence” and in light of his “general review and appreciation of – of the mobile market place at the time and the economics of – of this technology” (Dearborn Decl. Ex. H (Cockburn Dep. at 177:15–178:8.); (2) “looked at the information as to the scale of – of Sun’s Java licensing business,” as well as the revenues that “Sun was already realizing,” and “the volume anticipated” and determined that the projections were “reasonable in the context of the marketplace at the time” (Id. at 178:9–179:9.); and (3) considered the time-to-market and profit margin elements in that projection, and found them reasonable in light of his review of other evidence concerning Sun’s business, the Sun-Google partnership, and the mobile industry. (Id. at 183:5–187:8.) Google’s challenge rests on a mischaracterization and selective misreading of both Prof. Cockburn’s report and Prof. Cockburn’s deposition.⁶

Fourth, Google’s continued insistence that the Sun convoyed sales projections are unreliable is off-base, especially in the context of a Daubert challenge. Google certainly may wish to try to cast doubt on the reliability of the projections at trial, based on examination of the relevant witnesses (none of whom it saw fit to depose) and use of the documents relating to the projections. But that is not the stuff of a motion in limine. Even if it were, Google is incorrect. The reliability of the projections is confirmed by the fact that the kinds of convoyed and ancillary sales projected were precisely the types of sales—licenses to OEMs for commercial implementations of Java technology—with which Sun had extensive experience, and which formed a core portion of Sun’s Java business. (See, e.g., Dearborn

_________________________________

⁶ Google’s argument is particularly ironic in light of the fact that Google’s own damages experts rely on a single, unrelated license agreement between Sun and a third party called Danger Inc. as the basis for their conclusion that the upward adjustment properly attributable to the fact that the hypothetical license would have been for an incompatible implementation is an amount double the royalty for a compatible license. (Dearborn Decl. Ex. J (excerpts from Leonard Report at pp. 45–49).) The irony is multiplied by the fact that Google’s experts’ complete reliance on the Sun-Danger license simply misreads the license and incorrectly conflates the impact on the royalty (double) from whether Danger’s implementation would be branded as “Java” with the impact on the royalty if the Danger implementation were to be incompatible, which is not addressed in the Danger license because compatibility was absolutely required.

12

Decl. Ex. I (Singh Dep. Tr. 47:9-13) (“Sun had engineering services that would do implementations for customers, but consistent with the specification and consistent with the TCK.”); id. 99:18-23 (“Our business model is based on, as I mentioned, on runtime licensing revenue for – for the platform functionality as well as adding in services revenue on top of that.”).) Sun had offered engineering services on its commercial licenses for years, and the projections correspond to commercial licensing that Sun could have earned and expected to earn if it had been part of the Android story. (See Purcell 10/20 Decl. (Dkt. No. 550-4, 550-3) Ex. D & C (OAGOOGLE0100166873, OAGOOGLE0100166874 at 882).) In other words, the projections considered monetizing Android in a manner very similar to the way Sun was accustomed to monetizing Java. Google is incorrect in characterizing the projections as “a completely new line of business for Sun.” (Supp. Br. at 10.)

Nor does anything in the document itself suggest unreliability. The projections are reflected in a presentation created in early 2006, just before the infringement began, making it particularly relevant to shed light on Sun’s reasonable contemporaneous expectations. See Interactive Pictures, 274 F.3d at 1385. The projections were presented to Sun’s finance department. As described in Oracle’s prior brief, nothing in the cover e-mail suggests that the presentation containing the projections disclaims the accuracy of the projections. Kathleen Knopoff, the former Sun employee who is the author of the document and the sender of the e-mail that attached it is on Oracle’s trial witness list (see Dkt. No. 523- 2 at 3) and has been listed on Oracle’s initial disclosures since June 2011. Prof. Cockburn discussed and cited the projections in his May 2011 Report. Despite all of this, Google declined to take any fact discovery at all relating to the projections, including not taking Ms. Knopoff’s deposition. The only deposition testimony in the record relating to the projections is the testimony of another former Sun employee, Vineet Gupta, who participated in the 2005-2006 negotiations with Google and testified as follows with the Sun projections in front of him:

Q. Do you know Kathleen Knopoff?

A. Yes.

Q. Do you find her to be a trustworthy and valuable employee of Sun?

A. Yes. Yes.

Q. Do you have any reason to doubt the integrity of the information she would send you?

13

A. No, no. That's why I said I let other people do their jobs. And I would trust. But that's the cost, and that's the area. That's what it is.

(Dearborn Decl. Exh. K (Gupta Dep. at 307:24-309:24).). Google should not be permitted to use a Daubert challenge in lieu of its failure to take timely discovery or try to rebut this uncontested testimony.⁷

C. Cockburn’s Calculation Of A Hypothetical License Damages Award For Google’s
Copyright Infringement Is Valid And Supported

In its motion in limine, Google argued that Oracle must show “evidence of benchmark transactions, such as licenses previously negotiated for a comparable use of the infringed work” in order to prove a lost copyright hypothetical license fee. (Dkt. No. 494 at 4.) In its opposition, Oracle pointed out that the author of the quoted language, Judge Hamilton, clarified soon after issuing the opinion including that language that “the court did not hold as a matter of law” that a copyright hypothetical license is “available only if the copyright owner provides evidence of actual licenses . . . and/or actual ‘benchmark’ licenses.” Oracle USA, Inc. v. SAP AG, No. C07-1658 PJH, at Dkt. No. 1088. In its supplemental brief, Google concedes that Judge Hamilton rejected Google’s interpretation. (Supp. Br. at 13.) And yet Google again relies on the earlier, disowned standard. (Id. at 12, 13.)

Google’s supplemental brief also shows that Google misunderstands Prof. Cockburn’s copyright hypothetical-license analysis. Google claims that there can be no copyright hypothetical license because, in Google’s view, there is no “objective” evidence that Sun would have licensed an incompatible implementation of Java. But Prof. Cockburn relies precisely on objective evidence: documents underlying the parties’ actual negotiations, which, as this Court has put it, are the “realworld ‘comparable’ close on point.” (Dkt. No. 230 at 14.)

Google repeatedly ignores the objective evidence. For example, it argues that the Ninth Circuit has not “affirmed hypothetical-license damages in a case involving competitors (who do not commonly license each other).” (Supp. Br. at 14.) But to the extent they were competitors at all at the time, Sun and Google did not in 2006 consider their competitive relationship an impediment to a license. Indeed,

_________________________________

⁷ Ironically, Google complains that Prof. Cockburn did not interview Ms. Knopoff. (Supp. Br. at 10.) Unlike Google’s damages experts, Prof. Cockburn preferred to base his opinions on the contemporaneous documents and record evidence, rather than spoonfed facts in interviews.

14

even in his deposition, Mr. Rubin, the head of Android, did not identify any competitive relationship between the companies as an impediment to a license; he simply identified disagreements about control and security. (See Dearborn Decl. Ex. A (Rubin 4/5/2011 Dep. 24:12-26:4).)

Similarly, Prof. Cockburn quantifies the “starting point” license by taking the $100-million starting point⁸ and apportioning down to isolate the contributions of the copyrights-in-suit. (See Cockburn Report Exhibit 23.) The starting point is based on an actual, real-world offer, not “speculation.” Google understood that the purpose of the starting-point license was to compensate Sun not only for the right to license its intellectual property, but also for short-term revenue loss as it moved into an open-source business model. (See, e.g., Dearborn Decl. Ex. L (GOOGLE-26-00007930) (“I’ve been working with Sun and pushing them to open source Java. . . . Sun is prepared to walk away from a $100M annual J2ME licensing business into an open source business model that we together crafted.”); Dearborn Decl. Ex. M (GOOGLE-12-00044903) (“Alan [Brenner, of Sun] . . . just needs to be compensated for collateral short-term revenue loss and get comfortable that this won’t allow Google or anyone else to run away with the platform”).) Prof. Cockburn’s starting point is based on the objective, real-world facts. Google wants to characterize it as somehow “subjective” or “speculative,” but it nowhere deals with the indicia of objectivity in Prof. Cockburn’s report and deposition testimony.

Prof. Cockburn’s copyright damages calculations are consistent with the principles set forth by the Ninth Circuit in Polar Bear Prods. v. Timex Corp., 384 F.3d 700, 708 (9th Cir. 2004). Judge Hamilton’s since-clarified decision in Oracle v. SAP provides no basis for Google’s motion.

D. Cockburn’s Calculation Of Patent Damages By Features And Calculation Of Copyright
Damages Without Any Individual Copyright Breakdown Is Appropriate

In its motion in limine, Google claimed that it was error for Prof. Cockburn to assess patent damages on a patent-by-patent, rather than claim-by-claim, basis. (Dkt. No. 494 at 9.) Google cited no

__________________________________

⁸ Google continues to insist that this starting point is incorrect. (Supp. Br. at 1–2.) Prof. Cockburn used the $100 million offer not only because that is what the Court directed but because that $100 million was linked to a proposed contract that more closely resembles the hypothetical license—one where Sun has less control. Google mischaracterizes the $28 million offer as a lower offer for the same consideration. Google could have tried to calculate the value of the additional control granted to Sun with the new contract, and adjusted the $28 million offer accordingly to come up with an alternative starting point, but Google chose not to.

15

case in which any court required apportionment on a claim-by-claim basis. Despite the fact that Oracle noted that deficiency in its opposition to Google’s motion (Dkt. No. 494-1 at 16), although Google now repeats the argument in its supplemental brief, and although Google in fact has now expanded its argument and contends that Prof. Cockburn should have separately calculated damages for each of the different categories of copyrighted materials (Supp. Br. at 16), Google has still not cited a single case for the proposition that apportionment must be so granular. Nor does Google provide either any indication or any argument that the results would have been different on a claim-by-claim basis as opposed to a feature-by-feature basis.

As described in Oracle’s prior brief (Dkt. No. 494-1), neither the law nor this court’s prior order requires apportionment down to a granular claim-by-claim basis. Most of the cases instructing how to do a proper apportionment focus on the incremental benefit attributable to the infringing feature or work, without any mention of any claim-by-claim or copyright-by-copyright analysis—exactly the analysis that Prof. Cockburn performs. The cases all focus on the incremental benefit provided by the infringing feature—not the claim.

For example, in VS Techs., LLC v. Twitter, Inc., 2:11CV43, 2011 WL 4744572 (E.D. Va. Oct. 5, 2011), the court recently upheld an expert’s apportionment analysis that measured the incremental value of the infringing feature. There, the expert calculated damages based on three methods for measuring incremental value, including a calculation of how use of the infringing feature led to increased use of the Twitter system and revenue to Twitter, and a calculation of the expected incremental economic benefit attributable to the infringing feature, and an evaluation of how, in his experience, that benefit would have been shared between Twitter and VS Technologies. Id. Defendant challenged the expert under Daubert and cases that have emphasized the apportionment rule. The court rejected that challenge, because the “proposed testimony does not seek to extend damages to any features not encompassed within the claimed invention.” Id. at *7. See also Function Media, 2010 WL 272409, at *4 (refusing to exclude plaintiff’s expert on apportionment grounds when he evaluated only the incremental profit); Metallic Rubber Tire Co. v. Hartford Rubber Works Co., 275 F. 315, 322-23 (2d Cir. 1921) (using infringers’-profits rule, calculating damages from an improvement based on the difference in defendant’s profits on improved, infringing tires and ordinary, non-infringing tires).)

16

Similarly, for the copyrights, Prof. Cockburn calculates the incremental benefit to Android’s market share, and therefore to Google’s advertising revenues, of offering a platform with more available apps—a direct consequence of Google’s intentional choice to leverage the Java developer community by providing them with familiar Java APIs. (Dearborn Decl. Ex. H (Cockburn Dep. 201: 24–204:1).)

Prof. Cockburn’s analysis comports with established standards, and Google has cited no authority permitting exclusion of Prof. Cockburn’s analysis on the basis that he should have done an even more granular analysis. Oracle detailed in its responsive brief the reasons why more granular apportionment is not warranted as a matter of fact, law, case management, or economics. (Dkt. No. 494-1 at 16–17.) But even if there were a claim-by-claim apportionment requirement (which there isn’t), Google’s argument simply exalts form over substance. By insisting that Prof. Cockburn slice and dice damages ever more granularly, without once offering an illustration as to how one would accomplish that task, Google has invented an economic principle that is untethered both to facts and to the policies underlying intellectual-property damages. Prof. Cockburn’s analysis goes much farther than the “smallest salable unit” practicing the patents in suit, which is the economically rational royalty base; instead he measured only the incremental benefit to Google of the claimed invention. Cf. Cornell Univ. v. Hewlett-Packard Co., 609 F. Supp. 2d 279, 283–92 (N.D.N.Y. 2009) (smallest salable unit). There was no reason for Prof. Cockburn to do anything more, and there is no basis for Google’s Daubert challenge.

E. There Is No Reason To Bar Dr. Cockburn From Testifying About Other Relevant Licenses
And Settlements, Including The Sun v. Microsoft Agreements

In the last section of its supplement brief Google repeats yet another set of arguments from its original motion: that Prof. Cockburn’s citation to and consideration of the Qualcomm-Nokia license and the Sun-Microsoft settlement in his report is impermissible. (See Dkt. No. 494 at 9-10.) Google seeks an order barring Prof. Cockburn from “testifying about licenses for technologies other than those embodied in the patents-in-suit, or licenses involving neither Sun nor Google” and also barring Prof. Cockburn from making “any reference at trial to the Sun-Microsoft settlement.” (Supp. Br. at 16.) It does so even though Google’s own damages experts rely heavily on other licenses involving Sun, most

17

notably the Sun-Danger license, as discussed above. (See Dearborn Decl. Ex. J (Excerpts from Leonard Report at pp. 45–49).)

As explained in Oracle’s opposition to Google’s original motion, the Sun-Microsoft litigation and resulting settlement is relevant. That litigation involved—as does this case—an incompatible, platform-specific Java implementation that undermined the core value of Java. Sun and Microsoft resolved that dispute after years of litigation with a set of agreements resulting in nearly $2 billion paid by Microsoft to Sun. That included a license for Microsoft to use Sun’s Java patents. Among other things, the Sun-Microsoft litigation and resulting settlement are relevant to Sun’s perspective with respect to any hypothetical license negotiation here—precisely the sort of “objective” historical evidence that Google elsewhere erroneously complains is missing from Prof. Cockburn’s analysis. Google cannot have it both ways.

Moreover, neither of the two cases cited by Google warrant an order precluding Prof. Cockburn from testifying about any other agreements. In ResQNet.com, Inc. v. Lansa, Inc., 594 F.3d 860, 869 (Fed. Cir. 2010), the court cited Lucent Technologies, Inc. v. Gateway, Inc., 580 F.3d 1301 (Fed. Cir. 2009) for the proposition that courts should be “vigilant” when considering licenses other than for patents in suit, but neither case forbids consideration or testimony regarding such other licenses. Prof. Cockburn appropriately heeds this guidance, not using any other licenses as the basis for any quantification of damages (unlike Google’s damages experts, who rely on a misreading of the Sun- Danger license as the sole basis for their quantification of the upward adjustment for incompatibility). He simply reviews those mobile licenses and finds that their terms underscore the reasonableness of his calculations. The Federal Circuit decisions on which Google relies involved situations where the expert actually relied on the other agreements in calculating damages, which is not the case here. See ResQNet, 594 F.3d at 870 (expert “used licenses with no relationship to the claimed invention to drive the royalty rate up to unjustified double-digit levels”); Lucent, 580 F.3d at 1329 (expert used licenses to calculate damages where it was “doubtful that the technology of those license agreements is in any way similar to the technology being litigated” in that case). Prof. Cockburn’s use of other licenses is well within the legal limits. There is no basis for exclusion of any or all reference to them.

18

CONCLUSION

The court should reject Google’s meritless Daubert challenge.

Dated: October 27, 2011

BOIES, SCHILLER & FLEXNER LLP

By: /s/ _______________
Steven C. Holtzman

Attorneys for Plaintiff
ORACLE AMERICA, INC.

19

690

UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
SAN FRANCISCO DIVISION

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

Case No. CV 10-03561 WHA (DMR)

ORACLE’S RESPONSE TO FINAL
PRETRIAL ORDER

Dept.: Courtroom 9, 19th Floor
Judge: Honorable William H. Alsup

Oracle submits this response to the Final Pretrial Order of January 4, 2012. Our comments are directed first to the plan as set forth in the Final Pretrial Order and then to an alternative that could move the case to resolution more expeditiously and reduce the burden on jurors.

I. COMMENTS ON PLAN IN FINAL PRETRIAL ORDER

A. Trial Date

Oracle wants this case to be tried as soon as possible. If March 19, 2012 is the Court’s earliest available date, then Oracle would strongly prefer trial to proceed on that day or as soon thereafter as possible. Google is inflicting irreparable harm on Oracle by offering proprietary Java technology, most notably the patents and copyrights in suit, for free within Android, fragmenting the Java platform. Oracle needs to obtain injunctive relief quickly to preserve the Java ecosystem and restore the “write once, run anywhere” promise that is the core of Java’s value. (See ECF No. 649 at 1-3.) Google has also said that it “is prepared to try the case at the earliest time available for the Court, the parties, and counsel.” (Id. at 3.)

Oracle notes, however, that one of the scheduling conflicts that prevented an earlier trial date has now disappeared. The electronic docket indicates that Mr. Van Nest’s trial date in New York v. Intel Corp., which was previously set for February 14-24, was vacated on December 23, 2011, two days after the pretrial conference in this case. The Intel case is now stayed pending the Court’s consideration of several pretrial motions (the trial date is subject to being reset). See Order Cancelling Trial, No. 09-827 (LPS) (D. Del. Dec. 23, 2011), ECF No. 298. Oracle’s counsel has no conflicts until late May 2012. Accordingly, if the Court’s schedule permitted, the trial could begin in February without conflicting with Mr. Van Nest’s other trials.

Oracle also notes that, if it is not possible for trial to begin on March 19 or earlier, then given the time limits set by the Court in its Final Pretrial Order, it would be possible to schedule the full trifurcated trial to take place between April 23 and May 30, when all counsel are free. The Court has allocated a total of 79 ¼ hours for jury selection, opening statements, presentation of evidence and the FJC video. (See Final Pretrial Order at 2-3, ECF No. 675.) Thus, under the Court’s current plan, adding in additional time for closing arguments and instructions, less than

1

19 days of trial presentation time is required. Adding in additional time for jury deliberations, it should be possible to complete the trial in less than the 27 court days that the parties have available from April 23 through May 30.

B. Trifurcation

Oracle previously opposed trifurcation and renews those objections here. (See ECF No. 627; ECF No. 649 at 1.)

Oracle expects that a trifurcated trial will continue to present issues over whether evidence is “relevant only to a later phase,” and therefore can be introduced only in the later phase under the terms of the Final Pretrial Order. For example, the Final Pretrial Order says that more general “Java” evidence may be introduced in Phase One or Two to the extent it constitutes a copyright (and presumably patent) liability issue or defense, including implied license, laches, equitable estoppel, and waiver. Willfulness is reserved for Phase Three. If Google’s equitable defenses are to be presented in Phase One and Two, willfulness should be presented in those phases as well because of the significant overlap among these issues and because Google’s willful infringement negates its equitable defenses as a matter of law. See, e.g., WeddingChannel.com, Inc. v. Knot, Inc., No. 03 Civ. 7369 (RWS), 2004 U.S. Dist. LEXIS 25749, at *7 (S.D.N.Y. Dec. 23, 2004) (“[T]he infringer’s state of mind can support a finding of egregious conduct, thereby defeating equitable defenses such as laches.”); ECF No. 627 at 6-8 (discussing overlap among willfulness, liability, and equitable defenses). Thus, we believe, but wish to clarify, that under the Court’s Order, willfulness evidence may be introduced in Phases One or Two to the extent it establishes an element of liability (e.g., in connection with inducement) or counters a defense that Google has indicated it will introduce at trial.

II. ALTERNATIVE PROPOSAL

In the wake of the pretrial conference, we have given additional thought to concerns the Court identified about the constraints on the Court’s schedule and the burden on the jury of sitting through a lengthy trial. We also wish to maximize the chances that we can be set for a trial earlier rather than later.

2

The injunctive phase of this case is critical to Oracle and is the most time-sensitive aspect of the litigation. As we explained at the pretrial conference, Oracle’s goal is to employ its right to injunctive relief to bring Android back into the Java fold and end the fragmentation in the Java platform and Java community that Google’s lawless conduct has generated. Millions of new Android devices appear every week, and new flavors of Android are being introduced. More Android devices have likely been sold in the last five months, since fact discovery closed, than in the preceding three years. According to Google Chairman Eric Schmidt, speaking yesterday at the Consumer Electronics Show in Las Vegas, Google is seeing 700,000 Android activations a day, totaling more than 200 million Android phones in the world today. “Just do a little math with the numbers I just told you, and Android is on a billion-unit plan.” (Laurent Goode, Schmidt – Storm Alert: The Google Chairman Didn’t Like Your Question, AllThingsD (Jan. 10, 2012), allthingsd.com/20120110/schmidt-storm-alert-the-google-chairman-didnt-like-your-question/.)

But Android’s adoption has moved far beyond mobile phone deployment. Its adoption in tablet computers is well-known, as perhaps is its use in televisions. Less well-known is Android’s spread into embedded systems, a traditional strongpoint for Java. Oracle’s sales force is seeing increasing evidence that Android is threatening Java in this area. That expansion is, of course, Google’s objective. See, e.g., Ina Fried, Google’s Rubin Says Ice Cream Sandwich Is Filled With “Butter”, AllThingsD (Oct 19, 2011), allthingsd.com/20111019/googles-rubin-saysice- cream-sandwich-is-filled-with-butter/:

While Andy Rubin is most focused on getting Android running on as many phones, TVs and tablets as possible, he sees opportunities well beyond that. Already, he notes, the operating system has spread to watches from Motorola and car dashboards from Mercedes. “Tell me what screen Android shouldn’t be on,” Rubin said, speaking to AllThingsD just after he finished his onstage appearance at AsiaD in Hong Kong.

(See also Tim Simonite, Google’s Android Rampages Beyond Mobile, Technology Review (May 10, 2011), www.technologyreview.com/blog/editors/26751/ (“Combine all that with other tech announced today that will have Android devices control everything in your home from lighting to irrigation...and it becomes difficult to think of that cute green robot representing a

3

mobile operating system any more. Rather, Android is being pitched as the operating system, for a future brought to you by Google.”).)

The more Android penetrates into the market, the more significant its forking of Java becomes and the harder it will be to achieve Oracle’s desired result. A speedy trial and entry of injunctive relief is thus ever more critical. We therefore offer the following proposal for the Court’s consideration:

• Phases One and Two would proceed as outlined in the Final Pretrial Order (with one exception as noted below). A single jury would hear both liability phases.
• If Oracle prevails in Phase One or Two, the Court would then turn to the question of injunctive relief. Except for this change in sequencing, the injunction phase would also be handled as outlined in the Final Pretrial Order.
• After the injunction phase, the Court would turn to the question of damages (Phase Three). Assuming that there is a break between the end of Phase Two and the beginning of Phase Three, a new jury would be selected to hear that phase.

This proposal would cut about one week from the duration of the initial trial, comprised of Phases One and Two. That would make sitting as a juror in that trial less burdensome. It would also give the Court greater flexibility in selecting a trial date, as only 14 days of trial presentation time, plus time for closing argument and deliberations, would be required for the initial trial. Hence, the Court may more readily be able to schedule the case for an earlier trial.

This proposal entails one modification to the Court’s delineation of Phases One and Two: the treatment of willfulness. We propose that the jury address copyright willfulness in Phase One and patent willfulness in Phase Two. Because, as noted previously, questions of intent are relevant to issues of liability, we believe this would in any case be a more efficient allocation of decision-making responsibility. But under our alternative proposal, such a reallocation would be necessary so that the damages jury does not have to hear extensive liability evidence in order to render a verdict on willfulness.

4

We believe this alternative approach offers considerable advantages in moving this case to resolution as quickly and efficiently as possible.

Dated: January 11, 2012

MORRISON & FOERSTER LLP

By: /s/ Michael A. Jacobs

MORRISON & FOERSTER LLP
MICHAEL A. JACOBS (Bar No. 111664)
[email]
KENNETH A. KUWAYTI (Bar No. 145384)
[email]
MARC DAVID PETERS (Bar No. 211725)
[email]
DANIEL P. MUINO (Bar No. 209624)
[email address telephone fax]

BOIES, SCHILLER & FLEXNER LLP
DAVID BOIES (Admitted Pro Hac Vice)
[email address telephone fax]
STEVEN C. HOLTZMAN (Bar No. 144177)
[email address telephone fax]

ORACLE CORPORATION
DORIAN DALEY (Bar No. 129049)
[email]
DEBORAH K. MILLER (Bar No. 95527)
[email]
MATTHEW M. SARBORARIA (Bar No. 211600)
[email address telephone fax]

Attorneys for Plaintiff
ORACLE AMERICA, INC.

5

691

UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
SAN FRANCISCO DIVISION

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

Case No. CV 10-03561 WHA

ORACLE’S RESPONSIVE BRIEF
REGARDING OBJECTIONS TO
GOOGLE’S PROPOSED CLAIM
CONSTRUCTIONS

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

INTRODUCTION

Google’s brief makes clear that its proposed constructions aim to avoid infringement. But the intrinsic evidence controverts Google’s narrowing proposals, which should be rejected.

ARGUMENT

I. OBTAIN A REPRESENTATION OF AT LEAST ONE CLASS FROM A
SOURCE DEFINITION PROVIDED AS OBJECT-ORIENTED PROGRAM
CODE (’720 PATENT)

>>>>

Claim	Term or Phrase	Oracle Proposed Construction	Google Proposed Construction
’720 patent, Claims 1 and 10	obtain[ing] a representation of at least one class from a source definition provided as object-oriented program code	No construction necessary. The phrase has the ordinary meaning that its constituent words give it.	load at least one class definition by compiling object oriented source code

Google’s brief perpetuates its error in equating “source” in the “obtaining” step of Claims 1 and 10 with the technical concept of “source code.” The ’720 patent uses the word “source” to mean “a point of origin or procurement,” which is its plain and ordinary meaning. MICROSOFT PRESS COMPUTER DICTIONARY 443 (3d ed. 1997) (definition of “source”: “In information processing, a disk, file, document, or other collection of information from which data is taken or moved.”) (Declaration of Marc Peters in Support of Oracle’s Objections to Google’s Proposed Claim Constructions (ECF No. 646) (“Peters Decl.”) Ex. A.)

Google’s proposed construction is directly contradicted by the ’720 specification. Claims 5 and 14 require that the “source definition” be maintained as a class file. But even Google acknowledges, as it must, that a “class file” is a compiled binary file rather than source code. (ECF No. 647 at 3.) Claims 5 and 14 are original claims and so part of the specification, forming the written description of the invention. They are not “anomalies” to be cast aside, as Google does without any legal authority. (Id. at 6.)

Google’s argument rests on specification language and dependent claims that refer, not to the “obtaining” step it seeks to construe, but to a different claim limitation: the “interpreting and instantiating” step, which is not at issue here. The first three steps of Claim 10 are reproduced here in full to illustrate this point (Claim 1 has analogous structural limitations):

1

10. A method for dynamic preloading of classes through memory space cloning of a master runtime system process, comprising:

executing a master runtime system process;

obtaining a representation of at least one class from a source definition provided as object-oriented program code;

interpreting and instantiating the representation as a class definition in a memory space of the master runtime system process; . . . .

Google’s reliance on a single reference to compilation in the body of the specification— “an instance of the class is created by compiling the source and the class instance is installed in the system class dictionary (block 160)”—is thus misplaced. (ECF No. 647 at 4.) This sentence is part of the written description of the “interpreting and instantiating” step and not of the “obtaining” step. This is confirmed by Figure 10 of the ’720 patent, in which step 157 is the “obtaining” step (“load bytes for class”) and step 160 is the “interpreting and instantiating” step (“create instance of class”). There is no description of the “obtaining” step in the specification that includes compiling source code.

Likewise, the dependent claims that Google cites in its brief do not limit the “obtaining” step to compiling source code. Google is wrong when it states (ECF No. 647 at 6) that the “class definition” in Claims 2 and 3 is a class file, in order to argue that a “source definition” may not be a class file. The complete phrase in Claims 2 and 3 is “instantiated class definition,” which refers to an already-interpreted and instantiated “class object to represent the class in memory” according to the specification. ’720, 6:54. This is the output of the “master runtime system process to interpret and to instantiate the representation as a class definition in a memory space of the master runtime system process” in Claim 1, which provides the antecedent basis for the phrase “the instantiated class definition” in Claims 2 and 3. The phrase “class definition” does not appear in the “obtaining” phrase that Google seeks to construe.

Nor does Claim 8 support Google’s proposed construction. Claim 8 requires that the “object-oriented program code is written in the Java programming language.” Of course, Java class files were originally written in the Java programming language and were created by a Java compiler, as the specification of U.S. Patent No. 7,213,240, which is incorporated by reference

2

into the ’720 specification, makes clear. (The ’240 specification refers to “class files” maintained on a server as being “programs written in a platform-independent language, such as Java,” as explained on pages 4-5 of Oracle’s Objections to Google’s Proposed Claim Constructions (“Oracle’s Opening Brief”) (ECF No. 645).) Thus, the specification confirms that the claimed “obtaining” step includes obtaining an already-compiled class file, not source code. ’240, 9:46- 60 (Peters Decl. Ex. C). Google’s proposed interpretation of unasserted Claim 8—that the reference to Java means “source definition” equals “source code”— is contrary to this disclosure. (ECF No. 647 at 6.)

Google’s argument that the plain meaning of the “obtaining” phrase fails to give meaning to all of the claim language and renders portions of it superfluous is incorrect. (ECF No. 647 at 6.) Whether Claim 1’s recitation of “class preloader” standing alone covers a traditional Java class loader configured to preload classes from class files is irrelevant. The inclusion of the phrase “to obtain a representation of at least one class from a source definition provided as objectoriented program code” aids in understanding the claim, and that is a sufficient purpose. Bell & Howell Document Mgmt. Prods. Co. v. Altek Sys., 132 F.3d 701, 707 (Fed. Cir. 1997) (“[D]efining a state of affairs with multiple terms should help, rather than hinder, understanding. Being ‘integrally bonded’ and ‘free of adhesive’ are mutually reinforcing definitions rather than being superfluous.”); Infosint, S.A. v. H. Lundbeck A/S, 603 F. Supp. 2d 748, 757 (S.D.N.Y. 2009) (when specification and claims make construction of word clear, use of additional words “to help not hinder understanding” is not superfluous). The words in the “to obtain…” phrase in Claim 1 also serve the purpose of providing an antecedent basis for further claim limitations in dependent Claims 5 and 8; no dependent claim otherwise further limits “class preloader.”

Google’s two cases on “superfluous” claim language (Merck and Haemonetics) are distinguishable. The Federal Circuit rejected the district court’s constructions because they were directly contrary to the claim language and were not the ordinary and customary meaning as used in the specification. Haemonetics Corp. v. Baxter Healthcare Corp., 607 F.3d 776, 781 (Fed. Cir. 2010) (when claim recited “centrifugal unit comprising a centrifugal component and a plurality of tubes” and specification disclosed corresponding embodiment, “centrifugal unit” could not be

3

construed to lack tubes); Merck & Co. v Teva Pharms. USA, Inc., 395 F.3d 1364, 1372 (Fed. Cir. 2005) (“about” had its ordinary meaning of “approximately,” and not “exactly” as construed by district court, because patentee had not acted as lexicographer). In cases where the proper construction is the ordinary definition, such as in Bell & Howell and Infosint, additional claim terms with reinforcing meanings are “helpful” and not superfluous. It is also significant that the phrase “class preloader” does not appear in Claim 10, and so Google’s “superfluous” argument cannot apply to it.

Google’s proposed construction is based only on extrinsic materials beyond the specification. The phrase “source code” appears nowhere in the ’720 patent. It was known in the art that bytecode (contained in class files) could be compiled. See ’205, 2:1-13 (filed before the ’720). When the ’720 refers to compiling, it just as much refers to compiling bytecode as it does to compiling source code. The specification states that the exemplary routine for preloading classes will first “load the bytes for the class from the source associated with the applicable bootstrap class loader 39 and system application class loader 40” (’720, 9:49-51) and only afterwards “an instance of the class is created by compiling the source” (’720, 9:52-54). The use of the word “bytes” means that the representation of the class is in the form of bytecodes (class files), not source code. See Java Virtual Machine Specification Ch. 4 (1996) (“This chapter describes the Java Virtual Machine class file format. . . . A class file consists of a stream of 8-bit bytes.”), available at http://java.sun.com/docs/books/jvms/first_edition/html/ClassFile.doc.html. Google has provided no evidence that persons of ordinary skill in the art ever used the word “bytes” as shorthand for “source code.”

It bears repeating that Google’s argument to the Court is directly contrary to the position it took in the ’720 inter partes reexamination. In the PTO, Google argued that class files were the claimed “source definition provided as object-oriented program code.” (Peters Decl. Ex. D at 22 & Ex. 17.) But here, Google takes the opposite position and argues that a class file cannot possibly be the “source definition provided as object-oriented program code” from which a class preloader obtains a representation of a class when the class is to be loaded. Google has yet to inform the PTO that it changed its position.

4

There is no need to construe the phrase “obtain a representation of at least one class from a source definition provided as object-oriented program code,” which has the ordinary meaning of its constituent words. Obtaining a class representation from a source definition includes “identifying a binary form of a class type as identified by specific name.” ’720, 6:49-50 (emphasis added). Rejecting Google’s attempt to narrow this term does not require the Court to rephrase it for the jury. At most, the Court should instruct the jury that a “source definition” may be in binary or source code form. The ordinary meaning of the term embraces both.

II. RUNTIME (’205 PATENT)

Claim	Term or Phrase	Oracle Proposed Construction	Google Proposed Construction
’205 patent, Claim 1	runtime	No construction necessary. The ordinary meaning is “during execution of the virtual machine.” (per 2/22/2011 Joint Claim Construction Statement)	during execution of the virtual machine instructions (per 2/22/2011 Joint Claim Construction Statement)

Google attempts to redefine “runtime” by taking only a narrow slice of runtime— execution of “the” virtual machine instructions—and excluding everything else that occurs at runtime. This is not supported by the specification or the claim language.

Nothing in Claim 1 of the ’205 patent limits “runtime” to the execution of any particular virtual machine instructions. The preamble of Claim 1, which describes the context for the invention but does not limit it, states that the invention is “a method for increasing the execution speed of virtual machine instructions at runtime.” Oracle agrees that when virtual machine instructions are executed, it is at runtime. But the inverse is not true: many things may happen at “runtime” other than the execution of particular virtual machine instructions. Indeed, Claim 1 claims an invention when the “generating, at runtime” and “executing” steps are separate steps:

1. In a computer system, a method for increasing the execution speed of virtual machine instructions at runtime, the method comprising:

receiving a first virtual machine instruction;

generating, at runtime, a new virtual machine instruction that represents or references one or more native instructions that can be executed instead of said first virtual machine instruction; and

5

executing said new virtual machine instruction instead of said first virtual machine instruction.

See Phillips v. AWH Corp., 415 F.3d 1303, 1314 (Fed. Cir. 2005) (en banc) (“To begin with, the context in which a term is used in the asserted claim can be highly instructive.”). All that Claim 1 requires is that the system generate a new virtual machine instruction before it executes the new instruction. The claim does not require that the generation be performed during the execution of “the” virtual machine instructions, or any virtual machine instruction or instructions, for that matter. The generation of new instructions could be performed when a class is loaded into a virtual machine (which happens at runtime) and before any instructions of the class are executed. The only execution that the claim requires follows the “generating, at runtime” step.

The specification of the ’205 patent does not limit “runtime” to the execution of any particular virtual machine instructions. Code execution does occur at runtime, but the ’205 patent makes clear that native machine instructions may be executed:

In general, embodiments of the present invention provide innovative systems and methods for increasing the execution speed of computer programs executed by an interpreter. A portion of a function is compiled into at least one native machine instruction so that the function includes both virtual and native machine instructions during execution.

’205, 2:35-40 (emphasis added). The compiling performed by this embodiment need not occur during execution of the virtual machine instructions; it may occur before they are executed, so that when execution does occur, both virtual and native instructions are executed. This is the best understanding of the embodiment, because when the compiler is executing, the virtual machine instructions are not—and yet it is still “runtime” under the ordinary meaning of the word. See also ’205, Abstract (“A portion of the virtual machine instructions of the function are compiled into native machine instructions so that the function includes both virtual and native machine instructions. Execution of the native machine instructions may be accomplished by overwriting a virtual machine instruction of the function with a virtual machine instruction that specifies execution of the native machine instructions.”) (emphasis added).

Other examples of activities performed at runtime disclosed in the specification that are not execution of “the” virtual machine instructions include initialization, class loading,

6

verification, compiling, linking, bytecode table generation, and snippet generation, as explained in Oracle’s Opening Brief (ECF No. 645 at 10-13). When a program implementing the inventive method of Claim 1 performs the “generating” step, it is at runtime, but “the” virtual machine instructions are not being executed. Instead, the instructions of the program are being executed. Google’s argument thus fails because it is contradicted by the specification.

Google’s proposed construction is ambiguous because it leaves open the question: which are “the” virtual machine instructions? They are not in the body of the claim, which recites only “a first virtual machine instruction” which is never executed (“executing said new virtual machine instruction instead of said first virtual machine instruction”). The Court should reject a construction that invites additional questions and risks having to “construe the construction.” (So the Court is aware, the reason for the ambiguity is that Google wants to avoid its admissions in the Android documentation that its accused “dexopt” program requires information “only available at runtime” and argue that “runtime” doesn’t mean “runtime.”)

Google’s citations to the specification in its brief stand only for the noncontroversial proposition that when virtual machine instructions are being executed, it is runtime. (ECF No. 647 at 8 (citing ’205, 1:17-21 (“porting an existing computer program to run on a different computer platform”); 1:47-51 (“An advantage of utilizing virtual machine instructions is the flexibility that is achieved since the virtual machine instructions may be run, unmodified, on any computer system that has a virtual machine implementation, making for a truly portable language.”); 9:55-57 (“the faster the program will run”)).) But Google’s citations do not stand for the proposition that the ’205 patent has redefined “runtime” to only occur when some particular yet unspecified virtual machine instructions are being executed. The patentee did not act as a lexicographer to give the general term “runtime” such a limited meaning, and the ordinary meaning controls.

CONCLUSION

For the foregoing reasons, Oracle respectfully requests that the Court decline to construe the two terms in dispute.

7

Dated: January 11, 2011

MICHAEL A. JACOBS
MARC DAVID PETERS
DANIEL P. MUINO
MORRISON & FOERSTER LLP

By: /s/ Michael A. Jacobs

Attorneys for Plaintiff
ORACLE AMERICA, INC.

8

692

UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
SAN FRANCISCO DIVISION

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

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

GOOGLE’S RESPONSE AND
OBJECTIONS TO FINAL PRETRIAL
ORDER

Judge: Hon. William Alsup
Dept. Courtroom 9, 19th Floor

INTRODUCTION

On January 4, 2012 the Court issued a Final Pretrial Order (Dkt. No. 675) and directed the parties to submit any objections to that Order within seven calendar days. Google has no objections to the Court’s Final Pretrial Order, other than the following.

OBJECTIONS

A. The Court should allot more time for the patent and damages phases of the trial.

Google previously requested 15 hours of trial time per side for each phase of the trial.¹ The Court’s Order allocates 16 hours for the copyright phase, 12 hours for the patent phase, and 8 hours for the damages/willfulness phase. Google renews its request for a total of 30 hours per side for the liability phases, and requests that the Court allow 12 hours per side for the damages/willfulness phase of the trial.

The patent case—at least in its current form—cannot be tried in a reasonable amount of time, let alone in 12 hours of trial time per side. Oracle intends to try 26 claims of six unrelated patents. It has identified 20 percipient witnesses on its “will call” list as relevant to the patent case.² Google will call other percipient witnesses, and, together, the parties will need to call more than a half-dozen expert witnesses. The parties would need far more than 12 hours per side to try that patent case. Oracle must narrow its claims to a manageable number for trial in order to be able to present a comprehensible case in the limited time available.

Even if the patent case could be tried in a reasonable amount of time, there is no reason to expect that it will take less trial time than the copyright case. The parties will need to explain to the jury what the patents are, how the claim terms have been construed, and what the jury needs to know to determine infringement and validity. In order to do that, the parties will also need to educate the jury on other complex issues that are specific to the patents (and not implicated by the copyright claims), such as the difference between source code and byte code, the operation of the Dalvik virtual machine, the dx tool, the just-in-time compiler (among other technology), and an explanation of the asserted claims of the six patents-in-suit. Presentation of this background

___________________________________

¹ Second Supplemental Joint Pretrial Conference Statement (Dkt. No. 649) at 4:22-5:2.

² Oracle Witness List (Dkt. No. 491); Oracle Critique of Proposed Trial Plan (Dkt. No. 627).

1

material will take a significant amount of time. Accordingly, the Court should allot two more hours for the patent phase, for a total of 30 hours per side for the liability phases.

The Court should also allot an additional four hours per side for the damages/willfulness phase. There too Oracle has identified at least 20 percipient witnesses from its “will call” list.³ The damages/willfulness phase also will require testimony from five expert witnesses, including Dr. Kearl. Direct and cross examination of these expert witnesses alone likely will take close to six hours, and the parties will need time to lay the foundation through percipient witness testimony for various facts relied upon by the experts. Moreover, the parties will need to present evidence on the issue of willfulness. Accordingly, the Court should allow 12 hours per side for the damages phase. Cf. General Signal Corp. v. MCI Telecomm. Corp., 66 F.3d 1500, 1508 (9th Cir. 1995) (“[A] district court may impose reasonable time limits on a trial.”) (emphasis added)).

B. The Court should allow the parties to videotape post-trial-day testimony of third-
party witnesses and play that testimony as necessary in subsequent phases.

Google agrees with the Court’s Order to the extent that it prohibits the parties from videotaping testimony during phase one or two of the trial and replaying that same testimony again in a later trial phase. Google requests, however, that the Court allow the parties to videotape certain third-party testimony presented to the Court—after the jurors have been excused for the day—and play the non-duplicative testimony for the jury in later trial phases, as necessary. This procedure would minimize the burden that a two-month trifurcated trial would impose on third-party witnesses and ensure that the court and the jury hear all of the relevant evidence.

Both Google and Oracle intend to call third-party witnesses whose testimony will be relevant to more than one phase of the trial. These witnesses include: Jonathan Schwartz, Vineet Gupta, John Rizzo, and Bob Lee. Some of the topics covered by these witnesses may be relevant to one liability phase but not to the other, or to either liability phase but not to the damages phase. For example, Mr. Schwartz likely will testify about Sun’s practices regarding

___________________________________

³ Oracle Witness List (Dkt. No. 491); Oracle Critique of Proposed Trial Plan (Dkt. No. 627).

2

the use of Java APIs—an issue relevant only to the copyright phase.⁴ But Mr. Schwartz will also testify about Sun’s established policy of using its patents only defensively—an issue relevant only to the patent and damages/willfulness phases. Because the Court’s Order prevents witnesses from “giv[ing] testimony on a subject relevant only to a later phase,”⁵ Mr. Schwartz would have to appear once to provide copyright-specific testimony, again to provide patentspecific testimony, and again to provide damages/willfulness-specific testimony. This is an unreasonable burden to place on a nonparty. Indeed, some third-party witnesses might be unavailable to appear for successive trial phases, thereby depriving the court and the jury of relevant evidence and prejudicing the parties’ ability to present their cases.

The Court could eliminate these problems by adopting the following procedure:

1. Allow third-party witnesses who appear in an early trial phase to testify before the court—out of the presence of the jury—on topics relevant only to a later trial phase.

2. Videotape that testimony, and play it for the jury in later phases, as necessary.

This procedure respects the structure of a trifurcated trial, while minimizing the burden on third-parties and ensuring that the court and the jury hear all of the relevant evidence. The Court should adopt it. See F&G Scrolling Mouse, LLC v. IBM Corp., 190 F.R.D. 385, 387 (M.D.N.C. 1999) (noting that trial judge should consider the “convenience to the parties, witnesses, jurors, and the court” in a phased trial); see also Ciena Corp. v. Corvis Corp., 210 F.R.D. 519, 521 (D.Del. 2002) (noting that trial judge should consider possible prejudice to the parties in a phased trial).

Dated: January 11, 2012

KEKER & VAN NEST LLP

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

__________________________________

⁴ See generally, Google’s Trial Witness Disclosure Pursuant to Fed. R. Civ. P. 26(A)(3).

⁵ Order at 3:17.

3

693

UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
SAN FRANCISCO DIVISION

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

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

Honorable Judge William Alsup

GOOGLE’S REPLY TO ORACLE’S
PROPOSED CLAIM CONTRUCTIONS

Pursuant to the December 21, 2011 Interim Pretrial Conference (Dkt. No. 654 at 154:12– 155:20), Google hereby submits this response to Oracle’s supplemental claim construction brief filed December 9, 2011 (Dkt. No. 645) (“Oracle Br.”). Per the parties’ agreement (Dkt. No. 654 at 154:20–155:20), Google relies on its prior briefing with respect to the “computer-readable medium” term of claim 14 of the ‘476 patent. (See Dkt. Nos. 96–97, 102–03.)

A. The ‘720 Patent: “obtain a representation . . . from a source definition . . .”

Oracle’s arguments rely on two fundamentally flawed premises: (i) that Google’s proposed definition reads out the preferred embodiment, and (ii) that a cherry-picked, nontechnical dictionary definition overrides the patent specification. Neither is true. In fact, Oracle’s proposed construction at best renders the claim language superfluous, and at worst is in conflict with the intrinsic evidence from the specification.

1. Google’s—and not Oracle’s—proposed construction captures the preferred
embodiment by including a class loader capable of compiling source code.

Oracle incorrectly claims that Google and its expert “rely on a misreading of a sentence fragment in the ‘720 specification” to arrive at a construction that “excludes the preferred embodiments.” (Oracle Br. at 5:21–22, 5:13–14.) But Google’s proposed construction does not exclude systems that preload Java class files produced by a Java compiler; it explicitly includes them. Where Google’s proposed construction differs, however, is that it gives meaning to the requirement that “at least one” of those classes be “from a source definition provided as object oriented program code.” In other words, source code. As such, Google’s proposed construction includes the preferred embodiments disclosed in column 9 of the ‘720 patent specification.

Oracle concedes that the operation of the preferred preloader is described in column 9 of the ‘720 patent under the heading “Routine for Preloading Classes,” which identifies three mechanisms for preloading classes. (See Dkt. No. 647, Google Br. at 4:16–5:7.) The preloader first attempts to locate “the class” in the system class dictionary. (‘720 patent at 9:42–43.) If not found in the system class dictionary, the preloader then looks for “the class” in the file system. (Id. at 9:45–47.) If not found in the file system, then the preloader “attempts to load the bytes for the class from the source associated with the [core class loaders]” and creates “an instance of the

1

class . . . by compiling the source . . .”¹ (Id. at 9:48–54.) It is this source code compilation process² that is captured by the express claim language that refers to “obtaining a representation of at least one class from a source definition provided as object oriented program code.”

Oracle mistakenly contends that the phrase “compiling the source” refers to compiling Java bytecode (which is not source code) with a just-in-time (JIT) compiler. (Oracle Br. at 6–7.) This argument is made of whole cloth: the ‘720 patent contains neither discussion of nor reference to JIT compilation. This argument fails for at least two technical reasons. First, because classes are preloaded into the master process before they are actually required by a user application in a child process (‘720 patent at 2:66–3:1, 10:5–14), compilation by the preloader is speculative and well in advance of execution. Put another way, because the preloader performs the step of “compiling the source,” the patent addresses an approach wholly different from a JIT compiler that “compiles an entire Java function just before it is called,” i.e., during execution. (Oracle Br. at 6:24 (emphasis added).) Second, the ‘720 patent only uses the term “compiling” in conjunction with one of three loading mechanisms enumerated in the ‘720 patent. (See, e.g., ‘720 patent at 9:48–54.) Oracle’s construction does not address (but rather obfuscates) this aspect of the claimed system. In contrast, Google’s proposed construction provides a clear explanation: two mechanisms load classes from precompiled bytecode while the third loads classes from source code that must first be compiled.

2. Oracle’s proposed definition conflicts with the specification.

Oracle presents two extrinsic dictionary definitions of “source” in opposition to Google’s proposed construction, neither of which can be correct in the context of the patent. First, Oracle advances the second of several definitions for “source” provided by Webster’s dictionary: “a point of origin or procurement.” (Oracle Br. at 1:25–27.) This definition renders the preferred

_______________________________

¹ Oracle argues that “bytes” indicates that the contents of “the source” must be bytecode. (Oracle Br. at 6:9–15.) This argument is specious: “bytes” simply refers to a unit of storage, and both source and object code are represented as bytes in a file. (See Fenton Decl. Ex. A.)

² The straightforward idea of a source code compiling loader was explained in a contemporary Java tutorial published by IBM in 2001. (Fenton Decl. Ex. B at 7 (“If the class is not available, but the source is, call the Java compiler to generate the class file.”).)

2

embodiments meaningless—if we replace the word “source” with Oracle’s proposed definition of that word, then the sentence in the specification discussing compilation becomes nonsensical: “If successful (block 158), an instance of the class is created by compiling the > and the class instance is installed in the system class dictionary (block 160).” (‘720 patent at 9:51–54.) The same is true of substituting the definition of “source” found in the Microsoft dictionary (“a disk, file, document, . . .”), also advanced by Oracle. In contrast, Google’s definition of “source code” can be meaningfully substituted into the ‘720 patent specification for “source”: “If successful (block 158), an instance of the class is created by compiling the > . . .” Thus, “source definition provided as object oriented program code” must refer to object-oriented source code.

3. Oracle’s proposed construction renders the phrase at issue redundant.

As noted above, Oracle argues that the word “source” simply means “a point of origin or procurement.” (Oracle Br. at 2.) Because the claim limitation recites “obtaining a representation of at least one class from a source definition,” it is difficult, if not impossible, to imagine any system that “obtains a representation of at least one class” from anywhere other than the “point of origin or procurement” of that class. As such, Oracle’s proposal renders the claim language meaningless, and cannot be correct. See Curtiss-Wright Flow Control Corp. v. Velan, Inc. 438 F.3d 1374, 1379–80 (Fed. Cir. 2006). The claim-at-issue in Curtiss-Wright was drawn to a mechanical device, and the claim term “adjustable” was construed according to its ordinary meaning. See id. But in reviewing the district court’s claim construction, the Federal Circuit noted that it was “difficult, if not impossible, to imagine any mechanical device that is not ‘adjustable,’ under the ordinary meaning of that term adopted by the district court.” Id. While the Federal Circuit “commend[ed] the district court’s reluctance to narrow the claims,” it ultimately found the construction flawed because it “renders that [‘adjustable’] limitation meaningless.” Id. The same is true here: a preloader by its very nature finds classes at “a point of origin or procurements.” Thus construing the phrase “from a source definition provided as object oriented program code” as Oracle suggests adds nothing. In contrast, Google’s proposed definition gives meaning to the additional language in such a way that captures the preferred

3

embodiment—runtime compilation of source code when classes have not yet been created.

4. Oracle’s claim differentiation argument relies on a specific definition of the
term “class file” that is unsupported by the intrinsic or extrinsic record.

Oracle insists that when claim 5 requires a file system to “maintain the source definition as a class file,” the claim requires that the “class file” be “a compiled binary file satisfying the Java Virtual Machine Specification.” (Oracle Br. at 3:10–11.) But this “requirement” is not part of the claim, and Oracle provides no intrinsic or extrinsic evidence to support such a narrow interpretation of the phrase “class file.” The ‘720 patent neither defines the term “class file” nor incorporates the Java Virtual Machine Specification by reference, and refers to Java technologies only as “[a]n example of a suitable managed code platform . . .” (‘720 patent at 3:6–10.)

Oracle’s argument regarding graphic icons does not alter this result as Oracle is simply incorrect about the meaning of those icons. The icon for classes 36, a box with a wavy bottom, merely represents a document, or a file. (See Fenton Decl. Ex. C at 123, 144; Ex. D (showing a flowcharting template with the identical icon in a 2003 drafting program).)

B. The ‘205 Patent: “runtime”

Oracle’s “ordinary meaning” construction of the term “runtime” ignores both the plain language of claim 1 and the entirety of the specification, and relies on unsupported and inaccurate technical conclusions. Oracle’s attempt to expand the scope of the claims of the ‘205 patent beyond anything contemplated or disclosed by the patentee should be rejected.

1. Oracle presents no evidence of an “ordinary meaning.”

Oracle provides no evidence that its proposed “ordinary meaning” construction reflects the common meaning of the term “runtime.” (Oracle Br. at 9.) Indeed, the meaning of that term depends on the context in which it is used. Oracle fails in its five and a half pages of briefing to explain what a person of ordinary skill would understand the term to mean in the context of the ‘205 patent. Oracle provides only its present-day, litigation-driven interpretation of the term divorced from the context provided in both the claim itself and the specification. See Phillips v. AWH Corp., 415 F.3d 1303, 1313 (Fed. Cir. 2005) (ordinary artisan deemed to read the claim term not only in the context of the claim, but in the context of the entire patent).

4

2. The language of claim 1 does not support Oracle’s construction.

On the one hand, Oracle’s proposed construction disregards the preamble of claim 1, which explicitly refers to “increasing the speed of virtual machine instructions at runtime.” (See ‘205 patent, claim 1; see also Google Br. at 7–8.) On the other hand, Google’s proposal looks to the preamble as the only other intrinsic use of the word “runtime.” As such, the preamble provides insight into what the patentee intended when limiting the “generating” step to occur “at runtime,” namely, during execution of the virtual machine instructions. See Pitney Bowes, Inc. v. Hewlett-Packard Co., 182 F.3d 1298, 1306 (Fed. Cir. 1999) (where meaning of claim term “is only discernable from the claim preamble . . . , it is essential that the court . . . construe the preamble and the remainder of the claim . . . as one unified and internally consistent recitation”). Notably, Google does not rely on the preamble as a limitation itself, as the term runtime is an explicit limitation of the “generating” step. (Oracle’s reliance on Catalina Mktg. is therefore misplaced. (See Oracle Br. at 10.))

In disregarding the preamble, Oracle’s construction renders the “at runtime” phrase meaningless. Oracle argues that “all that is required [in the ‘generating, at runtime’ step] is the generation of a new virtual machine instruction ‘that can be’ executed instead of the first virtual machine instruction.” (Oracle Br. at 10.) If Oracle’s argument is correct, then the “at runtime” phrase in this limitation is meaningless. This cannot be the case. Curtiss-Wright, 438 F.3d at 1379 (court should not construe a term in such a way that renders it meaningless).

3. The specification does not support Oracle’s construction.

In its various citations to the ‘205 patent (and to other patents-in-suit), Oracle fails to show a single instance where a new virtual machine instruction is generated at a time other than when the virtual machine is executing virtual machine instructions. In fact, the specification consistently discloses the generation of a new virtual machine instruction during the execution of instructions of a program. In particular, the patent describes the “go_native” instruction as the new virtual machine instruction. (E.g., ‘205 patent at 7:23–26, 9:30–34.) The go_native instruction is generated whenever the interpreter decides to substitute a sequence of bytecodes with optimized native machine instructions, which the ‘205 patent calls a “snippet.” (E.g., ‘205

5

patent at 6:26–30 (“virtual machine instructions . . . may be compiled into native machine instructions in the form of a ‘snippet’”); 7:49–57 (“the interpreter decides when to substitute a sequence of bytecodes with a snippet” and generates a go_native bytecode that specifies the generated snippet).) The specification is explicit that this process occurs during execution of the virtual machine instructions of the then-executing program. (See ‘205 patent at 9:30–34 (“As snippets are generated during program execution, the new virtual machine instruction is executed . . .” (emphasis added)); 7:49–51 (“During execution of an interpreted program, the interpreter decides when to substitute a sequence of bytecodes with a snippet . . . [and then generate] a go_native bytecode . . . specifying the snippet.” (emphasis added)).)

In fact, nowhere does the patent disclose generating a new virtual machine instruction at a time other than during program execution. Thus, this case is similar to Edwards Lifesciences LLC v. Cook Inc., 582 F.3d 1322 (Fed. Cir. 2009). The issue in Edwards was whether the term “graft” was limited to “intraluminal grafts.” Id. at 1328–29. The court held that it was, reasoning, inter alia, that “the only devices described in the specification are intraluminal, supporting an interpretation of [the term ‘graft’] that is consistent with that description.” Id. at 1329. So too here. The generation of the new go_native virtual machine instruction is consistently described as occurring while the program being optimized is in execution. Thus, this Court should construe “runtime” to be consistent with that description.

Finally, Oracle misreads the description of Figure 13 and incorrectly concludes that generation of a new virtual machine instruction is occurring during the disclosed initialization. (See Oracle Br. at 12 (quoting ‘205 patent at 12:15–26 and concluding that this passage “describes the technique of generating a new virtual machine instruction . . . while the virtual machine is up-and-running, before executing any virtual machines.”).) To begin with, Oracle fails to explain why Figure 13 allegedly supports its proposed construction, as it is unclear why the “initialization” described with respect to Figure 13 should be considered “execution” of the virtual machine. In any event, the ‘205 patent describes Figure 13 as a bytecode table that is preferably generated “once when the Java virtual machine is initialized.” (‘205 patent at 13:3– 8.) But the patent goes on to say that “[t]he bytecode table may include a pointer to snippet code

6

1061 for each bytecode to which native machine instructions will be generated.” (‘205 patent at 13:15–17 (emphasis added).) Thus, while the table itself may be generated when the virtual machine is initialized, the native machine instructions of the snippet “will be generated” at a later time. As described above, the specification is clear that snippet generation—and the concurrent go_native generation—occurs during program execution. (See ‘205 patent at 9:30– 34; 7:49–51.) Thus, Figure 13 and its description in the patent does not support Oracle’s proposed construction of the term “runtime,” but rather is consistent with Google’s proposed construction where runtime is “during execution of the virtual machine instructions.”

4. Oracle’s factual assertions are unsupported and inaccurate.

Oracle misses the point by proclaiming that the dispositive question is whether “the patent’s claimed invention cover[s] the situation where the virtual machine is executing but is not necessarily executing virtual machine instructions.” (Oracle Br. at 10.) This is a distinction without a difference: a virtual machine is executing when it executes virtual machine instructions. None of the evidence on which Oracle relies is to the contrary. For example, while Oracle cites to the ‘702 patent’s description of the class initialization as an alleged example of when a virtual machine is executing but not executing virtual machine instructions (Oracle Br. at 12), the Court need look no further than another patent-in-suit, the ‘520 patent, to realize that this process involves executing virtual machine instructions.

Specifically, the ‘520 patent discloses that “the Java compiler generates a special method, , to perform class initialization, including initialization of static arrays” (‘520 patent at 1:59–61 (emphasis added)), and then goes on to disclose the very bytecode (i.e., virtual machine instructions (see Oracle Br. at 11 (bytecode is synonymous with virtual machine instructions))) that the Java virtual machine executes to perform the class initialization. (‘520 patent at 2:26–53 (Code Table #3).) Thus, Oracle’s initialization example actually supports Google’s proposed construction;³ it proves that an executing virtual machine executes virtual machine instructions.

____________________________________

³ Oracle’s other examples (class loading, etc.) also involve execution of virtual machine instructions and support Google’s construction. (See, e.g., Fenton Decl. Ex. B at 2 (“The Java ClassLoader, furthermore, is written in the Java language itself.”).)

7

DATED: January 11, 2012

KEKER & VAN NEST, LLP

By: /s/ Christa M. Anderson

ROBERT A. VAN NEST (SBN 84065)
[email]
CHRISTA M. ANDERSON (SBN 184325)
[email]
KEKER & VAN NEST LLP
[address telephone fax]

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

DONALD F. ZIMMER, JR. (SBN 112279)
[email]
CHERYL A. SABNIS (SBN 224323)
[email]
KING & SPALDING LLP
[address telephone fax]

IAN C. BALLON (SBN 141819)
[email]
HEATHER MEEKER (SBN 172148)
[email]
GREENBERG TRAURIG, LLP
[address telephone fax]

ATTORNEYS FOR DEFENDANT
GOOGLE INC.

8