Foregoing a hearing on the matter, yesterday Judge Alsup issued an order (512 [PDF; Text]) on Google's pending motion to strike portions of the Mitchell Patent Report
(473 [PDF; Text], including the reply report. Although Google did not win on every point, Judge Alsup largely upheld Google's motion.
There were four primary points that Google was seeking to strike:
- To Demonstrate Infringement of the ‘720 patent, the Mitchell Report Adds New Theories Regarding Android’s Use of The Functions vfork() and clone() - specifically, the vfork() and clone() functions;
- The Mitchell Report Improperly Adds A New “mBS Mobile” Infringement Theory for the ‘476 Patent - specifically, a theory predicated on mBS Mobile, which is a third party application not provided by Google in Android;
- The Mitchell Reply Report Adds a New Infringement Theory Regarding the ‘476 Patent - specifically, that Android infringes the Java security framework even if SecurityManager is disabled (which it is in Android); and
- The Mitchell Reply Report Adds New Infringement Theories Regarding the ‘520 Patent - specifically, (a) that the dx tool can infringe the ‘520 patent claims by “creat[ing] (and stor[ing]) one or more instructions requesting the static initialization of the array, and (b) that an unsuccessful simulated execution can somehow infringe the ‘520 patent, meaning that the claimed step of “simulating execution” doesn’t need to identify a series of instructions for initializing individual elements of the array.
To be clear, this is Google's second motion to strike portions of the Mitchell Patent Report, Google having won about half of its first round of requests to strike. (Orders, Orders, Orders; 464 [PDF; Text])
In this case Judge Alsup:
- Granted the request to strike where the Mitchell Patent Report "identif(ies) vfork() and clone() as satisfying the “process cloning mechanism” limitation of claim 6, but not with respect to those same functions "satisfying the “copy-on-write process cloning mechanism” limitation of claim 1 and dependent claim 6." In the latter case, Judge Alsup denied the request only because Google didn't show that the Mitchell report actually made this contention.
- Denied the request with respect to mBS Mobile, disagreeing that Oracle advanced "a new theory of infringement based on the mBS Mobile application."
- Granted Google's request to strike the references to SecurityManager and AccessController; and
- Granted Googles request to strike the reference "that the accused elements create “one or more” instructions amounts to a new infringement theory."
Each of these wins will help restrain the infringement contentions within the limits of what Oracle originally claimed.
In the other business of the day Oracle submitted a letter
(511 [PDF]) seeking leave to file a motion or motions to strike and exclude portions of the Google damages reports prepared by Dr. Leonard and Dr. Cox. This request comes at a point after which all motions are to have been filed, but in fairness to Oracle, the Google damages reports were not made available until that deadline had passed. It seems this was a bit of a loophole in the various deadlines, as Oracle does not contend that Google filed the reports late.
Oracle's principle arguments in seeking to strike portions of the two reports are: (1) the two doctors lack the expertise to form the technical opinions they offer; and (2) the two doctors relied on testimony and documents not timely (or ever) made available to Oracle during discovery. Of course, this is just Oracle's side of the story, but if these statements ring true to Judge Alsup, it would be surprising if Oracle was not permitted to move forward with its motion, even though the trial date is rapidly approaching.
511 – Filed and Effective: 10/10/2011
Document Text: Letter from Steven C. Holtzman Precis re Daubert Motions. (Holtzman, Steven) (Filed on 10/10/2011) (Entered: 10/10/2011)
512 – Filed and Effective: 10/11/2011
Document Text: ORDER PARTIALLY GRANTING AND PARTIALLY DENYING SECOND MOTION TO STRIKE PORTIONS OF MITCHELL REPORT AND VACATING HEARING by Judge Alsup granting in part and denying in part 473 Motion to Strike (whalc1, COURT STAFF) (Filed on 10/11/2011) (Entered: 10/11/2011)
[Boies, Schiller & Flexner LLP letterhead]
October 10, 2011
The Honorable William Alsup
U.S. District Court, Northern District of California
450 Golden Gate Avenue
San Francisco, California 94102
Re: Oracle America, Inc. v. Google Inc., No. C 10-3561 WHA
Dear Judge Alsup:
On October 3, 2011, two days after the Court’s deadline for motions in limine, Google
served the reports of its damages experts, Dr. Gregory K. Leonard (patent) and Dr. Alan J. Cox
(copyright) (the “Reports”). Oracle now respectfully requests leave to file motions, pursuant to
Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579 (1993) and Rules of Evidence 401, 402, 403,
702, and 703, to strike portions of the Reports and exclude testimony on the stricken opinions.
The Reports contain technical opinions that Drs. Leonard and Cox are not qualified to proffer;
rely on spoon-fed “facts” from last-minute interviews; rely on documents and witnesses that
Google refused to produce or disclose; rely on the experts’ ipse dixit instead of evidence; and are
inconsistent with established damages principles.
Neither Dr. Leonard nor Dr. Cox holds any degree in computer science or any other
technical field. Yet both Reports rely heavily on opinions, based on technical arguments, that
“Google had multiple acceptable and effective non-infringing alternatives” to the patents-in-suit
and the copyrights, and purport to reject the opinions to the contrary of Prof. Mitchell, Oracle’s
computer science expert. (Leonard, 10-30; Cox, 23-27.)
To reach these technical opinions, the Reports rely almost entirely on interviews with
Google employees apparently conducted in the week prior to the submission of the Reports. Dr.
Leonard cites such interviews forty two times in his opening section alone. Among these
interviews, both Drs. Leonard and Cox rely heavily on Google employee Tim Bray, a witness
The Honorable William Alsup
October 10, 2011
Page 2 of 3
never mentioned in Google’s Rule 26 disclosures, not disclosed at all until appearing on
Google’s witness list on October 1, 2011, and for whom Google produced no documents. Drs.
Leonard and Cox rely on Mr. Bray for core propositions, such as purported “facts” about
disadvantages of the Java language that supposedly would have led Google to use alternatives
rather than pay for a license. (Leonard, 75-76; Cox, 22-23).
Beyond the non-infringing alternatives sections, the Reports are riddled with interview
citations, including to John Rizzo, a third-party witness not timely disclosed. Drs. Leonard and
Cox rely on such “interviews” while ignoring all contrary record evidence and disregard the
mandate that an expert “sufficiently tie the expert testimony on damages to the facts of the case.”
Uniloc USA, Inc. v. Microsoft Corp., 632 F.3d 1292, 1315 (Fed. Cir. 2011) (citing Daubert). In
all, Dr. Cox cites interviews 45 times; Dr. Leonard does so 61 times. In contrast, Dr. Cox cites
only 15 contemporaneous Google documents in all, and Dr. Leonard cites only 14.
The Reports also rely on Google documents that Google never produced, despite Oracle’s
repeated requests. These include 33 agreements and material from Google “dashboards.”
(Appx. B of the Reports (“Produced Documents Without Bates Numbers”).) The unfairness
goes farther: Google provided its experts filtered data for critical quantitative claims and has
refused to produce the underlying source data because its experts never saw source data.
Incompetent expert testimony based on purported facts provided to experts through late or even
undisclosed witnesses, interviews, and documents violates the Rules of Evidence and of Civil
Procedure and this Court’s Case Management Order. See Therasense, Inc. v. Becton, Dickinson
and Co., 2008 WL 2323856, *1-3 (N.D. Cal. May 22, 2008).
The Reports also violate numerous other basic safeguards. For example, Dr. Leonard
claims the conjoint surveys used in Dr. Cockburn’s report are subject to “hypothetical bias” even
The Honorable William Alsup
October 10, 2011
Page 3 of 3
though the academic literature—even the study cited in Dr. Leonard’s own Report—says the
opposite. See, e.g., Cabrera v. Cordis Corp., 134 F.3d 1418 (9th Cir. 1998) (expert testimony
unreliable because it had no support in medical literature). Similarly, as part of his claim that
Oracle’s copyright damages should be zero, Dr. Cox says, contrary to settled law, that even
though Google copied code, the code must be valueless because Google removed it. Frank
Music Corp. v. MGM, Inc., 772 F.2d 505, 518 (9th Cir. 1985). Dr. Cox also suggests that there
should be no disgorgement of Google’s actual infringer’s profits if Google could have earned
those profits without infringing. That distorts the concept of disgorgement, as does his
calculation of incremental profits compared to Google’s supposed next-best alternatives. Hebert
v. Lisle Corp., 99 F.3d 1109, 1117 (Fed. Cir. 1996) (“Incorrect statements of law are no more
admissible through ‘experts’ than are falsifiable scientific theories.”).
In addition to their reliance on interviews and unproduced evidence, both Dr. Leonard
and Dr. Cox repeatedly make quasi-factual assertions based on nothing other than their own sayso.
Such ipse dixit cannot form the basis for a reliable expert opinion. General Elec. Co. v.
Joiner, 522 U.S. 136, 146 (1997) (“But nothing in either Daubert or the Federal Rules of
Evidence requires a district court to admit opinion evidence that is connected to existing data
only by the ipse dixit of the expert.”).
Oracle respectfully seeks leave to file a motion or motions to strike and exclude Drs.
Cox’s and Leonard’s unqualified technical opinions, “facts” from spoon-fed last-minute
interviews, documents and witnesses that Google refused to produce or disclose, “facts” based
on ipse dixit instead of evidence, and opinions inconsistent with established legal principles.
/s/ Steven C. Holtzman
IN THE UNITED STATES DISTRICT COURT
ORACLE AMERICA, INC.,
FOR THE NORTHERN DISTRICT OF CALIFORNIA
No. C 10-03561 WHA
GRANTING AND PARTIALLY
DENYING SECOND MOTION
TO STRIKE PORTIONS OF
MITCHELL REPORT AND
In this patent and copyright infringement action involving features of Java and Android,
defendant moves to strike portions of opening and reply expert reports regarding patent
infringement. For the reasons stated below, the motion is GRANTED IN PART AND DENIED
The factual background of this action has been set forth in previous orders (see Dkt.
Nos. 137, 230, 433). In August 2011, defendant Google Inc. sought leave to file a motion to
strike portions of the opening expert report of John C. Mitchell regarding patent infringement.
Google asserted that many aspects of the report were not supported by plaintiff Oracle
America, Inc.’s infringement contentions as required by the Patent Local Rules (Dkt. No. 377).
Over Oracle’s objections, Google was allowed to file a motion limited to three points of critique
(Dkt. Nos. 380, 382). The motion was granted in part and denied in part. Because Google raised
meritorious critiques, Google was allowed to file the instant additional motion to strike
(Dkt. No. 464). This order follows expedited briefing.
Our Patent Local Rule 3-1 requires detailed disclosure of a party’s patent infringement
contentions. See Patent L.R. 3-1. Oracle initially disclosed its infringement contentions in
December 2010, and Google challenged their sufficiency two months later (Dkt. No. 79). Oracle
agreed to supplement its initial infringement contentions and in fact did so twice. Oracle
ultimately chose to stand on its April 2011 second supplemental disclosures, despite Google’s
position that they still were inadequate. Oracle was warned that if its disclosures later proved
inadequate to support its infringement theories, there would be no second chance to cure the
defects (Dkt. No. 110 at 10–11).
Google now asserts that the opening and reply expert reports of John C. Mitchell contain
infringement theories that were not disclosed in Oracle’s April 2011 infringement contentions.
Google also asserts that the reply report contains new material that should have been disclosed in
the opening report. Oracle objects to the attacks on the reply report, because “Google did not
obtain leave to move to strike any part of the Reply Report, as this is beyond the scope of its
August 30, 2011 précis and the Court’s September 26, 2011 Order, which concerned only the
[opening] Report” (Opp. 4–5). This objection is overruled. The Mitchell reply report had not yet
been served at the time of Google’s original précis request. Google’s critiques of the reply report
are in the same vein as its critiques of the opening report, and all of Google’s critiques have been
fully briefed. These issues will be resolved now, without further procedural ado.
1. FUNCTIONS VFORK() AND CLONE() (’720 PATENT).
Patent Local Rule 3-1(c) required Oracle’s disclosure of infringement contentions to
include a “chart identifying specifically where each limitation of each asserted claim is found
within each Accused Instrumentality.” Google asserts that the functions vfork() and clone() were
not identified in Oracle’s claim charts, but that the Mitchell opening report now relies on them as
satisfying certain limitations of claims 1 and 6 of United States Patent Number 7,426,720.
Google seeks to strike a portion of paragraph 606 and all of paragraphs 622–27 from the Mitchell
report “because they improperly refer to and rely upon the vfork() and clone() mechanisms”
Claim 1 of the ’720 patent covers a “system for dynamic preloading of classes through
memory space cloning of a master runtime system process,” comprising multiple components
including “a copy-on-write process cloning mechanism” whose function is described in detail in
the claim. Claim 6 covers a “system according to claim 1, further comprising . . . a process
cloning mechanism” whose function also is described in detail (col. 10:2–40). The
“copy-on-write process cloning mechanism” of claim 1 and the “process cloning mechanism” of
claim 6 are similar, but not identical.
Paragraph 606 of the Mitchell report states: “The Linux fork executed by Android
provides the ‘copy-on-write process cloning mechanism’ in its fork() system call. Linux provides
additional ‘process cloning mechanisms’ in its vfork() and clone() system calls.” The paragraph
then quotes at length from a source explaining these three system calls, or “functions” (Banner
Exh. A at 301–06).
Google’s instant motion does not challenge Mitchell’s theory that the fork() function
satisfies the “copy-on-write process cloning mechanism” limitation of claims 1 and 6. Indeed, the
Mitchell report also identifies several other items as satisfying the “copy-on-write process cloning
mechanism,” and Google does not contend that any of those theories was not timely disclosed.
On the other hand, Mitchell’s theory that the vfork() and clone() functions satisfy the
“process cloning mechanism” of claim 6 is where the battle is joined. Paragraphs 622–27 of the
Mitchell report focus largely on the clone() function as an example of a “process cloning
mechanism” for claim 6 purposes (Banner Exh. A at 131–15). Oracle’s operative infringement
contentions did not identify the vfork() or clone() functions as satisfying the “process cloning
mechanism” limitation of claim 6. This is not disputed. Oracle therefore may not now rely on
those functions as satisfying that claim limitation.
Oracle states that in both its infringement contentions and in the Mitchell report, its “’720
patent infringement theory focuses on a function present in Android called do_fork().” According
to Oracle, the report references vfork() and clone() only to explain where do_fork() is found and
how do_fork is called (Opp. 1–3). True, the Mitchell report identifies the function do_fork() as
“provid[ing] the ‘process cloning mechanism’ of claim 6” in one of the very paragraphs Google
seeks to strike (Banner Exh. A at 314). Google does not challenge this theory in its instant
motion. The report, however, also seems to identify the functions vfork() and clone() as satisfying
that limitation, as described above. Such theories would be improper.
As to the functions vfork() and clone(), Google’s motion is GRANTED IN PART AND
DENIED IN PART as follows. Google has not shown that the Mitchell report identifies vfork() or
clone() as satisfying the “copy-on-write process cloning mechanism” limitation of claim 1 and
dependent claim 6. This order therefore does not address whether such theories were adequately
disclosed in Oracle’s infringement contentions. On the other hand, the Mitchell report does
identify vfork() and clone() as satisfying the “process cloning mechanism” limitation of claim 6.
Those theories were not disclosed in Oracle’s infringement contentions. Accordingly,
Mr. Mitchell may not offer direct testimony that the vfork() or clone() function itself satisfies the
“process cloning mechanism” limitation of claim 6. This ruling, however, does not preclude any
and all mention of those functions on direct. For example, they may be referenced to describe
where and how the do_fork function is called if indeed Oracle contends that do_fork itself
satisfies the “process cloning mechanism” limitation.
2. “MBS MOBILE” APPLICATION (’476 PATENT).
Again, the Patent Local Rules required disclosure of a chart identifying where each
limitation of each asserted claim was found within each accused item. Google asserts that no
application called “mBS Mobile” was disclosed in Oracle’s operative infringement contentions,
but that the Mitchell opening report now identifies such an application as providing the security
functionality claimed in United States Patent Number 6,192,476. Google seeks to strike
paragraphs 70 and 77 from the Mitchell report.
Paragraphs 70 and 77 are identical. They refer to the mBS Mobile application to illustrate
“[s]ome of the ways that application developers could benefit from the java.security framework”
(Banner Exh. A at 21–24). It is undisputed that the mBS Mobile application was not mentioned
in Oracle’s infringement contentions. The Mitchell report, however, does not accuse the mBS
Mobile application of satisfying any particular limitation of the asserted claim of the ’476 patent.
Rather, the java.security framework is accused, and the mBS Mobile application is cited as an
example of the benefit of the alleged infringement. Paragraphs 70 and 77 do not present a new
theory of infringement based on the mBS Mobile application. As explained with reference to
Google’s first motion to strike portions of the Mitchell report, the Patent Local Rules required
Oracle to disclose its theories of infringement, not to identify every evidentiary item of proof
supporting those theories.
In its reply brief, Google characterizes the discussion in paragraphs 70 and 77 as a new
theory of indirect infringement. Google, however, does not explain how Oracle’s disclosures
concerning indirect infringement are supposedly inadequate to support the challenged paragraphs.
As to the mBS Mobile application, Google’s motion is DENIED.
3. SECURITYMANAGER AND ACCESSCONTROLLER (’476 PATENT).
Google asserts that paragraph 115 of the Mitchell reply report “should be stricken because
it represents a new infringement theory that was never previously disclosed by Oracle” (Br. 4).
The entirety of challenged paragraph 115 states (Banner Exh. B at 47) (internal quotations and
Even if the SecurityManager is disabled, the Java security
framework may still be used. For example, the static methods in
AccessController are always available to be called. Even if no
SecurityManager is instantiated, Android code includes
AccessController.java and the static methods associated with the
class may be called. Therefore, the functionality of the Java
security framework is accessible via the AccessController.
Google argues that this theory is improper because it was not disclosed in Oracle’s infringement
contentions and was not presented in the opening Mitchell report. According to Google, both the
infringement contentions and the opening report relied on SecurityManager rather than
AccessController to show infringement of the ’476 patent (Br. 3–4).
Challenged paragraph 115 is located in a portion of the reply report that addresses
claim 10 as an example, stating that the other then-asserted claims “are similar” (Banner Exh. B
at 46). At present, only claim 14 of the ’476 patent remains at issue (Dkt. No. 471). The instant
record includes portions of the infringement contentions and opening report that address claim 10,
but not claim 14. These excerpts specifically address the “determining whether said action is
authorized” limitation of claim 10. Claim 14 includes a similar, but not identical, limitation.
Because both sides rely on the claim 10 portions of the infringement disclosures and opening
report, this order assumes for the limited purposes of the instant motion that those claim 10
arguments apply equally to claim 14.
Oracle’s infringement contentions disclose AccessController along with SecurityManager
with respect to the relevant limitation (Banner Exh. D at 22). Thus, the statements about
AccessController in paragraph 115 do not run afoul of the disclosure rules.
Whether the content of paragraph 115 should have been in the opening report is a separate
question. The parties were warned that reply reports must be limited to true rebuttal, should be
very brief, and should not add new material that should have been placed in the opening report
(Dkt. No. 56 at ¶ 8). The Mitchell opening report opined that SecurityManager.java was the
accused element that satisfied the “determining whether said action is authorized” claim
limitation (Banner Exh. A at 362). The report further explained that it did so “by calling the
AccessController.checkPermission method, which in turn calls the
AccessControlContext.checkPermission method to check the permissions” (id. at 363). The
opening report did not identify any other accused element as satisfying the limitation in question,
and it did not identify any other steps by which the AccessController could be called.
The statement in paragraph 115 of the reply report that the AccessController methods may
be called by something other than the SecurityManager presents a new infringement theory. This
statement is not true rebuttal; it is sandbagging. Oracle had the burden to set forth all of its
infringement theories in the opening report. If Oracle intended to accuse something other than the
SecurityManager of infringing the claims of the ’476 patent by calling the AccessController
methods, then the Mitchell opening report should have said so.
Google argues that this critique is analogous to the critique regarding DvmDex.h in
Google’s first motion to strike, and that this critique should be rejected for the same reason. Not
so. There, the infringement contentions identified specific accused functions, and the Mitchell
report later identified the code files in which those functions were stored. Here, the opening
report identified a series of accused steps, and the reply report then sought to replace one step in
the series with a different (and unspecified) accused element. Doing so would present a new
As to the theory of infringement by something other than the SecurityManager calling the
AccessController methods, Google’s motion is GRANTED. Paragraph 115 of the Mitchell reply
report is STRICKEN.
4. MULTIPLE INSTRUCTIONS AND UNSUCCESSFUL SIMULATED EXECUTION
Google asserts that the Mitchell reply report presents two new theories of infringement of
United States Patent Number 6,061,520. According to Google, Oracle’s infringement contentions
and the Mitchell opening report both advanced the theory that Android’s dx tool infringes the
asserted claims of the ’520 patent “where it successfully creates a single new instruction, referred
to as a fill-array-data instruction, that replaces a less efficient series of Java bytecode instructions
for initializing individual elements of a static array” (Br. 4). On the other hand, Google interprets
the Mitchell reply report as claiming that infringement includes “instances where more than one
instruction is created” and instances where simulated execution is not successful (id. at 5).
Google moves to strike the “one or more” phrase and the last three sentences of paragraph 92 of
the Mitchell reply report.
A. Creation of Multiple Instructions.
The relevant portions of Oracle’s infringement contentions and the Mitchell opening
report do refer to creating a single instruction, not one or more instructions (e.g., Banner Exh. E
at 29, Exh. A at 277). The portion of paragraph 92 that refers to creating “one or more
instructions” purports to further explain Oracle’s infringement theory; it does not directly
respond to any argument by the opposing expert.
Oracle does not attempt to explain or justify the reference to creating “one ore more”
instructions in the Mitchell reply report. Oracle cites only to Google’s interrogatory responses,
which are irrelevant to the question at hand (Opp. 7). The statement in paragraph 92 that the
accused elements create “one or more” instructions amounts to a new infringement theory. As
such, it is improper.
B. Unsuccessful Simulated Execution.
The last three sentences of paragraph 92 address “Prof. Parr’s example in his Appendix A,
Section II” (Banner Exh. B at 39). This example involved a test that was run on modified, i.e.,
not standard, Android code (Peters Exh. C). As such, it is a hypothetical example. The
statements addressing this hypothetical example are true rebuttal and need not be supported by
any previous disclosures. In particular, the statement that the claims of the ’520 patent “do not
require the simulating step to succeed” expresses a broad view on the scope of the claims as
applied to the hypothetical example. It does not amount to an entirely new infringement theory,
because it does not accuse any particular new item of infringement. Even if Oracle chose in its
infringement contentions to accuse only devices and methods in which simulated execution is
successful, Oracle’s expert still may rebut a hypothetical example by opining that circumstances
in which simulated execution does not succeed may also infringe. The last three sentences of
paragraph 92 are proper.
As to the reply report discussion of the ’520 patent, Google’s motion is GRANTED IN
PART AND DENIED IN PART. The reference to creating more than one instruction is STRICKEN
from the third sentence of paragraph 92 of the Mitchell reply report. The last three sentences of
that paragraph remain intact.
For the reasons stated above, defendant’s motion to strike portions of the opening and
reply expert reports of John C. Mitchell regarding patent infringement is GRANTED IN PART AND
DENIED IN PART. As to the functions vfork() and clone(), the motion is GRANTED IN PART AND
DENIED IN PART. Mr. Mitchell may not offer direct testimony that the vfork() or clone() function
itself satisfies the “process cloning mechanism” limitation of claim 6 of the ’720 patent. As to the
mBS Mobile application, the motion is DENIED. As to the theory that the ’476 patent can be
infringed by something other than the SecurityManager calling the AccessController methods, the
motion is GRANTED. Paragraph 115 of the Mitchell reply report is STRICKEN. As to the reply
report discussion of the ’520 patent, Google’s motion is GRANTED IN PART AND DENIED IN PART.
The reference to creating more than one instruction is STRICKEN from the third sentence of
paragraph 92 of the Mitchell reply report. The last three sentences of that paragraph
The hearing on the instant motion previously set for October 13, 2011, is VACATED. The
hearing on the motion for relief from a non-dispositive pretrial order of a magistrate judge, set for
the same day, will go forward as planned.
IT IS SO ORDERED.
Dated: October 11, 2011.
/s/ William Alsup
UNITED STATES DISTRICT JUDGE