Oracle America, Inc. v. Google Inc.

Filing 1166

Brief re #1151 MOTION for Judgment as a Matter of Law on Counts V and VII of Oracle's Amended Complaint filed byGoogle Inc.. (Related document(s) #1151 ) (Baber, Bruce) (Filed on 5/16/2012)

1
2
3
4
5
6
7
8
9
10
11
12
13

KEKER & VAN NEST LLP
ROBERT A. VAN NEST - # 84065
[e-mail]
CHRISTA M. ANDERSON - # 184325
[e-mail]
DANIEL PURCELL - # 191424
[e-mail]
633 Battery Street
San Francisco, CA 94111-1809
Telephone:
415 391 5400
Facsimile:
415 397 7188

KING & SPALDING LLP
DONALD F. ZIMMER, JR. - #112279
[e-mail]
CHERYL A. SABNIS - #224323
[e-mail]
101 Second Street, Suite 2300
San Francisco, CA 94105
Tel: [phone]
Fax: [phone]

KING & SPALDING LLP
SCOTT T. WEINGAERTNER
(Pro Hac Vice)
[e-mail]
ROBERT F. PERRY
[e-mail]
BRUCE W. BABER (Pro Hac Vice)
1185 Avenue of the Americas
New York, NY 10036
Tel: [phone]
Fax: [phone]

IAN C. BALLON - #141819
[e-mail]
HEATHER MEEKER - #172148
[e-mail]
GREENBERG TRAURIG, LLP
1900 University Avenue
East Palo Alto, CA 94303
Tel: [phone]
Fax: [phone]

Attorneys for Defendant
GOOGLE INC.

14

UNITED STATES DISTRICT COURT

15

NORTHERN DISTRICT OF CALIFORNIA

16

SAN FRANCISCO DIVISION

17
ORACLE AMERICA, INC.,

Case No. 3:10-cv-03561 WHA

18
Plaintiff,
19
v.
20

BRIEF IN SUPPORT OF GOOGLE’S
MOTION FOR JUDGMENT AS A MATTER
OF LAW ON COUNTS V AND VII OF
ORACLE’S AMENDED COMPLAINT

GOOGLE INC.,
21
Defendant.

Dept.:
Judge:

Courtroom 8, 19th Floor
Hon. William Alsup

22
23
24
25
26
27
28
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-CV-03561 WHA

1
I.
2
3
4
5

INTRODUCTION
In accordance with Rule 50 of the Federal Rules of Civil Procedure, as previously set

forth in open court on May 10, 2012 (RT 3707-3709), and as outlined on May 15, 2012 (Dkt.
1151) Google hereby files this brief in support of Google’s motion for judgment as a matter of
law on Counts V and VII of Oracle’s Amended Complaint.

6
7

II.

Judgment as a matter of law is proper when “a party has been fully heard on an issue and

8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

LEGAL STANDARDS

there is no legally sufficient evidentiary basis for a reasonable jury to find for that party on that
issue.” Fed. R. Civ. P. 50(a)(1). Rule 50 “allows the trial court to remove . . . issues from the
jury’s consideration when the facts are sufficiently clear that the law requires a particular result.”
Weisgram v. Marley Co., 528 U.S. 440, 448 (2000) (internal quotations omitted). The standard
for granting judgment as a matter of law, in practice, mirrors the standard for granting summary
judgment, and “the inquiry under each is the same.” Anderson v. Liberty Lobby, Inc., 477 U.S.
242, 250-51 (1986).
III.

AS TO COUNT V OF ORACLE’S AMENDED COMPLAINT: GOOGLE IS
ENTITLED TO JUDGMENT AS A MATTER OF LAW
To prove patent infringement, Oracle must show that Android devices meet “each and

every” element of the asserted claims. Star Scientific, Inc. v. R.J. Reynolds Tobacco Co., 655
F.3d 1364, 1378 (Fed. Cir. 2011). Oracle presented two theories of infringement based on (1) the
Resolve.c code in the Dalvik virtual machine and (2) the dexopt tool. Oracle has not carried its
burden to show that either of these Android features meets each and every element of any of the
asserted claims. Google is therefore entitled to judgment as a matter of law that it has not
infringed the ‘104 patent.

24
A.
25
26
27

Because Dalvik bytecode instructions do not contain symbolic references,
Google is entitled to judgment as a matter of law that it does not infringe any
of the asserted claims of the ’104 patent.

There is no evidence that either Resolve.c or the dexopt tool operate on “instructions
containing one or more symbolic references,” as required by asserted Claims 11, 27, 29, 39, 40,

28
1
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

and 41 of the ‘104 patent. The Court has construed the term “symbolic reference” to mean

2
3

a reference that identifies data by a name other than the numeric memory location
of the data, and that is resolved dynamically rather than statically

4

[Dkt. No. 137 at 22.] As Dr. Mitchell conceded, this construction makes a distinction between

5

using names to represent data (i.e., symbolic references) and using numeric memory locations.

6

[RT 3480:12-15 (Mitchell).] There is no limitation regarding the type of data being referenced.

7

It is undisputed, that each of the asserted claims of the ’104 patent requires that there be a

8

symbolic reference in the instructions themselves. Even Dr. Mitchell, Oracle’s expert admits this.

9

[RT 3477:6-12 (“Q. Because every claim that you’ve analyzed requires that the symbolic

10

reference be in the instructions, right? A. The wording varies across the claims. Q. But,

11

essentially, either contained in or in each one of the limitations requires that there be a symbolic

12

reference in the instructions themselves, right? A. Yes.”); RT 3487:13-16 (“But you would agree

13

with me that whether you’re talking about Resolve.c or dexopt, you’ve got to find a symbolic

14

reference in the instructions? A. Yes.”).] The requirement is also clear from the plain language

15

of the claims. See TX 4015 at col. 13 (claim 11 reciting “instructions containing one or more

16

symbolic references”), at col. 14 (claims 27 and 29 reciting “symbolic data references in the set of

17

original instructions”), at col. 15 (claims 39-41 reciting “instruction contains a symbolic field

18

reference.”). Thus, to prove infringement, Oracle bore the burden of showing that Resolve.c and

19

the dexopt tool operate on instructions that contain “a reference that identifies data by a name

20

other than the numeric memory location of the data.”

21

Because Dalvik bytecode instructions—the instructions evaluated by both Resolve.c and

22

the dexopt tool—contain references to numeric memory locations of data, they cannot satisfy all

23

of the limitations of the asserted claims. Dalvik bytecode instructions consist of an opcode and

24

operand. [TX 737 (“Each instruction starts with the named opcode and is optionally followed by

25

one or more arguments.”); RT 3590:23-3591:25 (Bornstein); RT 3844:14-3855:12, 3925:4-18

26

(August).] In Android, the accused operands are indexes that take the form “field@CCCC” and

27

reference locations in tables. [See TX 46.106; TX 735 (“There are separately enumerated and

28

indexed constant pools for references to strings, types, fields, and methods.”); TX 737; RT
2
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

3732:15-19, 3736:16-23, 3755:8-9, 3765:9-12 (McFadden); 3858:5-12, 3858:21-359:11, 3918:13-

2

23, 3923:20-24, 3925:19-3926-9 (August).] Dr. Mitchell agrees. [RT 3488:6-8, 3488:19-23,

3

3489:10-12 (Dr. Mitchell) (the “classIdx” index in Dalvik instructions provides information “to a

4

location in another table”); RT 3496:12-3497:6 (Dr. Mitchell) (the “field Idx” in Dalvik

5

instructions is “an index to a specific location in the field table”).] Applying the Court’s claim

6

construction, these indexes do not qualify as symbolic references.

7

As Dr. Mitchell confirmed via testimony, his report refers to indexes as numeric rather

8

than symbolic references.

9

4039:11, 4039:14-4040:9 (Mitchell).] This is not surprising in light of the fact that indexes point

10

to locations in tables. To try to circumvent his prior statements, Dr. Mitchell has conjured up a

11

re-interpretation of the term “symbolic reference” that does not conform with the Court’s

12

construction. Specifically, Dr. Mitchell bases his opinion on a meta claim construction argument

13

about what “data” means in the context of the Court’s construction:

14

[RT 3490:7-10, 3528:9-22, 3529:1-10, 4037:9-4038:3, 4038:10-

16

As far as I can tell, the only contested issue here is whether the field indexes and
instructions are symbolic references or not. And that in the end just seems to turn
on what we mean by data, according to the last bit of testimony from Professor
August.

17

[RT 4019:3-7 (Mitchell) (emphasis added).] Dr. Mitchell explains that his opinion is based on the

18

theory that “data” in the Court’s construction is something much more restrictive than what the

19

ordinary meaning of the word “data” implies. According to Dr. Mitchell, the “data” must be the

20

actual field data in the instance object, as opposed to any data in the .dex file, including the data

21

in the constant pool tables. [RT 4025:6-21 (Mitchell) (“So in order for that to happen, we have to

22

find where that actual data is in the object using this symbolic reference. . . .”).] In other words,

23

to turn an index, which is a numeric reference to a location in a table, into a symbolic reference,

24

Dr. Mitchell has re-defined “symbolic reference” to mean “a reference that identifies the actual

25

field data in the instance object by a name other than the numeric memory location of the actual

26

field data in the instance object” (emphasis indicating Dr. Mitchell’s apparent changes to Court’s

27

construction).

15

28

Oracle’s overly-restrictive, re-definition of the term “data” was again highlighted in
3
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1
2
3
4

closing arguments when Mr. Jacobs argued:
Now, there was another potential source of confusion which we clarified also
through Dr. August and through Professor Mitchell. What’s the data that this
claim is talking about? Is it talking about what Google wanted to focus on, which
is the constant pool information (indicating), or is it talking about the data in the
data object, the actual field value, the actual data?

5
...
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

So the whole goal of this exercise is to obtain meaningful data, the actual data. . . .
...
Now, Mr. Van Nest agreed that the distinction between this is at least the
distinction between meaningful data and unmeaningful data. We’re relying on
meaningful data. Google is relying on unmeaningful data.
[RT 4111:5-11, 4200:7-10 (Jacobs).] This simply does not conform to the Court’s ruling on
claim construction, which simply distinguishes between a reference to data—any data—by a
name rather than location. See Claim Construction Order (Dkt. No. 137) at 20-22.
Oracle never requested clarification of the meaning of the term “data” during the
Markman process. It also declined the Court’s offer to seek additional claim constructions in
December 2011, months after Dr. Mitchell changed his opinion in response to Dr. August’s
report. [Dkt. Nos. 603, 637.] Most relevant to this motion, neither Dr. Mitchell nor any other
evidence offered by Oracle suggested that “data” is a term of art that is understood by those
skilled in the art to correspond to or encompass only “actual field data in the instance object.” In
fact, there is no such suggestion anywhere in the record. Accordingly, no reasonable jury could
conclude, based on the evidence and expert testimony, that there is a basis for such a meaning.
As is clear on its face, the Court’s construction of “symbolic reference” does not
incorporate the restrictive meaning of “data” advanced by Oracle and Dr. Mitchell. See Claim
Construction Order (Dkt. No. 137) at 20-22. As such, Dr. Mitchell’s testimony cannot be
credited. [RT 4206:18-21 (Court) (“If a witness has based his view on meanings of the terms
contrary to my stated definitions, you should discount that part of his testimony accordingly.”).]
Without Dr. Mitchell’s testimony, Oracle cannot prove that Dalvik bytecode instructions

28
4
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

contain symbolic references, which each of the asserted claims requires. To the contrary, the

2

wealth of evidence presented at trial confirms that Dalvik bytecode instructions contain only

3

numeric references in the form of indexes to locations in memory where constant pool data is

4

stored in the dex files. [See TX 46.106; TX 735 (“There are separately enumerated and indexed

5

constant pools for references to strings, types, fields, and methods.”); TX 737; RT 3732:15-19,

6

3736:16-23, 3755:8-9, 3765:9-12 (McFadden); 3858:5-12, 3858:21-359:11, 3918:13-23, 3923:20-

7

24, 3925:19-3926-9 (August).] Thus, no reasonable jury could conclude that either the Resolve.c

8

code or the dexopt tool in Android infringe the asserted claims of the ‘104 patent.

9

B.

10

Google is also entitled to judgment as a matter of law that the dexopt
functionality does not infringe Claims 27 and 29 of the ’104 patent because
dexopt resolves symbolic references statically rather than dynamically.

11

Android’s dexopt functionality performs static—not dynamic—optimizations. 1

12

Court’s claim construction requires that symbolic references be “resolved dynamically rather than

13

statically.” In light of that requirement, there is no dispute that if dexopt performs functions

14

statically, then dexopt doesn’t infringe. [RT 3509:16-18 (Mitchell).] Dr. Mitchell’s infringement

15

opinion at trial was that dexopt is a dynamic process because it relies on information about the

16

runtime environment and that is available at runtime. [RT 4028:14-25 (Mitchell).] But Dr.

17

Mitchell’s conclusory statements do not provide an adequate basis for a reasonable jury to reach a

18

verdict of literal patent infringement. Indeed, there is no evidence to support Dr. Mitchell’s

19

opinion that an optimization (e.g., resolution) based on information about a runtime environment

20

is “dynamic” and not “static.” To the contrary, there is a wealth of documentary evidence, fact

21

testimony, and expert testimony that directly contradicts Dr. Mitchell’s opinion. [RT 3730:16-22

22

(McFadden); RT 3940:17-20 (August); RT 3595:21-24 (Bornstein); TX 32 at 35, TX 816 at

23

24:05; TX 735 (defining opcodes that “are reasonable candidates for static linking”); TX 737

24

(defining “statically linked” instruction formats); TX 739 (dexopt performs optimizations that

25

“can be inferred statically”).] Accordingly, no reasonable jury could conclude that the dexopt

26

tool infringes claims 27 and 29 of the ‘104 patent.

27
28

1

Only claims 27 and 29 are asserted against dexopt.
5
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

The

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

C.

There is no legal basis for the jury to conclude that Android infringes the ’104
patent where Oracle only presented a literal infringement theory.

To the extent Oracle now contends that certain claim elements are effectively present in
Android under the doctrine of equivalents, Google is entitled to a judgment as a matter of law.
First, having failed to disclose such a theory in its Rule 3-1 Patent Disclosures under the Patent
Local Rules, Oracle may not now assert an infringement theory under the doctrine of equivalents.
Second, even if Oracle were not barred from asserting such a theory, Dr. Mitchell’s testimony
would not support a conclusion of infringement under the doctrine of equivalents. The record
contains no evidence that the accused Dalvik bytecode instructions, which contain symbolic
references not in the instructions themselves but in a shared constant pool reached (directly or
indirectly) via various tables, are substantially similar to instructions that literally contain
symbolic references. See TIP Sys., LLC v. Phillips & Brooks/Gladwin, Inc., 529 F.3d 1364,
1376-77 (Fed. Cir. 2008) (infringement under doctrine of equivalents may only be found where
the accused device contains an “insubstantial” change from the claimed invention or where the
accused device functions in substantially the same way as the claimed invention) (citing Graver
Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 610 (1950)). Here, Dr. Mitchell has
testified that strings are stored in the string data table of the dex file. [RT 3290:2-3291:18
(Mitchell).] These strings, which are the names of the fields, classes, methods, and strings used in
a program, are not stored in the Dalvik bytecode instructions themselves. The Resolve.c code and
the dexopt tool use a series of numeric references (indexes) to eventually obtain a symbolic
reference (name) from the string data table, and then resolve that name to find a numeric
reference corresponding to that symbolic reference. In contrast, the asserted claims of the ‘104
patent require resolving a symbolic reference contained in the instruction; they do not have to
follow a series of numeric references to obtain the symbolic reference (name). The way in which
symbolic resolution is performed in Resolve.c and dexopt is advantageous compared to the
process disclosed in the ‘104 patent in that it allows the Dalvik VM to operate with fixed-width
instructions, which make processing virtual machine instructions easier. [RT 3592:4-3593:4
(Bornstein), 3263:7-3264:8 (McFadden), 3878:3-3879:9 (August).]

28
6
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1
2

AS TO COUNT VII OF ORACLE’S AMENDED COMPLAINT: GOOGLE IS
ENTITLED TO JUDGMENT OF NON-INFRINGEMENT OF THE ‘520 PATENT
AS A MATTER OF LAW

3

To prove patent infringement, Oracle must show that the accused functionality meets

4

“each and every” element of an asserted claim. Star Scientific, 655 F.3d at 1378. Oracle has not

5

carried that burden for either of the asserted claims of the ‘520 patent. In particular, Oracle has

6

not shown that the code it accuses of infringement ever simulates execution to identify the static

7

initialization of arrays because, as the undisputed evidence at trial showed, that is not how the dx

8

tool works. The requirement that identification of the initialization of static arrays be performed

9

through “simulating execution” is found in both asserted claims, Claim 1 and Claim 20. TX

10
11

III.

4011.
The dx tool does not simulate execution in order to identify the static initialization of an

12

array. Instead of going through the steps required to produce output showing what the static array

13

would look like if the Java byte code instructions were performed, the dx tool uses a pattern-

14

matching algorithm to determine what the static array would look like when initialized. The

15

pattern matcher (a method known as parseNewarray found in class bytecodeArray) looks for a

16

repeating sequence of instructions used to initialize static arrays. [RT 3800:2-3801:3 (Parr).]

17

Although the pattern matcher code is called by code in the class file Simulator.java, the pattern

18

matcher itself does not simulate execution of the bytecode, e.g., by manipulating a stack. [RT

19

3802:25-3804:1 (Parr).] Instead, it returns the operands associated with the appropriate

20

instructions in the pattern. [RT 3343:3-16 (Mitchell).] Moreover, the pattern matcher code will

21

not work properly when a useless instruction is injected into the Java bytecode. As would be

22

expected, the dx tool, when confronted with revised code that breaks the pattern, returns the

23

Dalvik byte code equivalents of the Java byte code that performs the steps required to statically

24

initialize the array rather than an efficient instruction that initializes a static array. [RT 3808:18-

25

3809:6; 3809:20-3810-3 (Parr).] A process using simulated execution would simulate the

26

execution of each instruction, including the useless instruction, and the useless instruction would

27

end up having no effect on the eventual result. [RT 3810:4-13 (Parr).]

28

To prevail on literal infringement of the ‘520 patent, Oracle needed to prove that the
7
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

pattern matching code used in Android literally simulates execution, as required by the asserted

2

claims. Dr. Mitchell’s infringement opinion at trial was that the “simulation” (not “simulating

3

execution”) carried out by the parseNewarray method fell within the claims. [RT 4032:7-22

4

(Mitchell).] But a conclusory statement equating pattern matching to simulation (and not even

5

“simulating execution”) does not provide an adequate basis for a reasonable jury to reach a

6

verdict of literal patent infringement.

7

The dx tool source code confirms the absence of any literal correspondence between the

8

“simulating execution” claim limitation and the pattern matching code. Dr. Mitchell testified that

9

the identification of the static initialization of the array was performed by the parseNewarray

10

code, and agreed that code was distinct from the Simulator.java class. [RT 3343:3-16; 4061:21-

11

4062:3 (Mitchell).] Both Mr. Bornstein and Dr. Parr testified that the dx tool includes code that

12

actually performs simulated execution by manipulating a stack; however, the simulated execution

13

is not used for identifying the initialization of arrays. [RT 3589:11-17; 3589:25-3590:6; 3611:22-

14

3612:10 (Bornstein); RT 3798:2-8; 3798:22-3799:3 (Parr); TX 46.16, 46.17.] The presence of the

15

word “simulates” in code comments thus do not show that the dx tool simulates execution for

16

purposes of identifying the static initialization of an array; instead, simulated execution in the dx

17

tool is used only for reasons other than identifying the initialization of static arrays. [RT

18

3611:22-3612:10 (Bornstein); 3798:3-8 (Parr).] And it is undisputed that the ‘520 patent does not

19

include any mention of simulating execution through the use of pattern matching. [TX 4011; RT

20

3521:10-19 (Mitchell).]

21

Oracle never requested a construction of the phrase “simulating execution” during the

22

Markman process, taking the position that it should be construed in accordance with its “plain

23

meaning.” [See Dkt. No. 91 (parties’ joint claim construction and prehearing statement).] Oracle

24

did not suggest that simulating execution is a term of art that is understood by those skilled in the

25

art to correspond to or encompass pattern matching. No reasonable jury could conclude, based on

26

the evidence and expert testimony, that there is a basis for such a construction of simulating

27

execution, nor that there is an understanding in the field of computer science as to either of these

28

terms that supports an assertion that pattern matching is literally simulating execution. Such a
8
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

construction was never sought by Oracle, even after the Court permitted the parties to seek

2

additional claim constructions earlier this year. [Dkt. Nos. 603, 637.] Because of Oracle’s failure

3

of proof and its inability to construe its claims to cover pattern matching, there is no basis for a

4

finding of literal infringement of the ‘520 patent by the Android dx tool. The fact that the

5

Android code actually does perform simulated execution for some functions is irrelevant—the

6

key issue, as proven by the record, is that the dx tool does not simulate execution for the purpose

7

of identifying the initialization of static arrays.

8
9

Oracle’s position, as articulated by Dr. Mitchell, depends on an implied theory of
equivalent infringement, that “pattern matching” achieves a result similar to that of simulated

10

execution—i.e., identifying the initialization of a static array without actual execution of the byte

11

code. [RT 4032:11-22 (Mitchell)] But even assuming, without conceding, that the two different

12

processes provide similar results, that is not a basis for a finding of infringement. To begin with,

13

Oracle may not assert an infringement theory under the doctrine of equivalents, having failed to

14

disclose such a theory in its Rule 3-1 Patent Disclosures under the Patent Local Rules. But even if

15

Oracle were not barred from asserting a belated theory of infringement under the doctrine of

16

equivalents, Dr. Mitchell’s testimony would not support a conclusion of infringement under the

17

doctrine of equivalents because the recited claim limitation of “simulating execution” does not

18

work in a way substantially similar to the accused pattern matching mechanism. See TIP Sys.,

19

LLC v. Phillips & Brooks/Gladwin, Inc., 529 F.3d 1364, 1376-77 (Fed. Cir. 2008) (infringement

20

under doctrine of equivalents may only be found where the accused device contains an

21

“insubstantial” change from the claimed invention or where the accused device functions in

22

substantially the same way as the claimed invention) (citing Graver Tank & Mfg. Co. v. Linde Air

23

Prods. Co., 339 U.S. 605, 610 (1950)). An argument based on a theory of infringement under the

24

doctrine of equivalents, not disclosed at the proper time, is also ripe for judgment as a matter of

25

law.

26

In order to support its literal infringement claims, Oracle needed to link up the accused

27

code with the claimed functionality and, if it could, with the other language it considered helpful.

28

Oracle could not do so. Unable to prove actual simulated execution, Oracle has tried to
9
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

reconstrue “simulating execution” to cover pattern matching with no legal basis to do so.

2

Therefore, no reasonable jury could find that the dx tool infringes the ‘520 Patent.

3

IV.

4

THE PATENT ACT DOES NOT DEFINE INFRINGEMENT TO INCLUDE
MAKING A PRODUCT AVAILABLE FOR FREE.
The evidence shows that Google does not charge a fee for Android source code. [RT

5
3079:6-13 (Poore).] As such, Google cannot, as a legal matter, infringe under the applicable
6
statutory language.
7
The Patent Act defines patent infringement to include only certain discrete acts: “Except
8
as otherwise provided in this title, whoever without authority makes, uses, offers to sell, or sells
9
any patented invention, within the United States, or imports into the United States any patented
10
invention during the term of the patent therefore, infringes the patent.” 35 U.S.C. § 271(a)
11
(emphases added). These terms take their well-known, ordinary meanings. See NTP, Inc. v.
12
Research in Motion, Ltd., 418 F.3d 1282, 1319 (Fed. Cir. 2005); see also Med. Sol’ns, Inc. v. C
13
Change Surgical LLC, 541 F.3d 1136, 1141 (Fed. Cir. 2008); Rotec Indus., Inc. v. Mitsubishi
14
Corp., 215 F.3d 1246, 1255 (Fed. Cir. 2000). None of the categories covers the free distribution
15
of source code via the Internet.
16
Labeling free distribution of a product as a “sale” would turn that term on its head. As the
17
Federal Circuit has held, “[t]he definition of ‘sale’ is: ‘1. The transfer of property or title for a
18
price.’” NTP, 418 F.3d at 1319 (quoting Black’s Law Dictionary 1337 (7th ed. 1999)) (emphasis
19
added). And one of the “four elements” of a sale is “‘a price in money paid or promised.’” NTP,
20
418 F.3d at 1319 (quoting Black’s Law Dictionary 1337 (7th Cir. 1999)) (emphasis added).
21
“Plainly, the common, or usual meaning of the term sale includes those situations in which a
22
contract has been made between two parties who agree to transfer title and possession of specific
23
property for a price.” Enercon GmbH v. ITC, 151 F.3d 1376, 1382 (Fed. Cir. 1998) (emphasis
24
added).
25
By definition, distribution of a product for free does not constitute a transfer for a price.
26
Thus, in a case involving patents for collecting and using newborn umbilical cord blood, the
27
Federal Circuit found no “sale” where “[t]he defendants simply transferred the cord blood units to
28
10
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

designated transplanters upon direction from the families. Such a transaction does not constitute

2

a ‘sale’ to a transplanter under any definition of the term ‘sale.’” PharmaStem Therapeutics, Inc.

3

v. Viacell, Inc., 491 F.3d 1342, 1359 (Fed. Cir. 2007). Similarly, district courts have concluded

4

that distributing free samples does not qualify as a sale within the meaning of section 271(a). See

5

Cabot Corp. v. WGM Safety Corp., 562 F. Supp. 891, 892 (D. Mass. 1983).

6

If free distribution is not a sale, then offering to distribute something for free cannot be an

7

offer for sale. See Rotec, 215 F.3d at 1255; HollyAnne Corp. v. TFT, Inc., 199 F.3d 1304, 1309-

8

10 (Fed. Cir. 1999). “If, as in this case, the offer included none of the hallmarks of a potential

9

commercial transaction (i.e., a quotation of a price and a product description, or a communication

10

that the item was available for purchase by the intended donee), it cannot be considered an offer

11

for sale.” HollyAnne, 199 F.3d at 1310. Applied to the facts of this case, Google does not “offer”

12

Android for sale. Rather, it places the Android source code on a website from which others can

13

download it free of charge.

14

Oracle has also argued that Google “uses” Android within the meaning of the statute when

15

it makes Android available for free download. That requires a strained reading of the term. “The

16

ordinary meaning of ‘use’ is ‘to put into action or service.’” Med. Sol’ns, Inc. v. C Change

17

Surgical LLC, 541 F.3d 1136, 1141 (Fed. Cir. 2008) (quoting NTP, 418 F.3d at 1317). The

18

Federal Circuit has held that displaying a product and distributing brochures were not “uses”

19

because neither act put the product into service. See Med. Sol’ns, 541 F.3d at 1141 (“Much more

20

would be needed to qualify as an infringing use, including that the device was used to heat

21

medical items at the show.”). And other courts have held that “using a device means using it to

22

perform its actual function or service, not using it as a demonstrative display.” Id. at 1141 n.4

23

(citing Union Asbestos & Rubber Co. v. Evans Prods. Co., 328 F.2d 949, 951 (7th Cir. 1964);

24

Advanced Semiconductor Materials Am., Inc. v. Applied Materials, Inc., No. 93-20853, 1995 WL

25

419747, at *6 (N.D. Cal. July 10, 1995)) (emphasis added). Simply making source code available

26

on a website—whether for a price or for free—does not, under that definition, put the product into

27

service. While stored on a network server and available via a website, the Android source code

28

does not “perform its actual function or service.” Id. To do that, it must first be downloaded and
11
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

compiled by a developer. To adopt Oracle’s contrary position would make the term “use” so

2

broad that it would render the other statutory terms superfluous.

3

In short, there is no evidence that Google performs any infringing acts. Judgment as a

4

matter of law is therefore appropriate.

5

V.

6

THE ’104 PATENT IS INVALID FOR LATE BROADENING OF REISSUE
CLAIMS UNDER 35 U.S.C. § 251.
The ’104 patent is invalid for failure to comply with 35 U.S.C. § 251, which requires

7
broadening reissue applications to be filed within two years of the date the original patent issued.
8
There is no dispute that the claims of the ’104 patent are broader than the claims of the patent
9
from which it reissued, U.S. Patent No. 5,367,685 (“the ’685 patent”). See TX 4015 (reflecting
10
that reissued claims 11-41 are not limited to a hybrid-compiler-interpreter, unlike original claims
11
1-10, which require a compiler). Nor is there any dispute that the ’104 patent resulted from a
12
continuing application that was filed more than two years after the ’685 patent issued. See id. at
13
[22], [64] (reflecting Mar. 3, 1999 as filing date of ’104 patent application, and Nov. 22, 1994 as
14
issue date of ’685 patent). The patent statute requires that broadening reissue applications be
15
filed within two years of issuance of the original patent. 35 U.S.C. § 251. Although the U.S.
16
Court of Appeals for the Federal Circuit has recently concluded that broadening claims filed more
17
than two years after the issue date of the original patent may be valid if the reissue matures out of
18
a continuation of a timely-filed broadening application, In re Staats, 671 F.3d 1350, 1356 (Fed.
19
Cir. 2012), the Court essentially invited a petition for rehearing en banc. Id. at 1356. Google
20
believes that the case will be reheard and the decision will be reversed, requiring the invalidation
21
of the claims of the ’104 patent under the plain terms of the statute. Google therefore moves for
22
judgment as a matter of law to preserve this issue for appeal.
23
Notably, while the claims of the ’104 patent are narrower than those of the ’685 patent
24
because they all include the limitation that symbolic references be contained in the instructions,
25
the claims of the ’104 patent are also broader in that they eliminate the requirement of generating
26
executable code and compilation means. See TX 4015. Thus, though narrower in one respect,
27
the claims of the ’104 patent have nevertheless been broadened for purposes of 35 U.S.C. § 251.
28
12
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

A claim of a reissue application is broader in scope than the original
claims if it contains within its scope any conceivable apparatus or process which
would not have infringed the original patent. A reissue claim that is broader in
any respect is considered to be broader than the original claims even though it
may be narrower in other respects.

2
3
4

See Tillotson, Ltd. v. Walbro Corp., 831 F.2d 1033, 1037 n.2 (Fed. Cir. 1987) (emphasis added).

5
6

For these reasons, Google is entitled to judgment as a matter of law that the claims of the
’104 patent are invalid.

7
8

VI.

AS TO COUNTS V AND VII, GOOGLE IS ENTITLED TO JUDGMENT AS A
MATTER OF LAW OF NO WILLFUL INFRINGEMENT

9

It is undisputed that Google did not receive actual notice of the ‘104 patent or the ’520

10

patent—a prerequisite for a finding of willful infringement—until July 20, 2010, only twenty-

11

three days before this lawsuit was filed on August 12, 2010. RT 3100:10-3101:2. See In re

12

Seagate Technology, LLC, 497 F.3d 1360, 1371 (Fed. Cir. 2007); see also Gustafson, Inc. v.

13

Intersystems Industrial Prods., Inc., 897 F.2d 508, 511 (Fed. Cir. 1990) (“[A] party cannot be

14

found to have ‘willfully’ infringed a patent of which the party had no knowledge.”).
To establish willful infringement, Oracle must first show by clear and convincing

15
16

evidence that Google acted “despite an objectively high likelihood that its actions constituted

17

infringement of a valid patent.” Seagate, 497 F.3d at 1371. Three weeks of total pre-suit

18

exposure to and awareness of each of the original seven patents-in-suit is per se insufficient to

19

support a finding of objective recklessness as required under Seagate. See Finisar Corp. v.

20

DirecTV Group, Inc., 523 F.3d 1323, 1339 (Fed. Cir. 2008) (finding no willful infringement even

21

where accused infringer did not obtain opinion of counsel until over one year after it received a

22

notice letter from patentee); see also Gustafson, 897 F.2d at 511 (“Nor is there a universal rule

23

that to avoid willfulness one must cease manufacture of a product immediately upon learning of a

24

patent, or upon receipt of a patentee’s charge of infringement, or upon the filing of suit.”).

25

Objective recklessness cannot be established by any action taken by, or inaction by, Google

26

during the three weeks following notice of seven complex patents. 2 Following actual notice of

27

2

28

As ordered by the Court on May 14, 2012, Oracle may not rely on any post-filing activity to
prove willful patent infringement by Google:
13
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

the patents-in-suit, Google promptly studied the patents, investigated the merits of Oracle’s patent

2

infringement allegations and developed sound legal defenses to those allegations. See, e.g.,

3

Answer to Plaintiff’s Complaint (Dkt. No. 32); see also Gustafson, 897 F.2d at 511 (“Exercising

4

due care, a party may continue to manufacture and may present what in good faith it believes to

5

be a legitimate defense without risk of being found on that basis alone a willful infringer. That

6

such a defense proves unsuccessful does not establish that infringement was willful.”) (citation

7

omitted).

8
9

No evidence at trial suggested that Google received actual notice of the ‘104 patent or the
’520 patent at any point in time prior to July 20, 2010. [See, e.g., RT 3771:20-:25 (Gupta Dep.

10

Tr. (July 26, 2011) 133:2-:15, 239:9-:20) (explaining that Java-related patents never came up

11

during the partnership negotiations between Google and Sun).] Oracle attempts to base Google’s

12

actual notice on a book co-authored by Tim Lindholm in 1997, which refers to U.S. Patent No.

13

5,367,685 (“the ’685 patent”), the original patent that eventually reissued as the ’104 patent. TX

14

25, TX 4015. This book alone is insufficient to establish that Google had actual notice of the

15

’104 patent prior to July 20, 2010. First, the reference to the ’685 patent in the book predates Mr.

16

Lindholm’s arrival at Google in 2005 by eight years [RT 3033:5-6] and his August 6, 2010 email

17

to Andy Rubin and Ben Lee regarding a Java license by thirteen years. [TX 10.] Since joining

18

Google, he has had no technical involvement with Android or Java. [RT 3033:5-3035:1.]

19

Second, Mr. Lindholm testified that he was personally unfamiliar with the ’685 patent and did not

20

recall who included it in the book. [RT 3022:17-3023:13.] Indeed, in the second version of the

21

book, published in 1999, the entire chapter referring to the ’685 patent was removed. [TX 987.]

22

Third, even if Mr. Lindholm had actual notice of the ’685 patent, such notice is insufficient

23
24
25
26

Here is the answer. This is going to be a mess if we get into post complaint. So
no post complaint. Period. We’re going to go with the ordinary rule under
Seagate and we will craft a sentence in the jury instruction that tells them not to
speculate about events that occurred after the filing of the complaint and that the
period for proving willfulness is up to the date of the complaint. . . . So the ruling
is going to be, we measure things up to the date of the filing of the complaint.

27
28

[RT 3902:15-3903:17 (Court).]
14
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

grounds to infer that he also had actual notice of a subsequent reissue patent. See Anascape, Ltd.

2

V. Microsoft Corp., Civil Action No. 9:06-CV-158, 2008 WL 7182476, *2-3 (E.D. Tex. Apr. 25,

3

2008) (finding pre-suit notice of parent patents to the patent-in-suit insufficient to establish actual

4

notice of the patent-in-suit and therefore insufficient as a matter of law to establish willful

5

infringement). Finally, as described in section V above, the claims of the ’104 patent are very

6

different from the claims of the ’685 patent—being both broader and narrower in scope.

7

Furthermore, even assuming Google had some general awareness that Sun had a patent

8

portfolio, general knowledge of a patent portfolio is insufficient to support a finding of that

9

Google acted in an objectively reckless manner specifically with respect to the patents-in-suit in

10

the years leading up to this lawsuit. This is particularly true in light of the fact that Oracle, and its

11

predecessor Sun Microsystems, actively encouraged and promoted Google’s development and

12

release of the Android platform. [See, e.g., TX 2352 (Sun CEO Jonathan Schwartz’s official

13

corporate blog post, dated Nov. 5, 2007, stating, “I just want to add my voice to the chorus of

14

others from Sun in offering my heartfelt congratulations to Google on the announcement of their

15

new Java/Linux phone platform, Android. Congratulations! . . . Google and the Open Handset

16

Alliance just strapped another set of rockets to the community’s momentum . . . . With friends

17

like Google and Red Hat, it sure seems like the momentum behind Java’s on the rise…”);3 TX

18

2939 (Larry Ellison’s public statement at JavaOne 2009, “We’re flattered by Android. We’re

19

flattered. Android is very exciting. Everyone should be flattered. I think we can see lots of Java

20

devices. Some coming from our friends at Google.”); TX 3103 (Eric Klein’s public

21

demonstration of Java FX running on Android at JavaOne 2008).] In light of these public

22

statements, it is unsurprising—and entirely reasonable—for Google to have believed that it had

23

no reason to investigate the thousands of patents in Sun’s portfolio. 4 [See, e.g., RT 1538:24-

24

1539:15 (Schmidt) (“Sun management was comfortable that we had done -- that what we had

25

3

26
27
28

The record demonstrates that Mr. Schwartz’s blog reflected the official corporate opinions of
Sun Microsystems rather than merely his own personal views. See TX 971 (referring to “CEO
blogs” as a source of Sun’s “notifications” of its “material news” in its 2008 Form 10-K annual
report to the SEC).
4
Oracle’s Chief Corporate Architect, Edward Screven, testified that Sun had thousands of patents
related to Java. RT 549:12-:14 (Screven).
15
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

done was free and clear of any intellectual property of Sun’s.”); TX 1128 (Rubin Dep. Tr. (July

2

27, 2011) 14:10-:12, 14:14-15:07, 15:09-:13) (explaining that it was “not feasible to look at all

3

the patents out there in the patent system”).] In the face of such overwhelming evidence of public

4

support by Oracle and Sun—and the lack of any evidence of actual notice of the specific patents

5

at issue prior to July 20, 2010—no reasonable jury could find that Google acted in an objectively

6

reckless manner.

7

Moreover, following the July 20, 2010 notice of the patents-in-suit and the filing of Oracle

8

complaint, Google developed a series of legitimate defenses to Oracle’s patent infringement

9

allegations following its analysis of the patents-in-suit. [See, e.g., Answer to Plaintiff’s

10

Complaint Dkt. No. 32 (asserting legal defenses of non-infringement, invalidity, unenforceability,

11

etc.).] Nothing in the record suggests that Google’s defenses were made frivolously or in bad

12

faith. Google’s good faith assertion of legitimate defenses to Oracle’s allegations of patent

13

infringement cannot by itself serve as a basis for finding Google a willful infringer. Gustafson,

14

897 F.2d at 511; see also Advanced Fiber Technologies (AFT) Trust v. J&L Fiber Services, Inc.,

15

674 F.3d 1365, 1377 (Fed. Cir. 2012) (affirming summary judgment of no willful infringement

16

based in part on defendant’s presentation of “compelling non-infringement and invalidity

17

arguments”); Spine Solutions, Inc. v. Medtronic Sofamor Danek USA, Inc., 620 F.3d 1305, 1319

18

(Fed. Cir. 2010) (The objective prong “tends not to be met where an accused infringer relies on a

19

reasonable defense to a charge of infringement.”).

20

Finally, where the threshold objective prong of Seagate is met, “the patentee must also

21

demonstrate that this objectively-defined risk (determined by the record developed in the

22

infringement proceeding) was either known or so obvious that it should have been known to the

23

accused infringer.” Seagate, 497 F.3d at 1371. Because Oracle failed to carry its burden of proof

24

by clear and convincing evidence that the objective prong has been met based on the record, there

25

can be no finding of willfulness and the Court need not consider the subjective prong. See

26

Accentra Inc. v. Staples, Inc., Case No. CV 07-5862 ABC (RZx), 2011 WL 7563039, *23 (C.D.

27

Cal. Dec. 19, 2011) (citing Uniloc USA, Inc. v. Microsoft Corp., 632 F.3d 1292, 1311 (Fed. Cir.

28

2011); DePuy Spine, Inc. v. Medtronic Sofamor Danek, Inc., 567 F.3d 1314, 1337 (Fed. Cir.
16
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1

2009)). In any event, even if Oracle had been able to meet the clear and convincing standard

2

required by the first, objective prong, Oracle failed to introduce any evidence sufficient to meet

3

the requirements of the second, subjective prong.

4
5

For the foregoing reasons, Google is entitled to judgment as a matter of law that Oracle
has not proven that Google willfully infringed the ’104 Patent or the ‘520 Patent.

6
7

*

*

*

8
9

Google’s Rule 50 motion is based on Google’s Motion (Dkt. 1151), Google’s

10

May 10, 2012 oral motion for judgment as a matter of law [RT 3707-09], this brief in

11

support and the evidence and authorities cited therein, the entire trial record, and such

12

argument as the Court allows on this Motion.

13
14
15
16

Dated: May 16, 2012

KEKER & VAN NEST LLP
By:

17

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

18
19
20
21
22
23
24
25
26
27
28
17

BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA

1
2

ATTESTATION
I, Bruce W. Baber, am the ECF User whose ID and password are being used to file this

3

Google’s Brief In Support Of Motion For Judgment As A Matter Of Law On Counts V And VII

4

of Oracle’s Amended Complaint. In compliance with General Order 45, X.B., I hereby attest that

5

Robert A. Van Nest has concurred in this filing.

6
7

Date: May 16, 2012

/s/ Bruce W. Baber
BRUCE W. BABER

8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
18
BRIEF IN SUPPORT OF GOOGLE’S MOTION FOR JMOL ON COUNTS V AND VII
Case No. 3:10-cv-03561 WHA