|
Oracle v. Google - The "Final" Damages Figures |
|
Tuesday, March 20 2012 @ 08:35 AM EDT
|
Oracle and Google have now responded to the directive issued by Judge Alsup pertaining to the valuation of the remaining patents at issue. In the order striking much of Dr. Cockburn's third attempt at a damages report, Judge Alsup stated:
At the hearing, the undersigned judge raised the question of whether Dr. Cockburn's report provides a way for the jury to calculate damages in the event that Oracle prevails on fewer than all patents in suit. Now the only two remaining patents in suit are the '520 and the '104 patents. The Court understands how Dr. Cockburn's group-and-value methodology could arrive at a value for the '104 patent. But in light of the rulings in this order, it seems to the Court that there is no remaining methodology to place a value on the '520 patent. While this was briefly discussed at the hearing, the parties shall also address how this affects, if at all, the copyright allocation, and how the one-half formula will be presented to the jury if Oracle only asserts one-third of the original denominator, the three patents in suit. By NOON ON MARCH 19, each side shall submit ten-page statements on this issue.
First up is Google. (817 [PDF]; Text]) Google suggests there are only two methods to be applied to determine the individual patent values and one of those is inherently wrong. The right method, according to Google, is provided by Exhibit 35 of the Cockburn report. According to Google, this methodology, which attributes equal value to the top 22 patents in the Sun/Oracle "mobile" portfolio is consistent with the Court's overall order pertaining to the Cockburn report, i.e., all patents are to be valued equally among that top 22.
The alternative approach, which Google suggests is inconsistent with the Court's order, would be based on Exhibit 36 of the Cockburn report. This approach attributes greater value to the six patents that were in contention than the other 16 from among the top 22 patents. It then values each of those six patents equally. The problem is, there are only two patents, not six, being asserted. Google believes this approach is inconsistent with the Court's earlier order.
Given that the copyright valuation is a function of the patent valuation, as suggested by Dr. Cockburn, these two approaches yield the following alternative royalties:
Alternative 1 (the "correct") approach
'104 Patent | $4.7 million |
'205 Patent | $0.2 million |
Copyrights | $32.6 million |
| |
Total | $37.5 million |
Alternative 2 (the "incorrect") approach
'104 Patent | $12.5 million |
'205 Patent | $0.5 million |
Copyrights | $32.6 million |
| |
Total | $46.6 million |
As you can see, in either case we are far from the $6 billion in damages figure that was being tossed around by some patent "experts" a year ago as a virtually certain recovery by Oracle.
So what does Oracle say. (816 [PDF; Text]) Of course, Oracle starts off by quibbling with the $561 million starting figure. In fairness to Oracle, there does appear to be some question as to whether the judge used the right number. So Oracle offers three possible right numbers ($561 million, $566 million, or $554 million) and then uses the most conservative of the three, $554 million.
Oracle then applies the same approach as Google (the Exhibit 35 approach, which Google states is the correct one) and arrives at the following:
'104 Patent | $4.1 million |
'205 Patent | $0.5 million |
Copyrights | $27.7 million |
| |
Total | $32.3 million |
Oracle, however, equivocates a bit on the copyright valuation saying that $27.7 is the lowest it could be. It is not clear to this writer exactly how Oracle would determine that number or whether Oracle would adopt an approach much like that of Google. It is also important to remember that these numbers, both those of Google and Oracle, presume that Google wins on the issue of patent marking. Oracle presumes the marking factor reduces the royalties by about 20%, so there would be some upward adjustment if the marking issue is decided in Oracle's favor.
In any case, we are looking at a total past damage figure that is likely under $100 million, not $6 billion.
In the only other material filing of the day, Oracle sets forth its request to amend the infringement contentions arising from the '205 patent. (814 [PDF; Text]) Oracle concedes this request may be moot if the court proceeds to trial on schedule in April, but it wishes to preserve its motion should, for whatever reason, the trial date slips to a much later date. Much of the Oracle rationale for this motion hinges on its belief, a belief that appears to be inconsistent with the facts, that the judge's interpretation of "at runtime" varies in some material way from the interpretation Google advocated. As we have stated before, that appears to be an astonishing leap. This motion will, in all likelihood, be moot because this trial is not going to get delayed.
*************
Docket
03/19/2012 - 814 - REPLY (re
782 MOTION to Amend/Correct Oracle's '205 Patent Infringement
Contentions and Supplement Expert Reports ) filed byOracle America,
Inc.. (Jacobs, Michael) (Filed on 3/19/2012) (Entered: 03/19/2012)
03/19/2012 - 815 -
Declaration of Marc Peters in Support of 814 Reply to
Opposition/Response, 782 MOTION to Amend/Correct Oracle's '205 Patent
Infringement Contentions and Supplement Expert Reports filed byOracle
America, Inc.. (Attachments: # 1 Exhibit
D)(Related document(s) 814 , 782 ) (Peters, Marc) (Filed on
3/19/2012) (Entered: 03/19/2012)
03/19/2012 - 816 - Statement
re 785 Order on Motion to Strike ORACLE AMERICA, INC.S STATEMENT
RESPONDING TO THE COURTS ORDER CONCERNING DR. COCKBURNS THRID REPORT
[DKT. NO. 785] by Oracle America, Inc.. (Holtzman, Steven) (Filed on
3/19/2012) (Entered: 03/19/2012)
03/19/2012 - 817 - Brief re
785 Order on Motion to Strike filed byGoogle Inc.. (Related document(s)
785 ) (Van Nest, Robert) (Filed on 3/19/2012) (Entered: 03/19/2012)
03/19/2012 - 818 -
Declaration of Dr. Gregory K. Leonard in Support of 817 Brief filed
byGoogle Inc.. (Related document(s) 817 ) (Van Nest, Robert) (Filed on
3/19/2012) (Entered: 03/19/2012)
*************
Documents
814
MORRISON & FOERSTER LLP
MICHAEL A. JACOBS (Bar No. 111664)
[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 adress telephone fax]
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.
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 REPLY
IN SUPPORT OF MOTION TO
AMEND ’205 PATENT
INFRINGEMENT CONTENTIONS
AND SUPPLEMENT EXPERT
REPORTS
Dept: Courtroom 8, 19th Floor
Judge: Honorable William H. Alsup
After Oracle filed its motion to amend its infringement contentions and supplement its
expert reports with respect to the ’205 patent, the Court set the case for trial to begin on April 16.
(ECF No. 786.) It does not appear likely that the ’205 examiner will address the ’205 patent
further before the patent phase of trial begins. Today, Oracle files its response to the ’205
examiner’s rejection, and the deadline for Google to file its comments is April 18. Given
Oracle’s offer to dismiss with prejudice the claims of the ’205 patent that remain rejected as of
the time of a spring trial (ECF No. 777 at 3), the Court may well decide to deny this motion as
moot. If the Court does so, it should be without prejudice. If, for some unforeseen reason, the
trial does not go forward in the spring, the yardsticks by which diligence and prejudice are
measured change substantially, and Oracle should be permitted to re-file its motion.
If the Court wishes to address the merits of Oracle’s motion at this time, it should allow
Oracle to amend its infringement contentions and supplement its expert reports to identify the
specific virtual machine instructions that are executing when Android dexopt tool generates new
virtual machine instructions, thereby infringing the asserted claims of the ’205 patent. Oracle’s
amendments do not alter its theory that dexopt’s inline substitution functionality infringes the
asserted claims of the ’205 patent. Oracle’s theory has always included the time when dexopt is
running. The infringement contentions disclose that dexopt can be run “at runtime” by the
Android installer as well as in a “just-in-time” fashion on a development device.
Google’s selectively quotes from Prof. Mitchell’s reports to remove context and argue that
what he said means something different. For example, Google quotes (using bold italics, no less)
Prof. Mitchell’s statement that Android’s dexopt obtains needed information “by initializing a
Dalvik VM and loading all the classes from an application’s .dex file into the VM.” (Google
Opp. at 2-3.) That is a true statement. But it has nothing to do with whether the application is
running. Neither does the full paragraph from which Google plucked its sentence:
Prof. August fails to consider the evidence that dexopt runs at runtime. As Google
wrote: “[t]he solution is to invoke a program called dexopt, which is really just a
back door into the VM. It performs an abbreviated VM initialization, loads zero or
more DEX files from the bootstrap class path, and then sets about verifying and
optimizing whatever it can from the target DEX. On completion, the process exits,
freeing all resources.” (See
http://android.git.kernel.org/?p=platform/dalvik.git;a=blob;f=docs/dexopt.html.)
1
The optimizations that dexopt performs (such as replacing symbolic references
with vtable offsets) are ones that require information that is only available at
runtime (such as a vtable offset), and that information is obtained by initializing a
Dalvik VM and loading all the classes from an application’s .dex file into the VM.
Google confirms it: “[s]ome of [the optimizations performed by dexopt] require
information only available at runtime, others can be inferred statically when
certain assumptions are made.” (See id.) The way Google wrote dexopt, rewriting
the DEX method invocation instructions with EXECUTE_INLINE and
EXECUTE_INLINE_RANGE instructions is one of the former kind of
optimizations.
(Francis Decl. Ex. B (Mitchell Reply Report) ¶ 60.) Prof. Mitchell’s argument here is that one
can tell that dexopt runs at runtime because it requires information available only at runtime, as
Google admits in its dexopt documentation. He did not argue that dexopt runs at runtime because
it then proceeds to run the application it just optimized. Prof. Mitchell acknowledges the
opposite, in fact, when he quotes Google’s dexopt documentation to the effect that dexopt loads
the application into a Dalvik VM but then exits upon completion of the optimization. (Id.)
Google’s quotations from Prof. Mitchell’s deposition likewise take his testimony out of
context and—worse—fail to quote testimony that directly contradicts its argument. With respect
to the testimony that Google did quote, Google’s elisions disguise the fact that Prof. Mitchell was
not testifying with respect to the ’205 patent about dexopt running on an Android phone but was
instead testifying with respect to the ’104 patent about an experiment he performed using
Android’s development environment in which he compiled an application and then installed and
ran it on the Android SDK’s emulator:
Q. BY MR. KAMBER: Professor Mitchell, I just want to ask a few questions to
finish up on the ’104, because we need to move to the ’205 patent. You would
agree with me that the output of running dexopt on a .dex file is a .odex file;
correct?
A. I believe that may be one option. When I installed the SDK and used this to run
code, I didn’t generate an output like that. It’s just part of the process of running
the application that I compiled and translated to Dalvik. And then the run process
of executing that in the Dalvik Virtual Machine, I believe, was what -- where the
dexopt got invoked.
Q. So did you only run this by way of the SDK?
A. That’s the -- an experiment that I did. Anytime you run an Android phone, you
may also run this, so that’s another example.
(Francis Decl. Ex. C (Mitchell Dep.) at 342:7-22 (portions quoted by Google highlighted).) (The
2
word “this” in the second answer refers to dexopt, which indeed may run anytime one runs an
Android phone, according to the source code in PackageManagerService.java.)
The testimony that Google should have quoted in its opposition is when Prof. Mitchell
testified that dexopt’s “install time” qualifies as “runtime,” and that an example of why that is so
could be found in the portion of his report that discusses the ’205 patent:
Q. How did you determine that dexopt was at runtime?
A. Through statements from Google describing the way that this operates. There
are slides, again, and quotations and sections from documentation, I believe. And
also looked at the code and understood how this behaves and the nature of the
steps and the information required in order to complete those steps.
Q. Let’s look at some of that documentation. Take a look at page 90. You have a
clip of a slide and then a quote from Mr. Bornstein’s presentation here on page 90;
correct?
A. This certainly is a slide and a quote. I believe it’s from one of Dan Bornstein’s
presentations. It may be the one we identified earlier.
Q. And the quote, you say that -- you give at 21:54 under, quote, “install time
work.” Do you see that?
A. Yes.
Q. Okay. So what Mr. Bornstein is talking about that you cite to relates to install
time on a device; correct?
A. That’s the way he’s characterized this portion of the operation, yes.
* * *
Q. Why do you think that this work that’s being done at install time qualifies as
runtime under the terms of the patent?
A. I think what Dan Bornstein is trying to emphasize is things that are done once.
Although, there are other documentation information that’s explicit on the runtime
nature. I believe, for example, the documentation that I reproduced on page 178
and 179, that explains it in slightly more technical terms.
Q. Well, what you said on 178 and 179 relates to the ’205 patent; correct?
A. It’s presented there. It’s an explanation of how -- it’s an explanation of dexopt
from the Android website. It’s the same dexopt in both patents. I don’t know if
it’s also -- that same section is also cited here in connection with ’104. That’s just
one place that I knew I had put that information. So that’s why I pointed to it now.
(Peters Supp. Decl. Ex. D (Mitchell Dep.) at 329:25-332:12.) Note that paragraph 402 of the
Mitchell report (which spans pages 178 and 179) includes the paragraph about the installer that
3
Google quotes on page 3 of its opposition. (Compare Google Opp. at 3 with Francis Decl. Ex. A
(Mitchell Report) at ¶ 402.) There can be no question that Prof. Mitchell advanced an
infringement theory based on the “install time work” performed by dexopt, because it is right
there in his testimony and his report. Oracle’s original contentions, which identify dexopt as
performing the step of “generating, at runtime, a new virtual machine instruction” when it
performs inline substitutions, likewise quote Google’s admission that dexopt performs these
optimizations “either in the build system or by the installer. On a development device, dexopt
may be run the first time a DEX file is used and whenever it or one of its dependencies is updated
(‘just-in-time’ optimization and verification).” (Peters Decl. Ex. A at ¶ 13.) In its infringement
contentions, Oracle accused non-build-system, non-development devices (such as the Droid) of
infringement—for these devices, how could Google reasonably believe that Oracle was basing its
dexopt infringement theory on anything but the installer?
In its opposition, Google fails to address the key fact that pops the balloon of its argument:
the Android installer, which runs dexopt, is itself an application that runs in a Dalvik virtual
machine. As Prof. Mitchell lays out in detail in his supplemental report, the virtual machine
instructions of the installer are the ones executing when dexopt optimizes an application during
its installation. (Peters Decl. Ex. B at 8-14.) As Google knows better than anyone, and as Prof.
Mitchell testified, “install time” on an Android device is runtime. This is why Google’s
documentation (disclosed in both Oracle’s infringement contentions and Prof. Mitchell’s reports)
states that dexopt requires information “only available at runtime.” (Francis Decl. Ex. B
(Mitchell Reply Report) ¶ 60.) In truth, Google’s argument about what Oracle’s position “has
always been” is based on either Google’s misunderstanding or wishful thinking colored by its
non-infringement position, rather than on an objective assessment of Oracle’s infringement
contentions.
As stated in its opening brief, Oracle has good cause to amend because the Court’s
construction of “at runtime” was not only different from Oracle’s proposal but also quite different
from Google’s ambiguous construction. Oracle provided the proposed amendment and
supplemental reports to Google about a month after the Court construed the phrase “at runtime.”
4
Oracle did so in less time than other parties did that were found to be diligent. Schindler Elevator
Corp. v. Otis Elevator Co., No. 06 Civ. 5377 (CM)(THK), 2010 U.S. Dist. LEXIS 110313, at
*11-12 (S.D.N.Y. Oct. 6, 2010) (allowing plaintiff’s expert to disclose revised infringement
theory seventy-four days after issuance of claim construction order because expert “needed some
time to digest the court’s rulings and revisit the file wrapper”). If, for some reason, the trial does
not take place in the spring, Google will not be prejudiced by the Court allowing this amendment,
for Google will have ample time to prepare its non-infringement analysis and expert report.
With respect to Google’s request for leave to amend its invalidity contentions, Google has
not provided the Court or Oracle with its proposed amendments or even specifically identified the
prior art that would be included in the amendment.1 Oracle cannot evaluate whether the Court’s
construction of “at runtime” would justify Google’s amendment under Patent Local Rule 3-6 or
whether Oracle would be prejudiced by it.
Dated: March 19, 2012
MICHAEL A. JACOBS
MARC DAVID PETERS
DANIEL P. MUINO
MORRISON & FOERSTER LLP
By: /s/ Michael A. Jacobs
Attorneys for Plaintiff
ORACLE AMERICA, INC.
___________________________________________
1 Google’s mention of a Deutsch & Schiffman publication is particularly hard to understand.
Google submitted a ’205 anticipation contention based on a Deutsch & Schiffman article in its
original invalidity contentions in January 2011, but withdrew the contention in June 2011.
Google moved the Court to amend its invalidity contentions to add an obviousness contention
based on a combination of the Deutsch & Schiffman article with a Magnusson article in July
2011, which the Court granted (ECF No. 281 at 7-8), but Google withdrew the contention in
October 2011. (ECF No. 475.) Google submitted a claim chart based on the Deutsch &
Schiffman article in the inter partes reexamination, but the examiner concluded that the article
did not raise a substantial new question of patentability, and so it was excluded from the
reexamination. Google may have a different Deutsch & Schiffman reference in mind.
5
816
MORRISON & FOERSTER LLP
MICHAEL A. JACOBS (Bar No. 111664)
[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 adress telephone fax]
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.
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 STATEMENT
RESPONDING TO THE COURT’S ORDER
CONCERNING DR. COCKBURN’S THRID
REPORT [DKT. NO. 785]
Dept: Courtroom 8, 19th Floor
Judge: Honorable William H. Alsup
Oracle America, Inc. submits this statement in response to the Court’s order concerning Prof.
Cockburn’s third report. (Dkt. 785.) The Court directed the parties to address three issues: (1) how
Prof. Cockburn calculates the value of the ’520 patent, (2) whether Oracle’s dismissal of patents other
than the ’104 and ’520 affects the copyright allocation, and (3) how the allocation of the value of the
copyrights in the 2006 bundle will be presented to the jury if Oracle asserts only the ’104 and ’520
patents at trial. (Id. at 18-19.)
I. Request For Clarification of Starting Point Adjustment
In order to provide calculations along with the explanations requested by the Court, Oracle
requests a clarification of the Court’s required adjustment to the starting point.
In the group and value approach described in his Report, Prof. Cockburn deducted $86.15
million from the 2006 starting point to reflect expenses that Sun expected to incur in connection with
Project Armstrong, including expenses to develop and implement additional class libraries and to
develop a virtual machine to Android’s specifications. (Cockburn Third Report ¶¶ 363, 371, 382.) In
its Order, the Court directed Prof. Cockburn to make an additional deduction: to “adjust his group-andvalue
calculation by deducting $37 million, his calculated value of the unasserted copyrights, from the
adjusted starting point of $598 million. Accordingly, $561 million shall be the total value of the
copyrights in suit and 569 patents in Sun’s Java mobile patent portfolio.” (Dkt. 785 at 10-11.)
Oracle seeks clarification of how the Court calculated the $37 million figure. Prof. Cockburn
determined that the cost of writing code to develop and implement all of the class libraries in Android,
including class libraries that did not yet exist, was no greater than $32.4 million. (Cockburn Third
Report ¶ 379 & Exh. 30.) Separately, Prof. Cockburn determined that the value to Google of taking an
already-written virtual machine from Sun could be no greater than $11.3 million, which is Google’s
entire actual cost of developing the Dalvik VM. (Id. ¶ 366.) The $37 million measure does not appear
in Prof Cockburn’s report as the value of any unasserted copyrights.
The Court may have calculated the $37 million by adding the $3.8 million that appears in
Exhibit 29 (all Google engineering internal engineering costs for all APIs) and the $32.41 million that
appears in Exhibit 30 (Google cost of developing Android packages, libraries, and APIs), of Prof.
Cockburn’s third report. (These two figures add up to $36.2 million.) However, it would be incorrect
1
to add those together because the relevant portion of $3.8 million calculated in Exhibit 29 is already
included in Exhibit 30. The total engineering expense associated with API packages that Google
possibly could have avoided by partnering with Sun is $32.4 million, not $37 million. (Id. ¶¶ 379-80.)
It is thus unclear to Oracle whether the Court intended to require
(1) a deduction for the cost of implementing all class libraries (both existing and to be
developed), which would be $32.4 million;
(2) a deduction for the cost of implementing all class libraries (both existing and to be
developed) and for the full cost of developing the Dalvik virtual machine, which would be
$43.7 million; or
(3) some other deduction or combination of deductions that equal $37 million.
Absent further guidance, Oracle will assume that the Court intended the second set of
deductions described above, amounting to $43.7 million.1 All calculations below assume a further
deduction of $43.7 million, and consequently adjust the relevant starting point to $553.8 million. If the
Court orders Prof. Cockburn to use some other amount, Prof. Cockburn can re-calculate these amounts.
II. Calculation of Patent Damages Pursuant to the Court’s Order
The Court ruled that Prof. Cockburn would be permitted to offer a reasonable royalty opinion
based only on his lower bound calculation, “in which each patent in his top 22 has equal value to each
other.” (Dkt. 785 at 5.) Using the adjusted starting point of $553.8 million, and after allocating $27.7
million of the 2006 Bundle to the copyrighted APIs,2 the 569 patents in the Sun mobile Java portfolio
would have a total value of at least $526.1 million (based only on Google’s anticipated payments to Sun
and Sun’s Project Armstrong projections, and only for the first three years). Applying the same patent
value distribution curve used in the third report, the top 22 patents in the portfolio of 569 would have an
_______________________________________
1 Oracle still contends that no such further adjustment is appropriate, for the reasons described in Prof.
Cockburn’s report and Oracle’s briefing to the Court. Nonetheless, Oracle understands that the Court
has ruled, and will of course abide by that ruling. Oracle seeks clarification solely to ensure that Prof.
Cockburn is able to explain to the jury the specific bases for, and amounts of, all adjustments to the
2006 starting point.
2 To apply the apportionment approach allowed by the Court, Prof. Cockburn would have to assume
that each of the patents in the top 22 of the 569 has equal value, and that the top 22 patents represent at
least 77.1% of the value of the patents in the 2006 Bundle. As previously disclosed, Prof. Cockburn
would further assume that the value of the copyrights in the Bundle is one half the combined value of
the ’104, ’205, and ’720 patents. A formula expressing this relationship, which can be solved to
provide numbers for the patents and copyrights, is (22x / 0.771) + (3x / 2) = $553.8 million, in which x
is the average value of each patent in the top 22.
2
aggregate value of $406 million, and an average value of $18.4 million per patent. Thus, under the only
approach permitted by the Court’s order, the value of the ’104 patent, before any other adjustments,
would be at least $18.4 million for the first three years.3 The other potential adjustments are identified
below:
Value of ’104 patent: | $18.4 million |
US only damages: | $18.1 million |
US only damages and reduced for marking: | $15.3 million |
US only damages, reduced for marking, limited to accused devices: | $4.1 million |
Prof. Cockburn can calculate the three-year value of the ’520 patent, under the only approach
permitted by the Court, based on the average value of the 547 patents in the portfolio that are not
included in the top 22. Again, the 569 patents in the Sun mobile Java portfolio had a total value of
$526.1 million over the first three years; the top 22 patents in the portfolio of 569 would have an
aggregate value of $406 million. Accordingly, the 547 lower-ranked patents would be worth $120
million ($526 – $406) in aggregate, and the average value of those patents, under the only approach
permitted by the Court, would be at least $0.22 million. Because the engineers’ rating of the ’520
patent and its technology block indicated that it was more important, in terms of its technological
significance, than most of the patents in the portfolio, using this average would likely understate the
value of the ’520 patent. Thus, the value of the ’520 patent, before any other adjustments, would be at
least $0.22 million for the first three years. The other potential adjustments are:
Value of ’520 patent: | $0.22 million |
US only damages: | $0.22 million |
US only damages and reduced for marking: | $0.18 million |
US only damages, reduced for marking, limited to accused devices: | $0.05 million |
These calculations are consistent with the methodology approved by the Court. If necessary,
Prof. Cockburn could serve a one-page supplement to his report that includes these calculations.
________________________________________
3 Oracle uses the term “value” in this statement to be consistent with the Court’s order. (Dkt. 785 at 18-
19.) As explained in Prof. Cockburn’s report, these calculations do not account for substantial
unquantifiable losses to Oracle (including fragmentation) and gains to Google. These calculations do
not represent the actual value of the intellectual property or the full measure of damages, but rather
represent the minimum value for the two patents and copyrights based on the lower bound of the groupand-
value method for a three-year license.
3
III. Calculation of Copyright Damages Pursuant to the Court’s Order
The Court directed the parties to address how the order on Prof. Cockburn’s third report
“affects, if at all, the copyright allocation, and how the one-half formula will be presented to the jury if
Oracle only asserts one-third of the original denominator, the three patents in suit.” (Dkt. 785 at 19
(emphasis in original).)
In his third report, Prof. Cockburn concluded that if the lower bound of the group and value
apportionment method must be applied, the copyright lost license fee is at least $34.7 million through
the end of 2011. (Cockburn Third Report ¶ 6.) This amount, like the patent apportionment, was based
only on Google’s anticipated payments to Sun and Sun’s Project Armstrong projections for the first
three years, and does not fully account for the substantial harm that Sun would have reasonably
expected from an incompatible Android. This amount must be re-calculated based on the Court’s Order
and the requirement of an additional reduction of the starting point. Removing $43.7 million (assuming
this is the required reductions for the value of unasserted copyrights) from the starting point and
assuming that each of the ’104, ’205, and ’720 patents have an equal value, this amount becomes at
least $27.7 million. (See footnote 2, above.)
Other than that adjustment, the Court’s order does not disturb Prof. Cockburn’s allocation
between the copyrights and the rest of the 2006 Bundle. At the March 7 hearing, both counsel for
Google and counsel for Oracle agreed that even if the patents were invalid or not infringed, the value
attributed to the copyrights would still be the same. (See 3/7/2012 Tr. at 107:16-108:3 (Google counsel
agreeing that if Prof. Cockburn were allowed to use the 2:1 ratio indicated by the conjoint survey, “that
would allow for a patent by patent liability determination” “[w]ith the copyright, still, I think being half
of what the broader number would have been had all those patents been invalid and infringed.”);
3/7/2012 Tr. at 108:25-109:21 (Oracle counsel agreeing that the “copyright API number remains the
same regardless, regardless of which patents are valid or not”).)
The conjoint survey did not directly measure the value of copyrights or patents. It measured the
consumer demand for particular Android attributes enabled by Google’s copyright and patent
infringement. As previously disclosed, Oracle engineers disabled the Android features that are enabled
by infringement of the ’104, ’205, and ’720 patents, alone and in various combinations. When they did
4
so, application launch speed was adversely affected. (Cockburn Third Report ¶¶ 24, 428, 456, Exhs. 4-
5). The conjoint survey measured the demand for the incremental increase in application launch speed
enabled by those infringing Android features (as well as the features that infringe only the ’104 and
’205 patents). The conjoint survey also measured consumer demand for having a large number of
available applications, an attribute associated with the infringing use of the copyrighted APIs.
As Prof. Cockburn has explained, the data yielded by the survey indicates that consumers value
the incremental speed benefits twice as much as they value the incremental number of applications.
That relationship exists independent of any particular claim of infringement. Moreover, using the
group and value approach in the manner permitted by the Court, a minimum value for the ’104, ’205,
and ’720 patents can be established. That minimum value is derived by apportioning the 2006 Bundle
in the manner permitted by the Court, using the Java engineers’ analysis and the PatVal distribution
curve. Consequently, that minimum apportioned value for those three patents, for a three-year license,
is independent of any claim that the ’104, ’205, and ’720 patents are in fact valid or infringed.
At trial, Prof. Cockburn can (1) explain to the jury the relationship between the value of
incremental speed and incremental applications as demonstrated by the conjoint analysis, (2) explain
that the combination of the ’104, ’205, and ’720 patents would be expected to provide, and in fact has
been demonstrated to provide, performance enhancements equivalent to the application launch speed
benefit tested in the conjoint analysis, and (3) testify as to the combined value of those three patents as
indicated by the lower bound group and value approach permitted by the Court. This testimony would
allow the jury to conclude that the value of the copyrighted APIs in the 2006 bundle would be worth at
least one half of the apportioned value of the ’104, ’205, and ’720 patents, using the lower bound group
and value approach permitted by the Court. This would be the case regardless of the number of patents
that Oracle asserts are infringed.
It should be unnecessary, in that testimony, to discuss whether the ’205 or ’720 patent was ever
asserted against Google, is valid, or is infringed. If, nonetheless, it became necessary to explain that
patents had been asserted and withdrawn, that fact can be explained to the jury without explaining why
they were withdrawn, and in a way that prevents any prejudice to either party. Indeed, any reference to
the reasons for withdrawal should be excluded, as irrelevant to the merits of the claims that remain in
5
suit, and unduly prejudicial to Oracle. See Fed.. R. Evid. 401, 403; Cardiovention, Inc. v. Medtronic,
Inc., 483 F. Supp. 2d 830, 838-39 (D. Minn. 2007).
Dated: March 19, 2012
BOIES, SCHILLER & FLEXNER LLP
By: /s/ Steven C. Holtzman
Steven C. Holtzman
Attorneys for Plaintiff
ORACLE AMERICA, INC.
6
817
KEKER & VAN NEST LLP
ROBERT A. VAN NEST - # 84065
[email]
CHRISTA M. ANDERSON - # 184325
[email]
DANIEL PURCELL - # 191424
dpurcell@kvn.com
[email address telephone fax]
KING & SPALDING LLP
SCOTT T. WEINGAERTNER
(Pro Hac Vice)
[email]
ROBERT F. PERRY
[email]
BRUCE W. BABER (Pro Hac Vice)
[address telephone fax]
KING & SPALDING LLP
DONALD F. ZIMMER, JR. - #112279
[email]
CHERYL A. SABNIS - #224323
[email address telephone fax]
IAN C. BALLON - #141819
[email]
HEATHER MEEKER - #172148
[email]
GREENBERG TRAURIG, LLP
[address telephone fax]
Attorneys for Defendant
GOOGLE INC.
UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
SAN FRANCISCO DIVISION
ORACLE AMERICA, INC.,
Plaintiff,
v.
GOOGLE INC.,
Defendant.
Case No. 3:10-cv-03651 WHA
GOOGLE INC.’S BRIEF RE
CALCULATION OF REASONABLE
ROYALTY FOR REMAINING PATENTS
AND COPYRIGHTS IN RESPONSE TO
COURT’S DAUBERT ORDER [DKT. NO.
785]
Dept.: Courtroom 8, 19th Floor
Judge: Hon. William Alsup
I. INTRODUCTION
In its Order Granting in Part and Denying in Part Google’s Daubert Motion to Exclude
Dr. Cockburn’s Third Report [Dkt. No. 785], the Court requested that the parties advise the Court
concerning how Dr. Cockburn’s report could be used to calculate a reasonable royalty for each
remaining patent in light of the items stricken by the Order. Dkt. No. 785 at 19. Dr. Cockburn’s
report provides two potential ways of calculating a reasonable royalty for the remaining patents.1
First, Dr. Cockburn could adjust the calculations in Exhibit 35 of his Report to reflect that
(1) he will only be permitted to opine that the entire bundle is worth $560.5 million rather than
$597.5 million, and (2) Oracle has dropped all patents other than the ’104 and ’520 from the case.
This adjustment would allow Dr. Cockburn to testify to an unadjusted reasonable royalty for the
’104 patent of $18.5 million and an unadjusted reasonable royalty for the ’520 patent of $0.7
million. Declaration of Dr. Gregory K. Leonard (“Leonard Decl.”) ¶ II.A.6. This accords with
the Court’s conclusion regarding what Dr. Cockburn believes to be the top set of patents that “Dr.
Cockburn can only opine on this ‘lower bound’ calculation ($20 million per patent before
offsets), and not the ‘upper bound’ calculation.” Dkt. No. 785 at 7-8 (emphasis in original).
Second, Oracle may argue that Dr. Cockburn can value the patents based on the figures in
Exhibit 36 of Dr. Cockburn’s report. In Exhibit 36, Dr. Cockburn purports to list the proportion
of the value of all six patents attributable to each individual patent. This would yield an
unadjusted reasonable royalty for the ’104 of $48.5 million and an unadjusted reasonable royalty
for the ’520 of $2.1 million. Leonard Decl. ¶ II.B. But this approach is not only inconsistent with
the more robust analysis set forth in Exhibit 35, it does not account for the critical fact that there
are now only two patents left in this case, rather than six. Furthermore, this approach contradicts
the logic of the Court’s most recent Daubert Order. Accordingly, and as is explained in more
detail in the declaration of Google’s expert Dr. Gregory Leonard, filed with this brief, the only
reliable approach under Dr. Cockburn’s methodology is to use Exhibit 35.
_________________________________________
1 In submitting this statement, Google does not intend to support the correctness of what remains
of Dr. Cockburn’s analysis, or the damages figures discussed herein, all of which Google intends
to contest at trial. This brief only provides Google’s position concerning what testimony Dr.
Cockburn’s may offer at trial in light of the Court’s recent Daubert Order.
1
Dr. Cockburn’s copyright valuation also must be adjusted to account for the Court's
adjustment to his valuation of the overall 2006 Sun intellectual property bundle, which
established a new cap of $560.5 million, down from the previous total of $597.5 million. Making
that adjustment, Dr. Cockburn’s value of the copyrights-in-suit should be adjusted from $34.7
million to $32.6 million. Leonard Decl. ¶ II.A.5.
II. DISCUSSION
A. Valuation of the remaining patents-in-suit (’104 and ’205)
Dr. Cockburn’s report provides two potential ways Dr. Cockburn could calculate the value
of the two remaining patents-in-suit.
First, and most logically, Dr. Cockburn could adjust his calculations in Exhibit 35 of his
report to account for the Court’s Order and Oracle’s decision to drop all patents other than the
’104 and ’205. This adjustment would require the following steps.
- Dr. Cockburn initially calculated the total value of the Sun intellectual property
bundle at $597.5 million. The bundle included the six Sun patents and the handful
of copyrights asserted in this case, along with hundreds of other patents and
copyrighted works. If the value of the unasserted copyrights in the bundle is $37
million, deducting that amount from the total leaves a value of $560.5 million for
the asserted copyrights and all the Sun patents in the bundle. Dkt. No. 785 at 19;
Leonard Decl. at ¶ II.A.1.
- Dr. Cockburn then separated out the value of the six asserted patents from the rest
of the bundle, beginning with an estimation of the value for the 22 purportedly
most valuable patents in the portfolio. Relying on the PatVal study, Dr. Cockburn
concluded that those 22 patents are worth 77% of the total portfolio. As the Court
recognized, the record contains no information that would allow Dr. Cockburn to
opine that any of what he considers to be the top 22 patents is more valuable than
any other. Using the logic of Dr. Cockburn’s lower bound,2 Dr. Cockburn can
testify that the ’104 is worth 1/22 of the total value of what he calls the top 22
patents. Thus Dr. Cockburn can testify that the ’104 is worth (1/22)*(0.77), or
3.5%, of all of the patents. Leonard Decl. ¶ II.A.2.
- Using the same methodology Dr. Cockburn used in Exhibit 35, Dr. Cockburn can
testify that the value of the ’520 patent is 3/75 of the value of the ’104 patent.
Leonard Decl. ¶ II.A.3.
_______________________________________
2 The Court ruled that Dr. Cockburn could value what he calls the top patents, including the ’104,
by using his “lower bound” calculation, in which each of those patents have equal value. Dkt.
No. 785, at 7. It then concluded that “Dr. Cockburn can only opine on this ‘lower bound’
calculation ($20 million per patent before offsets), and not the ‘upper bound’ calculation.” Dkt.
No. 785, at 7-8.
2
- With these adjustments, combined with the revised valuation of the asserted
copyrights at $32.6 million, it is possible to solve for the value of the ’104 patent
using the equation in paragraph 414 of the most recent Cockburn Report.
Cockburn Report ¶ 414. Solving the equation results in a reasonable royalty for
the ’104, prior to any downward adjustments for foreign sales, non-accused
devices, or failure to mark, of $18.5 million. Leonard Decl. ¶ II.A.6.
- Again applying the rule that the ’520 is worth 3/75 as much as the ’104, Dr.
Cockburn can testify that a reasonable royalty for the ’520, prior to any downward
adjustments, is $0.7 million. Leonard Decl. ¶ II.A.6.
- When downward adjustments for foreign sales, non-accused devices, and failure to
mark are made, Dr. Cockburn can testify that the adjusted reasonable royalty is
$4.7 million for the ’104 and $0.2 million for the ‘520, resulting in a total adjusted
royalty of $4.9 million for the two remaining patents. Leonard Decl. ¶ II.D.
Second, Oracle may argue that Dr. Cockburn can testify that the reasonable royalty for the
remaining individual patents could be calculated using Dr. Cockburn’s Exhibit 36, which purports
to apportion the value of the individual patents based on the proportion of the value of the sixpatent
bundle attributable to each of the individual six patents. See Cockburn Report Ex. 36. As
Dr. Leonard explains in his Declaration, adjusting the calculations in Exhibit 36 to account for the
lower total value of the 2006 bundle results in an unadjusted valuation of the ’104 patent of $48.5
million and an unadjusted valuation of the ’520 patent of $2 million. Leonard Decl. ¶ II.B. When
downward adjustments are made, the adjusted reasonable royalty for the ’104 is $12.5 million,
and the adjusted reasonable royalty for the ’520 is $0.5 million, resulting in a total adjusted
reasonable royalty of approximately $12.9 million. Leonard Decl. ¶ II.D.
But, as Dr. Leonard notes, there are several logical problems with valuing the ’104 and
’520 based on the figures in Dr. Cockburn’s Exhibit 36, rather than the calculations in Exhibit 35.
See Leonard Decl. ¶ II.B. Most seriously, the percentages in Exhibit 36 reflect the value of each
individual patent as a percentage of the total value of the six-patent bundle that Oracle was
asserting at the time of Dr. Cockburn’s report. With four of those six patents now out of the case,
it would make little sense (and likely confuse the jury) to offer evidence as to the value of the
‘104 or ‘520 as a percentage of a six-patent bundle that is no longer at issue, and about which the
jury will presumably hear no testimony at trial.
Exhibit 36 is also inconsistent with the Court’s most recent Order. For example, the Court
ordered that Dr. Cockburn can only testify to his “lower bound” calculation, in which, as the
3
Court described it, “each patent in his top 22 has equal value to each other.” Dkt. No. 785 at 7-8.
But Exhibit 36 offers calculations at odds with this principle. Although the ’104, ’205, and ’720
patents are all in Dr. Cockburn’s top 22, Exhibit 36 gives different valuations for those patents,
listing the ’104 and ’205 as each being worth 75% of the six-patent bundle, but the ’720 as being
worth only 40% of the six-patent bundle, about half as much. Cockburn Report, Ex. 36.
Similarly, if the ’104 is worth 75% of the six-patent bundle, that one patent alone would be worth
7.1% of the total 2006 bundle. Leonard Decl. ¶ II.B. That is twice the 3.5% value one would
calculate if one assumes, as the Court’s Order requires, that each of Dr. Cockburn's top 22 patents
(which collectively amount to 77% of the value of the total bundle) is of equal value. Leonard
Decl. ¶ II.A.2. Accordingly, to the extent Dr. Cockburn’s Exhibit 36 could be read to support an
apportionment that is inconsistent with Exhibit 35, the methodology in Exhibit 35 should control.
B. Valuation of the asserted copyrights
Based on the calculation of relative consumer preferences in Dr. Shugan’s conjoint study,
Dr. Cockburn opined that the value of the Sun copyrights at issue was roughly half of the value of
the six patents Oracle was asserting at the time of Dr. Cockburn’s third report. Because the set of
asserted copyrights has not changed, the value of the copyrights should remain constant despite
Oracle’s decision to abandon four of the six patents discussed in Dr. Cockburn’s most recent
report. Accordingly, the only adjustment that needs to be made to the copyrights is to account for
the Court's adjustment to his valuation of the overall 2006 Sun intellectual property bundle, which
established a new cap of $560.5 million, as compared to the previous cap of $597.5 million.
Once that adjustment is made, Dr. Cockburn’s valuation of the asserted copyrights drops slightly,
from $34.7 million to $32.6 million. Leonard Decl. ¶ II.A.5.
III. CONCLUSION
For all these reasons, the soundest way to apply Dr. Cockburn’s damages methodology in
light of the Court’s Order and Oracle’s decision to limit its patent case to just the ’104 and ’205
patents is to adjust the calculations in Exhibit 35 of Dr. Cockburn’s most recent Report. Under
the methodology set forth in Exhibit 35, as modified by the Court’s Order, Dr. Cockburn can
testify to an unadjusted reasonable royalty of $18.5 million for the ’104 patent and $0.7 million
4
for the ’205 patent, an adjusted reasonable royalty of $4.7 million for the ‘104 and $0.2 million
for the ’205, and a total reasonable royalty of $32.6 million for the asserted copyrights.
Dated: March 19, 2012
KEKER & VAN NEST LLP
/s/ Robert A. Van Nest
By: ROBERT A. VAN NEST
Attorneys for Defendant
GOOGLE INC.
5
|
|
Authored by: hardmath on Tuesday, March 20 2012 @ 08:49 AM EDT |
Use the subject lien -> line to indicate nature of correction.
Thansk -> Thanks
---
Do the arithmetic or be doomed to talk nonsense. -- John McCarthy (1927-2011)[ Reply to This | # ]
|
|
Authored by: Anonymous on Tuesday, March 20 2012 @ 08:49 AM EDT |
... [ Reply to This | # ]
|
|
Authored by: hardmath on Tuesday, March 20 2012 @ 08:51 AM EDT |
So many News Picks, so little time...
They scroll down quickly, so please include a link to the
underlying article if starting a new thread.
---
Do the arithmetic or be doomed to talk nonsense. -- John McCarthy (1927-2011)[ Reply to This | # ]
|
|
Authored by: hardmath on Tuesday, March 20 2012 @ 08:53 AM EDT |
It's all good.
But Off Topic is better.
Please include clickies as needed.
---
Do the arithmetic or be doomed to talk nonsense. -- John McCarthy (1927-2011)[ Reply to This | # ]
|
- Off Topic - Authored by: red floyd on Tuesday, March 20 2012 @ 11:53 AM EDT
- MAYO COLLABORATIVE SERVICES v. PROMETHEUS LABORATORIES, INC. - Authored by: darrellb on Tuesday, March 20 2012 @ 11:57 AM EDT
- secret, for-pay laws - "While whales may make bigger bets, we draw the line at $273 million." - Authored by: Anonymous on Tuesday, March 20 2012 @ 12:10 PM EDT
- Linux Kernel 3.3 Released, will support Android Apps - Authored by: jheisey on Tuesday, March 20 2012 @ 12:19 PM EDT
- "Unpatentable" - Authored by: Anonymous on Tuesday, March 20 2012 @ 12:30 PM EDT
- "Unpatentable" - Authored by: Anonymous on Tuesday, March 20 2012 @ 12:35 PM EDT
- Maturity? - Authored by: Anonymous on Tuesday, March 20 2012 @ 12:40 PM EDT
- Add it - Authored by: Anonymous on Tuesday, March 20 2012 @ 01:05 PM EDT
- "Unpatentable" - Authored by: DannyB on Tuesday, March 20 2012 @ 02:28 PM EDT
|
Authored by: hardmath on Tuesday, March 20 2012 @ 08:57 AM EDT |
Please post transcripts from the Comes cache in HTML format
but Plain Text
mode.
It makes it easier for PJ to incorporate.
The
Groklaw Comes transcripts
project is here.
--- Do the arithmetic
or be doomed to talk nonsense. -- John McCarthy (1927-2011) [ Reply to This | # ]
|
|
Authored by: jez_f on Tuesday, March 20 2012 @ 09:23 AM EDT |
I am sorry if I am not following this right. So Oracle are saying that Google
owes them less than Google does?
Can Google, say 'OK then' (in legalize) and let the lower figure stick?
Or am I missing something important?[ Reply to This | # ]
|
|
Authored by: hardmath on Tuesday, March 20 2012 @ 09:25 AM EDT |
I found the Groklaw article confusing, and it derives from
the confusing discussion in Google's filing.
Currently the '205 patent is not "in-suit" due to USPTO's
rejection of it and Oracle's agreement to withdraw it "with
prejudice" provided the trial goes forward this spring.
So that means the '104 and '520 are the only two remaining
patents in suit. But Google (PDF and text versions) refers
several times to the "remaining patents in-suit" as '104 and
'205. This material then gets synopsized by the Groklaw
article's opening.
The confusion is confounded by the first (extensive) filing
by Oracle with respect to wanting to revise its arguments
and claims about the '205 infringement. A lot of words for
a moot question...
The discussion is also confused by '205 being one of the six
"originally patents in-suit" in the "top 22" patents, and
this not being so for '520. Some of Google's filing
correctly discusses the '205 patent in connection with how
to revise the Cockburn material on patent values for
presentation to the jury per the Court's recent ruling and
request.
---
Do the arithmetic or be doomed to talk nonsense. -- John McCarthy (1927-2011)[ Reply to This | # ]
|
|
Authored by: stegu on Tuesday, March 20 2012 @ 09:41 AM EDT |
It seems weird for Oracle to further reduce their
damages claims to be more conservative than the
Court's estimate, which is already less than
1% of the original "6 billions".
Are they trying for a formal win at all costs,
to maintain the illusion that there is any value
at all to software patents in general, and their
software patents in particular? If these two patents
are considered irrelevant, or worse invalid, by
the court, their entire Sun patent portfolio is
at risk of being considered junk by everyone. (It
should be, because it is, but that's another matter.)
However, if the damages awarded are even $1 million,
it is something to spin in future licensing negotiations:
"A single patent inadvertently infringed by Google
was worth $1M, so do you dare risk infringing an
unknown number of our 10,000 patents, or would you
perhaps like to sign a licensing agreement for a
mere $20M to make the problem go away?"
I hope Google sees the big picture here and takes the
patents down, guns blazing, instead of looking at the
short term gain from just accepting the currently
small, almost symbolic, damages and walking away.
$50M is cpmparable to pocket change to both parties
in the context of this large, long and staff-intensive
litigation, but this case needs to go to trial if we
are ever to put a stop to the endless patent troll
lawsuits.
[ Reply to This | # ]
|
|
Authored by: maroberts on Tuesday, March 20 2012 @ 10:56 AM EDT |
The judge issues an order saying that as the disputed amount is peanuts, then do
Oracle and Google *really* want an 8 week trial over this and can they please
settle?[ Reply to This | # ]
|
- Next up - Authored by: Anonymous on Tuesday, March 20 2012 @ 11:22 AM EDT
- No settlement!! - Authored by: Anonymous on Tuesday, March 20 2012 @ 12:06 PM EDT
|
Authored by: Ian Al on Tuesday, March 20 2012 @ 01:41 PM EDT |
Mitchell is completely wrong.
From Mitchell's report:
On an
Android system, the PackageManagerService is responsible for
installing
applications. The Java source code defining the PackageManagerService
class is
found in PackageManagerService.java. The performDexOptLI() method of
the
PackageManagerService class is invoked as part of the application
installation
process. This method in turn invokes the dexopt() method of the
Installer class.
The PackageManagerService class invokes Installer’s dexopt()
under other
circumstances as well. The Java source code defining the Installer
class is
found in Installer.java. The dexopt() method of the Installer class,
through
invocations of execute(), transaction(), writeCommand(),readReply(), and
other
methods, causes dexopt to run
and then returns dexopt’s success code to
its
caller.
Because PackageManagerService.java and Installer.java
are
written in the Java programming language, they are compiled to virtual
machine
instructions for execution, not native instructions. The dexopt
process
described in the code above and in the Google documentation is performed
during
the execution of the virtual machine instructions comprising the
Installer and
PackageManagerService classes. Thus Android satisfies the “at
runtime”
limitation as construed by the Court.
From the Dalvik
description:
dexopt
We want to verify and optimize all
of
the classes in the DEX file. The easiest and safest way to do this is to
load
all of the classes into the VM and run through them. Anything that fails to
load
is simply not verified or optimized. Unfortunately, this can cause
allocation of
some resources that are difficult to release (e.g. loading of
native shared
libraries), so we don't want to do it in the same virtual machine
that we're
running applications in.
The solution is to invoke a
program called
dexopt, which is really just a back door into the VM. It performs
an abbreviated
VM initialization, loads zero or more DEX files from the
bootstrap class path,
and then sets about verifying and optimizing whatever it
can from the target
DEX. On completion, the process exits, freeing all
resources.
It is
possible for multiple VMs to want the same DEX file
at the same time. File
locking is used to ensure that dexopt is only run
once.
Preparation
There are at least
three
different ways to create a "prepared" DEX file, sometimes known as "ODEX"
(for
Optimized DEX):
The VM does it "just in time". The
output
goes into a special dalvik-cache directory. This works on the desktop
and
engineering-only device builds where the permissions on the
dalvik-cache
directory are not restricted. On production devices, this is
not
allowed.
The system installer does it when an application is
first
added. It has the privileges required to write to dalvik-cache.
The
build system does it ahead of time. The relevant jar / apk files
are present,
but the classes.dex is stripped out. The optimized DEX is stored
next to the
original zip archive, not in dalvik-cache, and is part of the
system
image. So, dexopt is not in the Android devices. It can be
run
'just in time' on the apps development computer or it can be invoked on
the
development machine by PackageManagerService.java, but for Android phone
apps
the build system does it ahead of time before the app is installed on
the
'phone. The optimized DEX file is stored next to the original
zip
archive in the development machine, not in dalvik-cache, and is part of
the
system image.
--- Regards
Ian Al
Software Patents: It's the disclosed functions in the patent, stupid! [ Reply to This | # ]
|
|
Authored by: Ian Al on Wednesday, March 21 2012 @ 04:24 AM EDT |
On Groklaw and in the court we have established that use of the Java language is
free and open.
On Groklaw we have established that the only definition of the
Java language includes the Java APIs so that the APIs are free as
well.
However, Oracle have proved to their satisfaction that Dalvik is far less
efficient when it does not use the Oracle patents. (At the end I pour scorn on
the methods used by their top engineers and their expert reporter.)
In my post,
above, about Mitchell being totally wrong, I quote Oracle saying
that:On an Android system, the PackageManagerService is responsible
for installing applications. The Java source code defining the
PackageManagerService class is found in PackageManagerService.java. The
performDexOptLI() method of the PackageManagerService class is invoked as part
of the application installation process. This method in turn invokes the
dexopt() method of the Installer class. The PackageManagerService class invokes
Installer’s dexopt() under other circumstances as well. The Java source code
defining the Installer class is found in Installer.java. The dexopt() method of
the Installer class, through invocations of execute(), transaction(),
writeCommand(),readReply(), and other methods, causes dexopt to run and then
returns dexopt’s success code to its caller.
Because
PackageManagerService.java and Installer.java are written in the Java
programming language, they are compiled to virtual machine instructions for
execution, not native instructions. The dexopt process described in the code
above and in the Google documentation is performed during the execution of the
virtual machine instructions comprising the Installer and PackageManagerService
classes. Thus Android satisfies the “at runtime” limitation as construed by the
Court.
In other words, dexopt is run by something written in the
Java language and we see from the following Dalvik writeup that it works on Java
class files.The build system does it ahead of time. The relevant jar
/ apk files are present, but the classes.dex is stripped out. The optimized DEX
is stored next to the original zip archive, not in dalvik-cache, and is part of
the system image.
Oracle are saying that the Java language is
free, but that the essential optimisation of files compiled from the Java
language using the patents is not free. According to Oracle, dx infringes on the
'520 patent and dexopt infringes on the 104 and the 205 patent. As shown above,
dx and dexopt only work on the Java language files. Thus, if Oracle's
performance test is right, the Java language is worthless without the
patents.
Judge Alsup asked his question because he did not (or, perhaps
he did) know that Sun include the APIs as an integral part of the Java language.
His question is also relevant to the patents. Since the patents only work with
Java files, does not a licence to the Java language give an implicit licence to
the patents?
Oracle's top engineers 'proved' that the performance of
Dalvik was greatly reduced by disabling the JIT. Oracle's expert say that this
shows how important the infringing dx and dexopt is to Android. However, they
did not test applications written in Java that used dx and dexopt. The tests did
not prove anything related to the Java tools that allegedly infringed on the
patents./blockquote --- Regards
Ian Al
Software Patents: It's the disclosed functions in the patent, stupid! [ Reply to This | # ]
|
|
|
|
|