We have the transcript now from the March 7th hearing in
Oracle v. Google, the one about whether or not to allow Oracle to file a third expert report on damages by Iain Cockburn. They won. Or maybe I should say Oracle can't win for losing, because they were told if they went forward, they'd have to pay Google's expenses on the motion, and the report was totally useless anyway. We had two observers at that
hearing, so now you can compare their reports with the official court transcript [PDF]. If anyone can do it as text, I'd greatly appreciate it. I find it absolutely hilarious to read them trying to evaluate Oracle's "damages". At the time, it wasn't so funny, but now, if you did this as a play, I think we'd all fall down laughing. Except at the judge, who knows the difference between mathematics and arithmetic. I love this judge, the very Hon. William Alsup, who seriously ought to take up poker.
Jump To Comments
[Update 1,
Update 2,
Update 3]
Here's where I see he knows the difference:
So while I have you here, yesterday I sent out a request that you check my mathematics, arithmetic really, on the 10 percent line on Exhibit 34. Now, if we could only teach that to patent lawyers, so they'd comprehend that software patents defy reality. And if we act it out, this hearing, I want to play Mr. Norton for Oracle in this scene, if I can keep from guffawing:
MR. NORTON: And there are two calculations on slide 7, simply because it wasn't clear to us whether the Court wanted to use the average of the three curves that are provided on Exhibit 34 of Professor Cockburn's report, or just the mid range, which is PatVal. The two numbers are very similar, and they don't have a significant difference on the ultimate result.
But taking the formula from paragraph 414 of Professor Cockburn's report, the formula is described on the upper left-hand quadrant of slide 7 as the value of patents-in-suit plus the variable A times the value of patents-in-suit plus one half times the value of the patents-in-suit should add up to $597.5 million.
And then the formula for A, which is 1 minus percentage of the patent portfolio attributable to the patents-in-suit divided by the percentage of patent portfolio attributable to the patents-in-suit. And that is the formula in paragraph 414.
And using the PatVal curve, which is the curve on which Professor Cockburn relies, in that scenario when you solve for A you get 11.85.
And then plugging that into the equation on the upper left-hand side, that would tell you that the value of the patents-in-suit, using the assumptions in the Court's order of yesterday, would be $44.8 million. And applying Professor
Cockburn's formula from paragraph 414, the copyrights would be half of that, or $22.4 million.
THE COURT: All right. I don't know how -- I came up with 32, and you come up with 22.4 or 22.5. Is that true? Ah, the eternal question, when lawyers speak. Is that true? See what I mean? It's funny. I could act it out with a whiteboard, and slides, and a hat and a mustache. Here's more:
MR. NORTON: Well, no, we come up with 44.8 for the patents, and 22.4 for the copyrights.
THE COURT: I see. I see. All right. So you come up with 44.8. Somehow I came up with 32. And that's for five, correct?
MR. NORTON: That would be for five patents, applying the Court's assumption that those five patents are of equal value and are all in the top 57.
The Court asked that we be prepared to comment on the reasoning of that assumption, as well. I'm prepared to address --
THE COURT: Wait, wait. No. MR. NORTON: Thank you.
THE COURT: I want to make sure that 44.8 is not per
patent, but the grand total of all five. MR. NORTON: That is correct. That 44.8, under those
assumptions, would be the value of those five patents-in-suit.
THE COURT: All right.
Of course, Oracle ended up with zero, not a dime for those patents.
That's the funny part.
Update 2: I should add that most of the third Cockburn report was tossed out by the judge. What was stricken, should there ever have been a damages phase to the trial, was: -
the independent-significance approach to valuation;
- the econometric analysis;
- the conjoint analysis as used to determine market share; and
- the "upper bound" calculation in the group-and-value approach.
The jury found that Google didn't infringe either of Oracle's patents, as it turned out, so there never was a damages phase to the trial. But Oracle still has to pay for the report and for Google's expenses in handling it, including attending the hearing. All that PatVal stuff for absolutely nothing. I have a simple suggestion. If the Groklaw community is making fun of your case, maybe you would be wise to seriously consider if they are right. You'd probably save a lot of money.
Update 3: I'm working on the trial exhibits -- and may I say a huge thank you to everyone for your contributed work -- and I see #0618 is the proposed Joint Development and License Agreement between Sun and Google. It was never signed, but the judge in this hearing asks how long it was to last. Here's the truthful answer, from the agreement:
12.1 Term of Agreement. This Agreement shall have an initial term of three (3) years, and shall automatically renew for two (2) additional annual renewals, unless either Party provides notice of its intent not to renew the Agreement at least thirty (30) days prior to an annual renewal date (the "Term")....
12.5 Survival ofRights and Obligations Upon Termination. The provisions of Sections 1, 2, 5, 7-12, 13 (including Exhibit B), and 15, shall survive any expiration or termination of this Agreement and any provisions (or parts thereof) which by their express terms survive. Notwithstanding the foregoing, Sections 2.5, 5, 10 and 13.2 shall not survive if the Agreement is terminated pursuant to Section 12.4. Now you have that to compare to what the lawyers told the judge. Neither side seems to have it precisely right. Of course, as Mr. Van Nest says, the negotiations were ongoing, so the lawyers may have access to other materials. But it appears to have been a 5-year term. And paragraph 5 says, about Google paying Sun: "Such fees shall be fully earned when due and non-refundable."
Also, I have some time over the weekend to finish up the trial exhibits, so if you posted some of them and they are not in the article yet, would you be so good as to post them again here, so I don't miss them and don't duplicate effort? Thank you. Sorry that this got shoved to the side for a bit, but
that's how it is when so much happens at the same time. I still have to do the text of the JMOL motion [PDF] Oracle had the running, stomping, screeching gall to file with the court. The longer Mark stays on vacation, the deeper in the hole I find myself, so any help would be greatly appreciated. [Update: I'm done with the exhibits and have placed a link in all the articles from the trial. Thank you, guys. This is exactly the kind of project I could never have done on my own.]
Update: Thanks to bugstomper, we have it now as text:
******************
Pages 1 - 112
UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
BEFORE THE HONORABLE WILLIAM H. ALSUP
____________
ORACLE AMERICA, INC.,
Plaintiff,
VS.
GOOGLE, INC.,
Defendant.
____________
No. C 10-3561 WHA
San Francisco, California
Wednesday
March 7, 2012
__________________
TRANSCRIPT OF PROCEEDINGS
APPEARANCES:
For Plaintiff:
MORRISON & FOERSTER
[address]
BY: MICHAEL A. JACOBS, ESQUIRE
BOIES, SCHILLER & FLEXNER
[address]
BY: WILLIAM FRED NORTON, ESQUIRE
STEVEN C. HOLTZMAN, ESQUIRE
ORACLE AMERICA, INC.
[address]
BY: ANDREW C. TEMPKIN, CORPORATE COUNSEL
(Appearances continued on next page)
Reported By:
Katherine Powell Sullivan, CSR #5812, RPR, CRR
Official Reporter - U.S. District Court
[phone]
1
APPEARANCES (CONTINUED):
For Defendant:
KEKER & VAN NEST [address]
BY: ROBERT ADDY VAN NEST, ESQUIRE
DANIEL PURCELL, ESQUIRE
MICHAEL S. KWUN, ESQUIRE
CHRISTA MARTINE ANDERSON, ESQUIRE
KING & SPALDING LLP
[address]
BY: BRUCE W. BABER, ESQUIRE
GOOGLE, INC.
[address]
BY: RENNY HWANG, LITIGATION COUNSEL
For Dr. Kearl: FARELLA BRAUN & MARTEL LLP
[address]
BY: JOHN L. COOPER, ESQUIRE
2
P R O C E E D I N G S
MARCH 7, 2012 - 7:30 A.M.
THE CLERK: Calling civil action 10-3561, Oracle
America, Inc. versus Google, Inc.
Counsel, can you please state your appearances for
the record.
MR. JACOBS: Michael Jacobs, Morrison & Foerster, for
Oracle. Good morning, Your Honor.
MR. NORTON: Fred Norton of Boies, Schiller & Flexner
for Oracle. Good morning, Your Honor.
THE COURT: Good morning.
MR. HOLTZMAN: Steve Holtzman, Boies, Schiller &
Flexner, for Oracle American.
MR. TEMKIN: Good morning. Andrew Temkin with Oracle
America.
THE COURT: Okay.
MR. VAN NEST: Good morning, Your Honor. Bob
Van Nest for Google. I'm here with Dan Purcell, Michael Kwun,
Christa Anderson. And we're here, also, with Bruce Baber of
King & Spalding, and Renny Hwang from Google.
Thank you.
THE COURT: Good morning. Welcome.
MR. COOPER: Good morning, Your Honor. John Cooper
on behalf of Dr. Kearl. Dr. Kearl is in the courtroom, as
3
well.
THE COURT: Thank you both for coming.
Is Dr. Cockburn here?
MR. NORTON: Yes, he is, Your Honor.
THE COURT: All right. Thank you.
Okay. We're here for a hearing on whether to exclude
Dr. Cockburn's expert damages report. And I sent out a number
of questions. I hope you've had a chance to study them and be
ready on those.
Before we turn to other matters, I'd like to
understand a bigger-picture item, and that concerns this
question:
In the original 2006 offer of $100 million by Sun,
was that a one time paid-up license, or was that going to be
for a limited number of years? Who knows the answer to that
question?
MR. NORTON: I believe I know it.
THE COURT: You've got to come up here.
MR. NORTON: If I can confirm with my colleague for
one moment before I answer the Court.
(Pause)
MR. NORTON: Yes, Your Honor. It was anticipated to
be a 3-year term. And that would have been paid over the
3-year term.
THE COURT: So at the end of three years did -- once,
4
let's say Google, had started developing its Android platform,
and at the end of three years it had to come back and negotiate
a new license, is that the way it was going to work?
MR. NORTON: That is where the negotiations would
lead you. The actual draft agreement that was exchanged
between the parties, the last one was sent from Google to Sun
on April 19, 2006. And at that point there is not a term in
the license.
There were still negotiations between the parties
even then as to how long -- what the ultimate term of the
license would be. But the term that had been discussed between
the parties was a 3-year term, which would, of course, require
a renegotiation.
THE COURT: All right. And your name?
MR. PURCELL: Your Honor, that's not our
understanding.
THE COURT: What is your name?
MR. PURCELL: I'm sorry. My name is Dan Purcell.
I'm with Keker & Van Nest, representing Google.
THE COURT: The court reporter and I don't know who
the lawyers you are. You need to identify yourself for the
record.
MR. PURCELL: Fair enough.
THE COURT: All right. Say that again. What was
your point?
5
MR. PURCELL: Our understanding is that it was a
fully paid-up license. It was to be paid over --
THE COURT: Your understanding?
MR. PURCELL: Is that that was the initial --
THE COURT: Hand up the proof right now. Do you have
the proof --
MR. PURCELL: I do not have the initial draft offer.
Neither does, I believe, Oracle.
THE COURT: All right. Tell me what you base your
understanding on.
MR. PURCELL: The initial offer was a payment term
over three years, $20 million per year, total of $60 million
plus then some amount of downstream revenue sharing. That was
Sun's initial proposal. But I don't believe there was anything
in the proposal that said the license would go away after three
years.
THE COURT: All right. Thank you.
Well, I think this is a very important point. Let me
explain why. If it was a one time paid-up license, then the
idea that the plaintiff here could then seek damages for 2012,
seek damages for 2013, seek damages for 2014, on a royalty
basis doesn't square with what the $100 million offer was. If
that's the case. Now, I don't know that. I think it's a
matter of proof.
In other words, it's an apples and oranges thing. If
6
the hundred million dollars was meant to go forever, to allow
Google to use the technology indefinitely, then you can't turn
around and say, okay, this hundred million dollars was only for
the first three years, and then we get to tag them with
additional damages every single year thereafter.
On the other hand, if it really was just for three
years, then I guess the plaintiff can do that.
Now, that leads to a different question, which is,
would it have been savvy and smart for Google to go down a
three-year path, get stuck on the technology, get addicted to
the technology, only then to find out that the price would go
up at the end of three years? I don't know. That doesn't
sound smart to me.
Has anyone looked at this? Mr. Van Nest, has your
side looked into this?
MR. VAN NEST: Your Honor, I thought that's what the
experts were trying to come to. In other words, both experts,
as I understand it -- and admittedly I haven't been following
it as closely as Mr. Purcell -- but they're talking about what
would the royalty payment have been for this license?
And, as you know, Mr. Cockburn's been up in large
numbers coming down, and Dr. Leonard has been in, you know, low
eight figures. But what I understood was this was a three
year -- they had three years to pay it, but that it was a fully
paid-up license.
7
And what the experts were opining on is, what would
the parties have reached as a negotiation, not how many units
have you sold times a running rate.
So I kind of think the parties are taking that into
account in looking at the hypothetical negotiation. We've been
debating a starting point and we've been debating apportionment
and all that. But I really think what they are trying to get
to is, what's the value of the license on these asserted
claims?
THE COURT: Are you saying a fully paid-up license --
MR. VAN NEST: Yes.
THE COURT: -- or a year-to-year license?
MR. VAN NEST: Fully paid up.
THE COURT: Did you not hear Mr. -- Mr. Norton or Mr.
Holtzman -- who was it that spoke --
MR. NORTON: It was Mr. Norton.
THE COURT: Mr. Norton. Didn't you hear what he just
said?
MR. VAN NEST: I did.
THE COURT: He said that come 2012 they are going to
be asking for more damages, come 2013, more damages, and it's
not a fully paid-up license.
This is like the sky isn't dark, the sky is bright.
You two are completely apart on this. This is a major --
billions of dollars could be at stake on this question.
8
MR. VAN NEST: Again, what I'm coming back to is the
idea that what the experts have been charged with is
determining what the value of that license would have been for
the asserted claims now.
Obviously, they were negotiating, as Your Honor
knows, for a much bigger package. But I know that in
Dr. Leonard's case he's asserting a number that would
compensate Oracle for the value of these asserted claims on the
patent copyright side until the end of the -- until the end of
the day.
So we may be apart on it, and Your Honor is right to
point it out, but I know that our view is it was a fully
paid-up deal, payable over a 3-year term.
THE COURT: All right. Let me ask, Mr. Norton -- I
want us to be clear. And maybe I just misunderstand what you
say.
Are you really saying that the hundred-million-dollar
deal was only for a 3-year license, and after that it would be
renegotiated?
MR. NORTON: Yes. The hundred-million-dollar offer,
when it was made as a hundred-million-dollar offer, had a
3-year term attached to it. The parties continued to
negotiate.
THE COURT: 3-year term to pay the money, or did it
expressly say the license will only last for three years?
9
MR. NORTON: It was a 3-year term. Not just three
years to pay, but a 3-year term.
Now, at different points of the negotiation the
parties continued to talk about the term. So even after Sun
had proposed a hundred-million-dollar offer with a 3-year term,
there were, of course, further negotiations. But starting with
the hundred-million-dollar offer, that had a 3-year term.
At different points -- and there's testimony from
Mr. Gupta that Sun wanted a shorter term. There are other
documents that also -- evidence that would be offered at trial
that would show the length of the term as the parties
negotiated going forward.
But to respond in part to what Mr. Van Nest just
said, keep in mind that this is a hypothetical negotiation to
determine the reasonable royalty in this case. Google's
infringement, of course, is different and greater than what Sun
proposed to license.
And it would be inappropriate -- even if there had
been a term limit on this license, Sun would not have been
prepared to give up all of its rights to these patents and
copyrights forever, as Google now claims, if it were only being
compensated for a compatible license that --
THE COURT: Yes, but if you picked a
hundred-million-dollar offer as your starting point, which you
have done, and that was a fully paid-up license offer, then
10
you've got to adjust it for all of these variables. You can't
just assume away that it was not a paid-up license. That ought
to be -- maybe that's just a matter of proof.
All right. Has this point been adequately briefed in
the existing motion or not?
MR. NORTON: I don't believe it's been briefed at all
in the existing motion. The parties have not joined the issue.
Professor Cockburn has calculated the damages over
approximately a 3-year term, which brings us to the present.
And that issue has not been joined by the parties.
THE COURT: All right. Thank you.
Now, I have -- my next question for you is to
understand whether or not -- I understand that Dr. Cockburn has
dropped 2012 from his calculation. I assume that's without
prejudice to asking for 2012 damages in the future. But with
respect to the time period that he does ask for damages, is
that number lower or higher, or what, the same as the numbers
in report number two?
MR. NORTON: The numbers are -- I'm sorry. The
numbers are similar in some respects. And the reason why I
don't say "the same" is that Professor Cockburn, in his
February report, has several different approaches.
And if you use the lowest bound of his group and
value approach, the numbers are smaller. If you use the upper
bound of his group and value approach, the numbers are slightly
11
larger. But they are comparable.
We did prepare a demonstrative that shows --
THE COURT: I would be interested to see it. Thank
you for doing that.
MR. NORTON: I now have copies. I previously
provided it to Google, and have copies today.
THE COURT: Do you have a set -- did my law clerk get
a set? Okay. Good.
MR. NORTON: Yes, Your Honor.
THE COURT: Which chart do you want me to look at?
MR. NORTON: If you turn to tab 1 of the binder
that's been handed up to the Court, what we've done is in a
single demonstrative tried to answer a couple of questions the
Court had.
One was to walk through the steps that Professor
Cockburn employs to adjust the starting point; and, at the same
time, to compare those steps from his September report and
reply and his most recent report of February of this year.
And so you'll see -- we've done it for both patent
and copyright. Copyright is at the second tab, but I'll start
with patent.
And you can see that once Professor Cockburn applied
the adjustments for marking, and limitation to specific accused
devices, his damages as calculated in the fall, including his
October reply report, were $46.7 million for patents. Under
12
the group in lower bound, his damages for patents would be
$17.7 million.
Now, for this demonstrative we have included the '476
Patent -- although, we no longer assert it -- just so the Court
could actually see these numbers as an apples-to-apples
comparison. Obviously, dropping the '476 Patent would -- I
apologize. These numbers do include the elimination of the
'476 Patent.
On the group and value upper bound, it's
57.1 million. And the independent significance approach
damages are rather close to where they were in October, $43.7
million.
The steps -- I can walk through, if it would be
helpful for the Court to walk through the steps Professor
Cockburn has employed to make that adjustment.
THE COURT: No, this will do. Perhaps we'll get into
it later. But your chart is very helpful. Thank you.
MR. NORTON: And if you were to look at tab 2, the
steps are the same. Essentially, there's some patent-specific
steps that must be taken that do not apply to copyright, and so
this chart is a little bit smaller.
But, once again, you will see that the range of
damages is smaller for the group and value lower bound, and
slightly greater for the group and value upper bound, and very
similar for the independent significance approach as the
13
numbers were last fall.
THE COURT: Okay. Good. Thank you.
So while I have you here, yesterday I sent out a
request that you check my mathematics, arithmetic really, on
the 10 percent line on Exhibit 34. I was trying to see if I
had worked it out correctly. It would be fine if you say no, I
goofed it up. I just need to -- I'm trying to understand the
approach and, also, I'm particularly interested in that line
item.
So did you work through that 10 percent line item?
MR. NORTON: We did, Your Honor. And, again, we
prepared a demonstratives.
THE COURT: All right. Which one is it?
MR. NORTON: It is 6 and 7. One may be easier for
the Court than the other. Why don't I start with 6.
The answer to the Court's question is, we did have to
make an adjustment to the Court's math.
THE COURT: That's fine.
MR. NORTON: And we weren't quite able to figure out
where we parted ways, but --
THE COURT: That's okay. Just tell me how you did
it.
MR. NORTON: Sure. I think it's easier, then, if we
looked at slide 7.
THE COURT: Okay.
14
MR. NORTON: And there are two calculations on slide
7, simply because it wasn't clear to us whether the Court
wanted to use the average of the three curves that are provided
on Exhibit 34 of Professor Cockburn's report, or just the mid
range, which is PatVal. The two numbers are very similar, and
they don't have a significant difference on the ultimate
result.
But taking the formula from paragraph 414 of
Professor Cockburn's report, the formula is described on the
upper left-hand quadrant of slide 7 as the value of
patents-in-suit plus the variable A times the value of
patents-in-suit plus one half times the value of the
patents-in-suit should add up to $597.5 million.
And then the formula for A, which is 1 minus
percentage of the patent portfolio attributable to the
patents-in-suit divided by the percentage of patent portfolio
attributable to the patents-in-suit. And that is the formula
in paragraph 414.
And using the PatVal curve, which is the curve on
which Professor Cockburn relies, in that scenario when you
solve for A you get 11.85.
And then plugging that into the equation on the upper
left-hand side, that would tell you that the value of the
patents-in-suit, using the assumptions in the Court's order of
yesterday, would be $44.8 million. And applying Professor
15
Cockburn's formula from paragraph 414, the copyrights would be
half of that, or $22.4 million.
THE COURT: All right. I don't know how -- I came up
with 32, and you come up with 22.4 or 22.5. Is that true?
MR. NORTON: Well, no, we come up with 44.8 for the
patents, and 22.4 for the copyrights.
THE COURT: I see. I see. All right. So you come
up with 44.8. Somehow I came up with 32. And that's for five,
correct?
MR. NORTON: That would be for five patents, applying
the Court's assumption that those five patents are of equal
value and are all in the top 57.
The Court asked that we be prepared to comment on the
reasoning of that assumption, as well. I'm prepared to
address --
THE COURT: Wait, wait. No.
MR. NORTON: Thank you.
THE COURT: I want to make sure that 44.8 is not per
patent, but the grand total of all five.
MR. NORTON: That is correct. That 44.8, under those
assumptions, would be the value of those five patents-in-suit.
THE COURT: All right. So you wanted to say
something, and I interrupted you. Please go ahead.
MR. NORTON: So with respect to the reasoning of
that, the quarrel we would have with the logic of those
16
assumptions is that the analysis done by Dr. Reinhold and the
other JAVA engineers indicates that three of the asserted
patents are not merely in the top 57, not just in the top
10 percent, but are, in fact, in the top 3.9 percent.
And because this is not -- that the curve is not a
linear one -- in fact, it was very far from linear -- by
treating patents that are in the top 3.9 percent as though they
were in the top 10 percent, you significantly discount the
value of those patents.
At the same time, you actually probably overstate the
value of the '520 Patent because it probably doesn't belong in
the top 57. But we know from the evidence provided by those
engineers that three of the patents that are asserted are, in
fact, in the top 22.
And there is very good reason to think that they are
the top three. But even if you only say they are in the top
22, treating them as though they are only in the top 57
significantly reduces their value.
And so that's not an assumption that we think is
warranted by the evidence, and would, in fact, be inconsistent
with what the engineers determined based on their assessment of
all the patents.
I think that's the primary difference we have with
the logic of the approach.
THE COURT: What is the -- let me tell you what I
17
have a fundamental question about. You have three samples,
three sample portfolios from the history of the universe. Two
taken from Europe, and one from the United States. So it's a
sample of three. I want you to focus on the word "three." All
right.
No one doubts that these curves have a
disproportionate value at the far end. It's obvious from just
the ordinary application of the everyday 80/20 rule that that
would be the case. But that's not what is at issue here.
Your Dr. Cockburn -- am I saying that correct?
MR. NORTON: Cockburn.
THE COURT: -- Cockburn is trying to draw huge
conclusions from tiny data points.
You have three samples. Three. Not hundreds. You
have three samples. Three portfolios that were sampled. And
you looked at the curves on each one of them.
And when you're dealing with the top 20 percent,
where you're going to have a lot more data points, the numbers
come out fairly close. 94.4, 90.8, and 98.4. They vary
somewhat.
So you take three samples, and for that far
right-hand side of the distribution curve you're getting
somewhat similar results. When you get down to 1 percent, the
results vary widely. 52.6, 42.1, 78.4. Those aren't even
close.
18
In some studies you would treat those as outliers,
that top 1 percent, as very few data points under the curve
there. And some statisticians would treat those as outliers
and not even consider them, and yet that's the basis of your
study.
So let me come back to my main point. In statistics
when you do a sample of three -- which is what you've done
here. And how they were selected, I don't know. That's
another question. Let's assume they were randomly selected.
And you get widely varying results on three samples when we're
focusing on the 1 percent now. That's all I'm focusing on
because that's what you wanted to put before the jury. Three
samples that come out with widely varying results.
Then you ask the question, what are the odds that a
fourth, randomly selected portfolio taken somewhere in the
United States is going to be even in that range? It could be
lower than 52, when you get numbers of that -- varying to that
extent.
So we're not talking about the 20 percent part, which
seems pretty consistent. But as you get down to the 1 percent,
and the .5 percent, and even the 3 percent, even the 5 percent,
the results start to diverge. And I think statisticians would
ask the question, How many sample portfolios do you need?
You have three. That's all. Somehow he selected
three. Don't you need ten? Don't you need a hundred? Don't
19
you need to see a lot more results before you can draw
conclusions with any confidence as to that tiny tip of the
tail?
It's not even the tail. It's the tip of the tail.
It's the outliers that you're basing your whole case on. That
is my most fundamental -- and this is before you even get to
the way in which your own people started ranking these things.
I'm not even getting to that point. I'm asking just
on -- has -- what conclusions and confidence levels can you
place in the very tip-of-the-tail analysis that you have here?
I saw zero in this report. Zero where your Dr. Cockburn
analyzed that problem.
So I'm going to give you a chance to answer that
question.
MR. NORTON: There is a fair amount for me to respond
to there. I have some additional materials, if I can --
including the studies the Court requested.
THE COURT: Yes. Please, hand those up.
MR. NORTON: The Court ordered us to produce all
surveys that had patent value distribution curves. All
studies, I should say. And we've done that.
I had my back turned. Does the Court have a copy of
that binder?
THE COURT: You gave me this one right here. Is that
it?
20
MR. NORTON: Yes, Your Honor.
THE COURT: Yes, I do have it.
MR. NORTON: All right. Thank you.
So, again, a number of points to respond to.
Dr. Cockburn doesn't rely -- I think it would be more accurate
to say that Professor Cockburn relies upon the PatVal survey.
THE COURT: Then it's got a sample of one.
MR. NORTON: That is the one that he uses. Just as
Dr. Putnam, in the LG Display case, relied upon a single study
that had a single distribution curve.
But Professor Cockburn --
THE COURT: Is that a federal circuit case?
MR. NORTON: That is a District of Delaware case.
THE COURT: Well, that's just a district judge
talking. Did the district judge focus on the point that I'm
raising?
MR. NORTON: The adequacy of Professor Putnam's
analysis was challenged. It was a bench trial. The defendant
argued that his analysis was too speculative to support an
award of damages. And the Court not only allowed the
testimony, it awarded damages based on that testimony.
THE COURT: Did the judge raise the point about
relying on the tip of the tail, a tiny set of data points to
draw huge conclusions of the type you're trying to draw? Did
that point get addressed? I don't think so.
21
MR. NORTON: That specific point is not addressed.
THE COURT: All right. Well, I'm raising that point.
MR. NORTON: Right.
THE COURT: And that is, to me, a big problem with
your approach.
MR. NORTON: So let me --
THE COURT: I'm not quarreling with the 20 percent
part. I think -- because the numbers, you know, with three
samples are coming out the same pretty close. But they start
to diverge wildly when you get down to the small, tiny tip of
the tail thing. You know good and well if you did a fourth one
the number would be different, too.
MR. NORTON: So let me address that. That last
observation is probably a helpful place to start.
In the binder of demonstratives, if you turn to
number 5 --
THE COURT: All right. Wait. I'm sorry. The one
you just gave me, or the smaller one?
MR. NORTON: The small one. Wherever possible, I'll
try to use the small one.
THE COURT: Okay.
MR. NORTON: So the Court said, if we were to find a
fourth one the curve would be radically different. And what we
found, actually, is Professor Cockburn -- who, of course, is
here today -- has gathered together all of the studies that he
22
can locate that have surveys of patent value, direct evidence
of the economic value of patents, and report sufficient
information that one can actually see a curve in the study
itself.
And so these include the PatVal study, which is the
one that he relies upon for the purposes of this report, as
well as four others. And what you see -- and the PatVal curve
on this one is the blue one. And you see the curves do
intersect at various points --
THE COURT: I'm sorry. The blue one says here,
"Harhoff." PatVal is the red one in your chart.
MR. NORTON: I'm sorry. You may be looking at 5.
MR. PURCELL: It's 4. It's 4.
THE COURT: I'm sorry. Which one did you want me to
look at?
MR. NORTON: Slide 4, which should look like this
(indicating).
THE COURT: I am looking at the wrong one. I see.
All right. Okay.
MR. NORTON: So these are the available curves. And
what you see is there is a fairly tight range here.
Now, the Court asked about the confidence interval.
And Professor Cockburn is certainly better qualified than I am
to speak to this, but to answer the Court's question you
wouldn't talk about a confidence interval, normally, for a set
23
of studies where you only have five or six observations.
For a confidence interval you would need something
more like --
THE COURT: But the problem is, you have three sample
portfolios. That's it. You don't have hundreds of samples of
different portfolios.
But let me look at your -- your tab 4 for a minute.
Where is Barney on here?
MR. NORTON: So, Barney is not on here because --
THE COURT: That's because that number is 91 point --
see, you've selected the lines that help you make your point.
But look at your Exhibit 34.
MR. NORTON: Yes.
THE COURT: Look at Exhibit 34. Let's just take the
1 percent number. PatVal is 52.6.
That's not even the way it comes across on your
chart. I don't understand this chart.
Okay. 52.6. And then for Harhoff it's 42.1. And
then for Barney it's 78.4. But you conveniently left Barney
off.
MR. NORTON: The reason why Barney is not included on
slide 4 is that Barney is not a study -- is not a survey. And
Professor Cockburn will testify that the best evidence of skew
distributions comes from surveys. We have included not just
survey articles in the binder --
24
THE COURT: Why did he rely on it in the first place?
It looks like when it became inconvenient for him to rely on
it, he dropped it like a hot potato and then resorted to some
other surveys.
MR. NORTON: No, that's not correct, Your Honor.
What Professor Cockburn did in his report is he included the
Barney study and the Harhoff study as other examples of studies
that demonstrate a high degree of skew for patents.
And the Barney study happens to discuss U.S. patents,
and was helpful to demonstrate the -- anticipate the argument
that Google, of course, made, which was that, oh, well, you
can't assume that U.S. patents are skewed just because European
patents are skewed.
But Professor Cockburn never relied upon the Barney
study. He didn't do an average of the curves. He didn't
assimilate the curves.
What he said is he said the PatVal study is the
single most reliable study, and he explained why that's the
case at his deposition.
There are 23,000 observations, 23,000 data points in
the PatVal study, across several countries, based on extensive
survey data.
He testified about his familiarity with the study,
with the way the study was conducted, his experience with the
professional qualifications of the authors, and why he believed
25
that that single study was the best study to rely upon.
So -- and other studies that have similarly used
surveys to obtain data about the value of patents have very
similar curves.
But I would -- I have to strongly disagree with the
suggestion that he relied upon the Barney study and has done
something different. He has consistently said that the PatVal
study is the study that best predicts the distribution of
patent value. That is the only study that he used to actually
do the calculations.
THE COURT: I don't know what you've been reading.
Listen to this. Paragraph 405 of his report: "Based on three
recent studies of patent value" -- that's him talking, and then
at footnote 418 -- "I document distribution curves of that
value."
So he's relying on those three studies. He says it
right there in paragraph 405. And then he cites to the Barney
study as one of the three in that footnote.
MR. NORTON: Oh, absolutely. He cites that for the
proposition that there is a high degree of skew in patent
distribution. And he could cite dozens more articles for the
same proposition.
But if you then want to take the step of what curve
should he use to best predict the extent to which patent values
are, in fact, skewed, the only curve that he employed to do
26
that calculation anywhere in the report, anywhere in the
exhibits, at any point in his deposition testimony, is the
PatVal curve, because that is the one that he believes is the
best-conducted study that best predicts that value.
So the fact that he uses one study doesn't -- doesn't
in and of itself undermine our confidence.
THE COURT: He didn't use one study.
Listen to paragraph 406, "In each of these studies,
the top 1 percent of the patents collectively account for 42 to
78 percent of the value of all patents in that study as shown
on Exhibit 34."
So then you turn to your Exhibit 34, which I tried to
do, and read your -- and that's exactly where he gets the 42 to
78 percent. And Barney is 78 percent.
I don't see how you can possibly say he did not rely
on the Barney study.
MR. NORTON: Perhaps I'm not being clear. Professor
Cockburn relies on the Barney study and the wealth of
literature in this area. He specifically cites the Barney
study. He discusses the literature more generally, but he
specifically cites the Barney study as evidence of the degree
of skew in patent portfolio distributions.
But when it comes to the point where he has to do a
calculation, the only curve that informs his calculation, the
only curve that he uses to calculate damages in this case, is
27
PatVal.
THE COURT: So he has got a sample of one. A sample
of one. And you want to assume that any randomly -- any
randomly selected portfolio in the United States will conform
to that one at the very tip of the tail. That's what you -- it
comes down to; doesn't it?
MR. NORTON: Well, we rely on that survey. I don't
think --
THE COURT: What makes you think it's similar to
the -- there's nothing to indicate other than Sun had a patent
portfolio. Okay. But, so does a lot of other companies.
So, your proposition has to be that that one PatVal
study is identical to or very close to, at the tip of the tail,
any other randomly selected patent portfolio in the
United States.
If you don't accept that proposition, then I don't
see how you can apply that one study to any other selected
portfolio in the United States, which in this case happens to
be Sun, but in some other case would be somebody else.
MR. NORTON: So Professor Cockburn has previously
explained why he believes that the PatVal study does, in fact,
predict the distribution of value in the Sun portfolio. It is
not merely an assumption.
THE COURT: Where is that in the report?
MR. NORTON: That's in his deposition testimony.
28
THE COURT: In his deposition. Where is it in the
report? It doesn't count if it's in the deposition, except for
impeachment. The report is the place where he's supposed to
lay it out. Where is it in the report?
MR. NORTON: In the report he explains that he has
used the PatVal survey. In the report he does not say
specifically why he has concluded that the PatVal survey is the
best one.
THE COURT: He didn't. So his analysis in the report
has to come down to any randomly selected portfolio anywhere in
the United States is going to conform to the PatVal study.
Conceivably, the federal circuit could buy that.
But, to me, that is a remarkable proposition unless you're
saying it's going to, in general, conform. That's not so
remarkable.
But when you say the top half percent of the top
1 percent is going to conform, that is, to me, a very high
confidence that I don't see where you do any confidence level
assessment, period. There's none in there.
MR. NORTON: Let me try to address two of the points
that Your Honor just made. Three, in fact.
The first is that the justification for the reliance
on the survey must be in the report. Respectfully, I don't
believe that that is the law.
THE COURT: Well, tell me the law that says Rule 26
29
says every single opinion and every single justification has
got to be in that report. You can't fill in the blanks at the
deposition.
MR. NORTON: It would be an insurmountable burden of
disclosure if the reasons why a method are scientifically valid
had to be disclosed in the report itself. One would not demand
that a ballistics expert justify the gravitational constant.
That's an easy one.
Nonetheless, it cannot be the case that the reasons
why the methodology is sound have to be disclosed in the report
itself. And, in fact, I would be unable to do so right now,
but one can cite any number of Daubert opinions at which part
of the analysis at the Daubert hearing is, Well, what is the
literature? What do other people say? How --
THE COURT: I read this literature. Some of it,
anyway. I agree with you to a point. I agree that the curve
is going to show that at least the top 20 percent will have a
very disproportionate percentage of the value. So let's
concede that right off the top.
That is -- if you took the example that I sent out
and asked you to do, the top 10 percent even, you've come up
with some lower numbers than you are asking for. But the top
10 percent will have a disproportionate value.
And I think your studies -- your three studies come
out pretty close numbers on that. But once you drop below
30
10 percent, the numbers begin to wildly diverge in the three
studies that are in the report and which, contrary to your
representation to me, he did say he relied in these paragraphs
I quoted to you.
And when you have that kind of divergence and only
three samples, then you have to ask the question, if we did a
fourth sample what are the odds it's even going to fall in that
range?
To my mind, it cries out for more samples. And each
one of these patent portfolios is a sample. And you've only
got three. And you're talking about the very tip of the tail.
Not the 20 percent. The half percent of the 1 percent where
there are almost no data points. And many statisticians would
call those outliers, and yet you want to build a case on the
outliers. That's the problem.
All right. I'm going to give you a couple more
minutes. You're not convincing me very much. Maybe I'm just
missing.
Listen. What is the point of a Daubert hearing?
That your guy can just blow off -- just make an assumption that
one study is going to equal every single other patent portfolio
in the United States, and somehow that is a scientific
principle? Where do you find that in the literature? It's not
in the literature. All that is in the literature is the basic
proposition that it's going to be disproportionate. It is
31
not -- there's nowhere in the literature that it says that the
tip of the tail is going to be the same in every case.
Maybe you can find it. But I don't see it there.
MR. NORTON: We do not assert, nor do I believe we
need to assert, that the tip of the tail would be identical in
every case.
What we assert and what I think is sufficient under
Lucent -- which says that there's a degree of uncertainty and
approximation in this analysis -- and what's sufficient under
Daubert in Kumho Tire is that the study gives us enough
information that it allows an approximation of the
apportionment.
Now --
THE COURT: A reasonable approximation.
MR. NORTON: Absolutely, a reasonable approximation.
THE COURT: A reasonable approximation.
MR. NORTON: So when the Court says if we had another
study, well, in response to the Court's question we did
identify additional studies. And those studies are on chart 4.
And those studies show that, in fact, there is a fairly tight
range for the additional studies even --
THE COURT: After you conveniently leave off the
one -- you know, I don't know -- I can't imagine you thought I
was dumb enough not to realize you had left Barney off your
chart. You know, your new chart. I can't imagine that you
32
thought I wouldn't pick up on that.
MR. NORTON: Your Honor, I fully expected Your Honor
to pick up on it. I am prepared to address it, as is Professor
Cockburn.
Had we included the Barney study, it would simply
suggest the damages should be higher. And I would expect there
would be an argument from Google counsel that that upper bound
outlier study should be disregarded.
But the point is, is the PatVal study a good
predictor of the distribution of patents even at the high end
of the range? And if you look at the other studies which do
calculate these values -- they do calculate them, if you don't
limit yourself to the PatVal study but look at other studies,
you're going to get similar results.
Now, Google is free to come in and say, well, there
are other studies, and those other studies suggest a range, and
how can Professor Cockburn or the jury be confident that this
patent portfolio really looks like the others?
But it's going to be a dispute within that range.
And courts under Daubert have recognized that a certain amount
of uncertainty in the studies, variation in the inputs, is not
a reason to strike the analysis. It goes to its weight.
Here, there is an argument to be made -- and Google
is certainly prepared to make it -- that the analysis suggests
a range of results. But the fact that the analysis suggests a
33
range of results doesn't make it unsound.
There is a reasonable approximation here, which is --
leads to damages that are a lower bound and upper bound that
are within multiples of each other. Okay. And those are
ranges that experts often present to juries. The damages could
be anywhere between X and 3X. The fact there is a range
doesn't, in fact, make the methodology itself unsound.
There is a case, it is -- out of the District of
Colorado, Cook vs. Rockwell International Corporation, 580
F.Supp 2d 1071. And that really involved a class action trial
where the plaintiff class asserted they had been exposed to
plutonium from a nuclear plant. And the Court allowed the
plaintiffs' expert to testify at trial about -- it was a dose
expert on the effect of radiation doses.
And there the challenge made by the plaintiffs, like
the argument that's been made by Google here, is that the range
implied by all the different studies about what is the safe
level of exposure to plutonium, the ranges were too big.
And what the Court said is, "It is undisputed that
risk assessors commonly deal with such uncertainties in their
analyses, and that broad, estimated range of exposure, dose,
and risk may result from that. The existence of such
uncertainties and the consequences that flow from them in
estimating plutonium exposure, dose, and health risk may affect
the weight to be accorded to Dr. Fogel's testimony, but do not
34
provide a basis for finding his work and conclusions unreliable
within the meaning of Rule 702."
The standard here is not one of perfection. And the
question is, even, you know, where the Court has correctly
identified that the top of the range is where there is the most
sensitivity, but it is still possible to make a reasonable
approximation based on those studies.
And it's possible for the adversary to come in and
say that reasonable approximation has to be discounted because
there are other studies that suggest the numbers should be
lower, and those studies are better.
They are better for some reason that an expert can
point to because, unlike the PatVal study, they don't have
23,000 observations, they have 50,000, 60,000. Whatever
reason, they have to say that there is a better study that
better predicts patent value.
But that's a cross-examination question. That's a
jury question, because we have a narrow enough range here that
the jury can make a determination which is the better fit.
THE COURT: All right. I need to let the other side
have their say on this point. So I'll let you make one more
point, and then we'll hear from the other side.
MR. NORTON: It is not simply Dr. Cockburn coming in
and saying that this study is the best study. He's pointing to
the attributes of the study itself, the attributes --
35
THE COURT: Here's one of the problems I've got now.
It's like you're trying to bring in a brand-new report.
Exhibit 34 is what he said he relied on before. There were
three studies, including Barney. All right.
Is it fair now to say, okay, we're going to junk that
and we're not going to rely on Barney; we are throwing that out
and now we've got these other studies?
Yes, it's true that I asked to see the other studies.
I wanted to see what the literature requires here or suggests
here. But in fairness to the other side, don't we have to go
back to square one now? Or do we toss out this report as --
you're changing the report.
MR. NORTON: Absolutely not, Your Honor. Let's say
that this were a case where the issue had to do with exposure
to some carcinogen. And the expert came in and said, I have
one study. I have one study that I rely upon that was
conducted by the National Institutes of Health. It was done by
the best people. It went on over a number of years.
THE COURT: And concluded what?
MR. NORTON: And it concluded that there is a
correlation between exposure to this carcinogen and cancer.
And I, as a result, have a -- can conclude to a reasonable
certainty that, in fact, the plaintiff's exposure to this
carcinogen could have been the cause of their ultimate cancer.
One study. It is a gold standard study. It is the best study
36
out there. I don't know of any better. All the best people
worked on it. This is the one I rely on.
And the other side comes in and says, Well, you know,
one study. How can you rely on one study?
And so then the Daubert challenge is, is it -- is it
reasonable as a matter of science or as technical knowledge, is
it reasonable to draw those conclusions from that single study?
Well, what reasons are there to assume that it is
not? Are there other studies that come to different results?
Is there a methodological flaw in the underlying study on which
the expert relied? Would a cancer researcher not rely upon
this particular study in trying to assess carcinogenic effects?
Those are the questions that would be asked. But the
fact there is a single study doesn't mean that the study is a
bad one, and doesn't mean that you can't draw conclusions from
it.
So, Your Honor, we're not asking to put in these
other surveys. We are asking that the Court recognize that
when one looks to other surveys they tell us that the PatVal
survey is an awfully good one.
THE COURT: Let's hear from the other side.
I would like for -- and your name is what?
MR. PURCELL: Dan Purcell, Your Honor.
THE COURT: Mr. Purcell, I want you to start by
addressing the chart counsel handed up, the chart that leaves
37
off Barney.
I understand it leaves off Barney. Counsel's point
is that these other studies have a similar curve, and the range
is not as wildly varying as I had suggested earlier.
So my question to you is, what's your answer to this
tab number 4, that counsel has handed up?
MR. PURCELL: Our answer to that is pretty much what
was in our brief, which is that these studies are a bad fit for
the portfolio that we actually have here.
Your Honor was talking about randomly-selected
portfolios and how would I know that the PatVal study would map
to a randomly selected portfolio.
The Sun portfolio here is not randomly selected.
Perhaps, if Oracle had looked at the entire 14,000-some patents
that Sun had at the time Oracle acquired Sun, that would be
randomly selected.
What they did, instead, was winnow those 14,000 down
to 1300 that might be relevant.
THE COURT: I don't want to get into, yet, the way in
which they ranked the Sun portfolio.
MR. PURCELL: I'm not talking about ranking, Your
Honor. I'm just talking about the winnowing down of a randomly
selected group of patents to a narrowly focused group of
patents that are specifically relevant to a technology area.
And there actually is one study in the binder that
38
Your Honor has, the larger binder. It's at tab 11. It's the
Schankerman study. And I have --
THE COURT: I don't have any tab 11.
MR. PURCELL: The larger binder, the thicker binder
with the studies in it.
THE COURT: This?
MR. PURCELL: Correct. And I actually have a loose
copy of it here, if that would be easier, along with --
THE COURT: No, I have it here.
MR. PURCELL: -- our demonstratives.
THE COURT: I want to make sure I understand what you
just said, though.
I don't think you're coming to grips -- or maybe I
just don't understand it -- with counsel's point. I'm going to
hold up this chart. You see he's got this chart.
MR. PURCELL: I do.
THE COURT: And the colored lines, they vary somewhat
but they vary maybe 20 percent, not 50 percent.
And, oddly enough, when they get down to the tiniest
numbers, the variation is less than when it's up higher, for
some strange reason.
But down at the 1 percent level the variation is 52,
42, 42, 44, 48. So what they are suggesting is, all right,
that's pretty good proof that if we looked at the Sun portfolio
that the top 1 percent of the patents would be in that range of
39
42 to 52 percent of the overall value.
Now, I don't hear you -- what is so special about the
Sun portfolio that would distinguish it from these five
portfolios?
MR. PURCELL: The answer to that is that they are
looking at the wrong portfolio. They are not looking at the
Sun portfolio. They are not looking at all of the patents that
Sun had, which would be equivalent to this randomly selected
group of patents that were studied in these studies.
What they've done is specifically selected a
subcategory of patents that are relevant to this technology
area and that are presumptively more valuable to Google during
the course of a hypothetical negotiation.
THE COURT: So I see your point. So Sun actually had
more than 597 patents.
MR. PURCELL: Correct.
THE COURT: How many did it have?
MR. PURCELL: In the -- the number that was reported
at the time of the Oracle acquisition of Sun was around 14,000
total patents.
THE COURT: So what you're saying is if we looked at
the 14,000 -- I don't know if you're conceding it or just
saying, even if the 14,000 portfolio would be in the range of
this multi-color chart, we're not dealing with 14,000, we're
dealing with 597 that were specially selected to enable the
40
smartphone --
MR. PURCELL: Right, correct.
THE COURT: And so, therefore, -- doesn't that
mean -- doesn't that sort of indicate that all of the 597 would
be more valuable? Or would they be less valuable? What is
the -- what effect would that self-selection have on the
analysis?
MR. PURCELL: Well, what they've done, what Google --
sorry, Oracle has done by the methodology Dr. Cockburn has
used, they've got rid of all of the chaff. They have got rid
of all the patents that are the least valuable in the context
of a hypothetical negotiation. So you're starting with a pool
of patents that are presumptively at least of some value to
Google. And then they winnowed those down and selected the
ones that are the most valuable.
So what they've done, essentially, is they've gotten
rid of the patents that are in the 14,000 pool, that are of the
least value. And so you're starting with a pool of patents
that are all presumptively more valuable, which is inflated in
number.
THE COURT: So you're saying that of 597, the least
of the 597 are of greater value than the least of the 14,000?
MR. PURCELL: Presumably, yeah.
I think it's actually 569, Your Honor, that are in
the narrowed portfolio, are presumably already in the top
41
10 percent of the 14,000, at least in the context of this
hypothetical negotiation.
THE COURT: Therefore, the value of the hundred
million should be more evenly distributed over the 569 than if
you were distributing it over -- because the hundred million
dealt with the 569, not with 14,000.
MR. PURCELL: Correct. This is a -- not an argument
about the results of these studies. I mean, the results of
these studies are accurately captured on the chart. We're not
quarreling with those. What we're saying is these studies are
a bad fit for the portfolio at issue here.
And the one point I wanted to make has to do with the
Schankerman study, which we passed up a copy to Your Honor.
This is the one that's at tab 11. And the demonstrative slides
that we also handed you, slides 2 and 3 in that slim packet,
talk about Schankerman.
What Schankerman did alone, I believe, among the
studies in that binder was actually do what we suggested ought
to be done in our brief, which is look at a specific technology
area and see if the skew is different or the same in that
technology area as it is across the entire spectrum of patents
generally. A better faucet. A better mousetrap.
And what Schankerman did -- and this is slide 3 of
the slides that we handed up -- was look at how the skew works
in different industries. And Schankerman looked at four:
42
electronics -- which is the closest thing we have here; this is
an electronics product, the smartphone -- pharmaceuticals,
chemicals, and mechanical. And in all those industries there
was skew.
Again, we're not doubting the existence of skew.
Obviously, a lot of patents are really not very valuable at
all, and a small number are very valuable. But the range is
different and lower in these industries than any of the studies
that are in the binders.
And what Schankerman found is in the electronics
industry the top 1 percent of patents only account for about
24 percent of the value.
THE COURT: Where is that in your report?
MR. PURCELL: This is the slide 3 of the
demonstratives. And it's Schankerman at 95 and 96.
THE COURT: Well, let me -- I'm open to Schankerman
now.
MR. VAN NEST: We handed up a small set of slides,
Your Honor. They look like this. I'm sorry --
THE COURT: Oh.
MR. PURCELL: Right. So there's a summary chart at
slide 3. The third page of our demonstratives summarize it.
But it's at pages 95 and 96 of the Schankerman article.
And so what we see from Schankerman is that in the
electronics field 24 percent of the value is in the top
43
1 percent, 55 in the top 5 percent. Which is a significantly
lower value than in the studies that Dr. Cockburn used, which,
again, don't attempt to focus on the technology area at issue
here.
And that was the beef that we raised in our brief,
which was, why did you look at a random sampling of patents
across technology areas, not a single company portfolio? Why
didn't you focus your inquiry on what exactly was going on
here?
Schankerman tried to do that and ended up at a
significantly lower number.
THE COURT: All right. So what other issues do you
want to raise before you get to the way in which the company
ranked the patents, and just based upon the curves, the tip of
the tail and all of that? What other issues do you want to
raise?
MR. PURCELL: Your Honor, I don't have anything
further to say about the studies.
THE COURT: Mr. Norton, what is your answer to
Schankerman and the point that counsel just made about the
14,000 versus 569?
MR. NORTON: So let me start with the 14,000. First,
the number is misleading. It's 14,000 worldwide patents, which
counts every patent issued by a separate country as a unique
patent. So 14,000 --
44
THE COURT: How many were in the USA?
MR. NORTON: I did not know the answer to how many
U.S.-issued patents Sun had in 2006.
THE COURT: Was it more than 569?
MR. NORTON: Absolutely.
So, then, Google counsel's argument was that the 569
are disproportionately valuable.
There is no basis for that assertion, which is
contrary to the testimony of the JAVA engineers.
What the JAVA engineers testified they did was they
had -- asked for a word search for every patent that had the
word "JAVA" in it, "bite code" in it, or had been identified as
an inventor, Mr. Gosling or Mr. Fresko, and limited those to
the patents that were issued prior to June 2006.
Once they had that set of patents, they said, well,
which ones are smartphone patents? Not, which ones are
valuable patents? Just, which ones are smartphone patents? So
it's a portfolio that's limited to a technology area, but it is
limited to more valuable patents.
And Google's assertion, which they have made
repeatedly in the briefs, is somehow these were selected for
value to Google is just false. There is no basis for it
whatsoever.
And that is their entire argument for why this
particular portfolio shouldn't look like the distributions we
45
see in the studies.
So with respect to the Schankerman study, the
demonstrative Google handed up you will see you have
electronics, pharmaceuticals, chemical, and mechanical. And
you will see electronics actually have a higher degree of skew
in the Schankerman study than other industries.
And that's important because what it suggests is that
if we had used, instead of the PatVal study, a study that
provided a curve for the electronics industry alone -- because
PatVal has lots of different kinds of patents -- that the
PatVal study would suggest higher results.
And the reason for that is that there is a
significant methodological difference between what
Mr. Schankerman did -- I'm sure it's Dr. Schankerman -- what
Dr. Schankerman did and what the authors of the PatVal study
did.
And the difference is that the Schankerman study is
based on renewals. And, as Dr. Schankerman explains in the
study itself -- and do we have a copy of the binders? Thank
you.
The challenge of the renewal survey -- and Professor
Cockburn is more than prepared to explain the methodologies of
these different surveys. But the renewal surveys, what they do
not do is break down in a granular fashion the patents at the
top of the range.
46
So what they tend to do, and what I believe
Dr. Schankerman's study does, is it says, well, how many
patents are above X?
And if you don't make X big enough, then it doesn't
allow you to draw very fine conclusions about the value of the
patents in the top 1 percent or 5 percent. It tends to -- all
of the renewal surveys, renewal studies, will tend to flatten
the curve at the top end of the range.
And they do so because there is a systematic bias in
every single one of those surveys, because rather than to
continue to ask, how much is your most valuable patent worth,
they say, how many patents do you have that are worth more than
X? Because that's what they're interested in knowing.
That's fine for their purposes. But what they don't
do is go all the way up to figure out, well, where are the most
valuable patents?
And so the Schankerman study, as one would expect,
has a flatter distribution curve than the surveys because
surveys test the actual value of the patents that are being
measured.
But what's interesting about Schankerman is that
Schankerman tells us that applying that methodology
consistently across industries, we see a higher degree of skew
in the electronics industry than in pharmaceuticals, chemicals,
and mechanical industries or mechanical patents.
47
And so what that tells us is that if you correct
Dr. Schankerman's methodology, and did it as a survey, one
would expect that electronics patents would be more skewed.
And to the extent that electronics in this context
includes software patents or smartphone patents -- as it likely
would -- I would suggest that the PatVal study is probably a
conservative approach.
I think I have responded to Mr. Purcell's points.
THE COURT: All right. Thank you.
At this point we are going to change the subject and
go to the next step of the analysis, which is the ranking by
the in-house engineers.
So let's hear from Google on what your arguments are
against that and why it would be so unreliable. And keep in
mind that I'll let your side use in-house people. And you said
that was fine, and I agreed with you. But now it seems like
whenever the other side does it you're somehow offended. So
you've got to explain that part.
MR. PURCELL: Let me clarify that, Your Honor.
We are not saying that it is improper for Oracle to
use in-house people. We recognize that that's a bias issue for
cross. We're content to leave that for trial.
The only point we were making was the point, really,
that I just made, which was that what Oracle did was select
from a much larger group of patents a group of patents that are
48
presumptively more valuable to Google because they relate to
the technology Google was trying to commercialize.
Google is not building a car here. Google is not
building, you know, a sewer system. Google is building a
smartphone platform. So when Oracle selects smartphone-related
patents, they are necessarily selecting a group of patents more
valuable to Google's purpose.
And Oracle used engineers not only to do that, just
inherently, they used engineers that had a background in
preparing the case for litigation. And they admitted at
deposition that they weren't able to put away that experience
and they, in fact, did rely on that.
So it goes to the selection of the portfolio as a
Daubert issue. Beyond that, it's a point for cross at trial.
So I don't know that there's a whole lot to add to the argument
that I just made.
I do want to point out one other thing, though.
Mr. Norton was trying to draw a distinction between surveys and
citation -- sorry, renewal studies.
The Barney study that Dr. Cockburn relied on was a
renewal study, and it had a much, much higher -- higher-sloping
curve that had the top 1 percent in the 90 range. But that's
all I have to say, really, about the engineers.
THE COURT: Sounds like -- it does sound like that is
just a cross-examination point.
49
MR. PURCELL: As I said, Your Honor, it is,
primarily.
The only relevance it has to Daubert is in the
selection of the portfolio and whether or not the studies
Dr. Cockburn used are a proper fit for the portfolio that's at
issue here, which isn't a random sampling, which was prior
vetted by Oracle engineers for value to Google in the context
of the hypothetical negotiation.
THE COURT: All right. I'm going to take that as a
concession that your point is not very strong, and it's going
to be for cross-examination that the in-house people did the
ranking.
What's your next objection to the report?
MR. PURCELL: Really, Your Honor, our objections --
the main one, which relates to both the independent
significance approach and the group and value approach in
Dr. Cockburn's report, is indeterminacy.
This is something we've been struggling with from the
beginning of the damages proceedings in this case. The
independent significance approach, he says 25 percent of the
portfolio at least with respect to the patents, 12 and a half
at least with respect to the copyrights.
At deposition, when he was asked what does "at least"
mean he said, well, the patents could be 50 percent, they could
be more; the copyrights could be a hundred percent of the value
50
because it could have been the case that without these APIs
Google could not have commercialized Android at all.
So we don't want to be in a situation where
Dr. Cockburn has the freedom going to trial to say damages
could be 17 million; they could be 600 million; they could be
the entire value of the portfolio.
And that's the objection under the independent
significance approach, is that there's absolutely no limit to
it. And there's also absolutely no methodology. It's just:
Here, I looked at all of the evidence in the case. I
synthesized it, and the number that came out of my brain was
25 percent.
That's not replicable by anyone else.
THE COURT: In his report -- let's be clear what you
extracted in the deposition on cross-examination versus what's
in the report.
I want to say to you, at trial you experts will
only -- you have to stick to chapter and verse of what's in the
report. You cannot even get into what's in the rebuttal reply
report. You must limit yourself to the conclusions in the
original report.
And I'll keep it up here, and if somebody says,
"Beyond the scope of the report" I sustain that objection. I
turn to you and say, Where are those words in the report? And
if those words are not in the report, or something very close
51
to it, it is stricken.
Now, the problem, though, is on cross-examination the
lawyers will often go way beyond what's in the report. They're
not -- the expert is not limited to what's in the report, just
like you're not limited to what's in the report on
cross-examination.
So if you want to get into things that he didn't
opine on in the report, that's your problem; isn't it?
MR. PURCELL: I understand the danger of opening the
door, and that would be our problem.
But what we are moving on is to try to limit him to
what's in the report at least for purposes -- if Your Honor
says he can offer --
THE COURT: Of course, he's going to be limited. In
the report he doesn't give -- does he give up -- does he say
"at least" and then give some number? I don't think so.
MR. PURCELL: He just says "at least."
THE COURT: Then that's what he'll say at trial. And
if he tries to go one inch beyond that and you object, I'll
sustain the objection because you cannot do that.
Now, the problem is, again, though, on
cross-examination if you open the door he can walk right
through it and say things that are not in the report.
MR. PURCELL: The problem with "at least" is that it
doesn't give any guidance to the jury. It gives the jury one
52
number, but then what does "at least" mean?
THE COURT: There are many situations where somebody
is permitted to say "at least." They can have a lot of
confidence that a number is at least X, and it could be higher.
That happens all the time. What's wrong with that?
In other words, they're trying to be conservative.
They say the number is at least X. And they have a lot of
confidence that that number can't be attacked. So that's what
they go with. But then on cross-examination they say, well, I
guess if you test me on this, it could be higher. It could be
2X.
But you're calling for a level of precision that I
don't think the law does require that. So long as he doesn't
go beyond what's in his report, then you have the tough choice
how you're going to cross-examine him.
Believe me, if you ask the right question or the
wrong question I'm going to let him say whatever he wants to
say in response.
MR. PURCELL: Fair enough, Your Honor.
The group and value approach, if I can move on to
that. It has to do -- again, we're talking about a very broad
range here of three times, three and a half times as much
damages at the high end versus the low end.
The problem that we see there is that the upper bound
of the report depends on an assumption that the three patents
53
that Dr. Cockburn identified -- or, rather, the Oracle
engineers identified as being in the top 22, that's the '104,
the '205, and the '720 patents, are, in fact, the three most
valuable patents in the entire Sun portfolio.
THE COURT: I'm sorry. Show me where this is in the
report so I can -- it's kind of ringing true but -- is that
paragraph 409?
MR. PURCELL: Your Honor, I'd have to look at the
report. Again, this is something that was further fleshed out
at deposition, I think.
THE COURT: I want to stick with what's in the report
first, and then maybe we get into the deposition.
MR. PURCELL: Well, actually, Your Honor, paragraph
409 makes my point, the point that I wanted to make, which is
that there is no data to distinguish among the top 22 patents.
And Dr. Cockburn concedes that. He said that the Oracle
engineers basically told him, no, we can't break it down any
further than that. All those 22 patents are roughly equally
valuable.
When we asked the engineers about it at their
depositions, they said: We have to do a lot of testing. We
haven't done it, and I'm not even sure you could do it.
And so the lower bound of the report assumes,
basically, that those 22 patents are equally valuable.
THE COURT: Where is the lower bound?
54
MR. PURCELL: The lower bound is the 17.7 million, I
believe.
THE COURT: I need to see the report.
MR. PURCELL: All right.
THE COURT: I'm sure it's here somewhere.
MR. PURCELL: I think the easier place to find it
probably is at the front of the report. And this would be on
page 3, paragraph 5.
So the group and value approach on the patents leads
to a range between 17.7 million and 57.1 million. So when I'm
talking about the lower bound, I'm talking about the
17.7 million.
THE COURT: I don't see on -- page 2?
MR. PURCELL: Page 3, Your Honor.
THE COURT: Page what?
MR. PURCELL: Page 3.
THE COURT: 3. All right.
MR. PURCELL: Sorry.
THE COURT: And then what paragraph do I look at?
MR. PURCELL: It's line 7 in the chart at the top of
page 3.
THE COURT: All right. Adjust downward to limit
damages to accused devices. 17.7 million to 57.1 million.
All right. I see that. Now, so what accounts for
that range?
55
MR. PURCELL: Well, I don't believe this is really
spelled out in the report, to be honest. And Oracle can
correct me if I'm wrong.
The explanation we got at the deposition is that the
lower bound would be an appropriate jury finding if the jury
concludes that the top 22 patents are roughly equally valuable.
The upper bound would be an appropriate result if the
jury concludes that the '104, '205, and '720 patents are the
three most valuable patents in the portfolio.
And our argument here is that there is no support for
the upper bound. The Oracle engineers have said, We can't
offer any foundational testimony suggesting that there's any
distinction in value among those patents. We just can't do it.
And Dr. Cockburn hasn't cited anything. There's
nothing in the report that would support that conclusion.
THE COURT: All right. Wait a minute. I've got my
own chart. I have -- I've got a different question. Number
22. There were 22 groups, then there were 22 top patents.
Those 22 22s have nothing to do with each other --
MR. PURCELL: Correct.
THE COURT: -- it's a coincidence.
MR. PURCELL: It's a coincidence.
THE COURT: This is a good trial point. You should
never have that kind of coincidence. It's going to confuse the
jury to no end. I spent hours trying to figure that out. You
56
should have 21 or 23. But to come out with 22, it just sounds
like it's the same. I kept saying, it doesn't sound like it's
the same, but why would it be 22? Do you see the point?
MR. PURCELL: I do.
THE COURT: All right.
MR. PURCELL: It's not our report.
(Laughter)
THE COURT: Then you've got this business of the 1s,
2s and 3s.
MR. PURCELL: Right. So the top 22, the way that
that was explained that those were selected, those are all of
the patents that had a 1 ranking in the three most valuable
technology groups.
So out of the 22 technology groups, you take the top
three. And then any patent in those three groups that had a 1
ranking is in that top range. And that added up to 22 total
patents.
THE COURT: But, wait a second. I thought the -- all
right. So the 22 top -- of the 22 top categories, there are
three.
MR. PURCELL: Right. The 22 total categories they
took the top three.
THE COURT: And then there were 22 patents that were
ranked number 1 within those top three categories?
MR. PURCELL: Correct. And this actually is in the
57
report, Exhibit 33.
THE COURT: All right. I got that part now, but --
so you're saying that the testimony is that those top three
categories could not be distinguished among or ranked, I guess
is the word, ranked by the engineers? Is that what you're
saying?
MR. PURCELL: Correct. Dr. --
THE COURT: Read that testimony to me.
MR. PURCELL: I would need to go --
THE COURT: I asked you to come here prepared to do
it. I wanted to hear the testimony without spin. So I --
because if that's true, that's a very good point.
You find it. I don't want to interrupt you while you
look for it.
MR. PURCELL: Okay. I have one cite for you. This
is at paragraph -- page 105 of the Reinhold deposition. This
is a direct quote. He said:
"A quantitative analysis that would somehow
rank all of these patents in linear order
from 1 to 569 is actually intellectually
infeasible."
I would need to get the transcripts to read other --
THE COURT: Where did he limit his answer to the 22?
MR. PURCELL: I will need to get the transcript, Your
Honor.
58
THE COURT: You said that they could not rank within
the 22. That's what I want to hear.
MR. PURCELL: Your Honor, we can pull the transcript.
I don't have it at the ready. This was admitted, though, both
by Dr. Cockburn and the Oracle engineers. The testimony is
unanimous on that.
THE COURT: Who is going to answer for Oracle on
this?
Mr. Norton, I just want to hear one sentence. Do you
agree with what counsel just said, that the top 22 could not be
ranked within those top 22?
MR. NORTON: I have one sentence. The engineers
could not distinguish between the top 22 patents based on their
technical significance based on what they would have known in
2006.
THE COURT: So if that's true, then why do you -- why
don't you just assume they are all of average value of the top
22?
MR. NORTON: Because their economic significance and
their technical significance as known to engineers are two
different propositions. And Professor Cockburn is an
economist. Dr. Reinhold is a computer scientist.
So what the engineers did is they said, if we were
sitting down with Google in spring of 2006, and we didn't know
yet what Google was going to do, we just knew that we had a lot
59
of patents, which patents do we think would be most important
to a smartphone platform along the lines of what Google said
Android would be in 2006. And so with that information, these
are 22 very important patents. And these would be ones that
Google would want to have available to it as the result of the
negotiation.
THE COURT: Isn't that the way it should be done,
though? We're talking about apportioning a number that was in
existence in 2006, to figure out how much of that was
attributable to these particular patents. And the future had
not yet unfolded.
MR. NORTON: Absolutely. That is the right way to do
it, and that is why they did it that way. So then the question
is, well, but is Professor Cockburn merely apportioning that
original 2006 negotiation, or is he doing a hypothetical
license negotiation under federal circuit law? And it's the
latter.
So he now needs to make an adjustment. And the
adjustment is disclosed in paragraph 410 of the report. And I
have to -- Mr. Purcell said that Professor Cockburn says there
is no data, which is a little unfair.
What Professor Cockburn says, in the absence of data
you can't make a distinction. Then he goes on in paragraph 410
and explains that we do know something. Right.
What we know is that Google chose to use the
60
technology that is comprised by these specific inventions. We
know this because Professor Cockburn is a damages expert and
he's going to take the stand at a point in which infringement
and validity are established.
Now, if this were a case in which Google said, We
closely examined the JAVA patents, and we tried to choose ones
that weren't necessarily all that great, but at least we
thought they were invalid or we thought we had to workaround,
you know, and it turns out we were wrong, but we didn't choose
these because they were so great, we chose them because we
thought we weren't infringing. But that is not the case.
The evidence in this case is that Google maintains
that they didn't look at these patents. They didn't consider
that. So what we know is Google designed Android in the way
that Google thought best fit its business needs, its technical
and business needs.
And when it made those decisions, what technology
should we incorporate in Android -- it could have infringed any
JAVA patent. They designed the Android however they want it.
They could have infringed any of those 22. But they chose
specific functionality to incorporate into Android that
infringes those three patents in the top 22, not any of the
others.
THE COURT: Doesn't that mean you're allocating more
of the 2006 bundle to the ones that they turned out later, in
61
the future, to infringe as opposed to -- to me, that sounds
like you're apportioning the original number, more of it, to
the ones that happened to be infringed.
MR. NORTON: I think the better way to describe it
is, no, there are two steps. The first is an apportionment
step, which is in the 2006 negotiation in which Google would
have obtained rights to a collection of patents -- and the
collection of patents was never specified in the parties'
negotiations. Right.
What they said was Google would get the rights to the
Sun technology that it ultimately incorporates into Android and
releases under an open source license. So they would have been
able to choose the stuff they wanted, but they wouldn't have
gotten everything. They just would have gotten the right to
open source the things they actually used.
But what we're trying to do is apportion the value --
Google is basically buying -- in 2006, they are buying a whole
bunch of options. They are buying options to use a lot of
different patents. And they are going to choose a smaller set.
So we are going to apportion the value of that bundle
across the patents and other components. And when we do that,
we see that these 22 patents and the patents in general are an
important part of that bundle.
But then we have to do something different, right?
Because Google didn't take that deal. Google said no to that
62
deal. Wanted to just go ahead on their own. And, instead,
they infringed, we would say, five patents.
So focusing on the three that are in the top 22, now
what Professor Cockburn has to do is say, if they had just been
negotiating for those -- let's focus on the three in the top
22 -- if they had just chosen those three to negotiate for, and
said these are the ones we want, well, then that tells us that
those three are the most important to Google.
We know that these are more important than the others
because when Google decided how to design Android free from
constraint -- we are just going to make Android the way we
want -- they chose to incorporate this technology.
And there needs to be an adjustment made in the
hypothetical negotiation to account for the fact that Google
selected these inventions and not others. And that's described
in paragraph 410 of the report.
In his deposition, Professor Cockburn used the term
"revealed preference" to describe the same phenomenon. But
that's what it is. It's Google's revealed preference. Their
actions tell us what they thought was important. And what they
thought was important is what is valuable.
So that is an additional piece of information that
the engineers lacked for purposes of this apportionment
analysis but that Professor Cockburn would be remiss to ignore.
And we also -- we know more, frankly. What the
63
engineers did was they said, well, based on our experience with
JAVA and the technology described by these patents, what kind
of benefit would we expect this patent to provide to a
smartphone platform?
And if they thought it would provide a very
substantial benefit based on their experience in order of
magnitude over leaving it out, then they gave it a 1.
But that's what their expectation was. For the
asserted patents, we know -- it's not guesswork. We now know,
and Google would have known at the time of the hypothetical
negotiation, that these particular patents are, in fact, very
valuable to a smartphone platform, because at the time they
commenced to infringe it would have been possible for them to
conduct the same kinds of studies that JAVA engineers have
conducted in this case, which show enormous benefits from using
these patents.
So we know that these patents are not just patents
that --
THE COURT: You said that the engineers could not --
thought all 22 patents were important, and could not
distinguish among them just based on the engineering, right?
MR. NORTON: I did say that, and that is correct.
THE COURT: So if we put ourself back in 2006, why
would the Google engineers have been able to be more precise
than your own engineers are now?
64
MR. NORTON: Well, because it's partly the artifice
of the exercise and partly the limitations on what Sun/Oracle
engineers know versus what Google engineers know.
So the exercise the engineers did was, if we were
looking at this in 2006, which patents would you expect to be
valuable based on the product requirements document and the
other information Google gave Sun back at that time?
So not knowing as much as Google, in fact, knew.
Because -- and not even knowing as much what Professor Cockburn
knows because we have a protective order in this case and I
can't show Sun engineers, those JAVA engineers, all the
internal documents from Google which explain precisely what
Google was going to do.
THE COURT: But do you have 2006 e-mails and
documents that reveal that these three patents were going to be
the ones that they were likely to infringe?
MR. NORTON: These three patents, no. I don't have
that and don't need that.
What I have is the internal documents and the
documents that they actually gave to Sun that say, this is,
basically, what Android is going to look like.
THE COURT: What year is that?
MR. NORTON: 2006. Because they provided a document
called the "product requirements document," which they gave it
to Sun. And it sketched out, and here are the things that
65
Android will have, and here is how we want to do it, and this
is what the virtual machine will look like, the JAVA virtual
machine, and these are the things we need you to do. And it
spelled it out.
And there were discussions, and then people at Sun
began to create work plans for, How do we to do our part of
this job? What will we need to contribute? How many engineers
will we need? And so on.
So the engineers looked over those documents and
said, okay, we have a sense now for back in 2006 what everybody
was thinking about. And that helps us identify which groups of
technology, those 22, would have been relevant, and which
particular patents would have risen to the top in the parties'
discussions. But we don't have enough information to say, oh,
they would have used, you know-- for boot, they would have
chosen these particular patents. They would have used our
in-class technology as opposed to some other technology. They
would have had choices. And the engineers weren't able to say,
we could predict, based on what we know, we the engineers know,
what Google would do.
THE COURT: But are there e-mails internal to Google
from 2006 that were not provided to Oracle at that time, but
which zero in on these three patents or the technology that
were internal to Google, which would lead us to believe that
even at the time of the hypothetical negotiation Google knew
66
good and well these are the three patents it was going to
infringe or it was going to need?
MR. NORTON: Not in the way Your Honor phrased it,
because there's -- there are no documents in which Google says,
We are going to infringe the '104.
THE COURT: Or that, We need this technology.
MR. NORTON: What I would say -- but I would answer
Your Honor's question yes, in this sense. There are the
documents that have been analyzed by the technical experts,
which show that by 2006 Google had begun to infringe.
So, Google had chosen this technology by 2006. They
had built it into Android. And that's why they're infringing.
So the -- there are e-mails that say, we need --
THE COURT: 2006 they were infringing, is that
your --
MR. NORTON: It is not the same for every patent
but -- and I can't --
THE COURT: These three, were they infringing these
three?
MR. NORTON: Not all of these three. But I believe
by 2006 they had already begun to infringe the '104.
This is not an issue that is in dispute between the
damages experts in the sense that Professor Cockburn has a
hypothetical negotiation in spring of 2006, and Google doesn't
contest that that is the date on which the negotiations take
67
place; although, their experts do assume that Google waited
until longer to infringe some patents. But, there was
infringement as early as 2006.
So we do have documents from Google at the time that
they begin to infringe, that, of course, show that they chose
this particular technology to incorporate in Android. That's
how they infringe.
THE COURT: I've got to let the other side respond.
Thank you.
What do you say to that over there, Mr. Purcell, that
the -- in the hypothetical negotiation, yes, it's true that the
engineers can only give a broader, more crude group of 22, but
that Google going into those negotiations would have had better
information as to the ones that it needed, and that it's
reasonable to presume that it would have known that it needed
the ones it was about to infringe? What's wrong with that
analysis?
MR. PURCELL: Well, all they have to go on is this
product requirements document which shows Google's expectation,
which doesn't say anything about specific patents. It doesn't
say anything about specific functionalities. It talks about
the general structure of what Android would be.
And, really, their argument is circular. Their
argument is that the three most valuable patents are three of
the patents that they chose to accuse Google of infringing in
68
this case. But if you want to talk about revealed preference,
this case, as Your Honor knows, originally involved seven
patents. Four of those seven aren't in the top 22. One of
them, the '520, which is still in the case, is somewhere down
around number 100 among the 569.
So it's obviously not the case and can't be the case
that you can just look at what Oracle alleges Google has done
in the way of infringement, and infer from that that you've got
the most valuable patents.
THE COURT: But you don't want to argue to the jury
that you infringe even more and, therefore, your logic is not
right. That would mean you owe more money, not less money.
MR. PURCELL: Well, we obviously don't plan to argue
to the jury that we infringe the patents we have been accused
of infringing.
But even using Oracle's logic, which is that the
patents that they've accused us of infringing are necessarily
the most valuable to Google because they reveal Google's
preference, their own damages report doesn't support that.
THE COURT: But you deny you infringe anything. So
those accusations just cancel each other out.
MR. PURCELL: That's my point, Your Honor. Their
argument is based on the patents they have chose to infringe.
So I don't think it has any probative value at all, this
business of revealed preference.
69
THE COURT: But if they prove that you infringe those
three -- that's the assumption we have here. We don't get to
damages otherwise.
If they prove that these three have been infringed,
then why isn't it a reasonable assumption to presume that the
negotiator in 2006, from Google, would have known that in due
course Google was going to need a license for those three
patents; and, therefore, be willing to pay a little more for
those three patents than for the others?
MR. PURCELL: There are a couple of reasons why.
Number one is that I believe one of the top three -- I think
it's the '720 didn't even exist, wasn't issued until 2008. So
it wasn't in a fact at the time of the hypothetical
negotiation.
THE COURT: I didn't know that. Say that again.
MR. PURCELL: I believe the '720 Patent wasn't issued
until 2008.
THE COURT: How could there be infringement of that
until it's issued? Right?
MR. PURCELL: There isn't. Both experts have assumed
that the proper date for the hypothetical negotiation was in
2006 when --
THE COURT: That's what you wanted. Please don't
back up. The other side was willing to go with a much later
date, and you argued for 2006. Right?
70
MR. PURCELL: That was the date of first infringement
of one of the patents.
THE COURT: So if you want to push the date back so
that -- I'm pretty sure you don't want to do that, but if you
want to push the date of the hypothetical negotiation back,
we'll reopen that subject.
MR. PURCELL: There are a bunch of different ways
that Google could have gone in 2006. There are a bunch of
different paths that it could have taken. There's not any
evidence that suggests that Google, at the time of the
hypothetical negotiation, focused on these specific patents or
this specific technology.
THE COURT: Maybe the way to deal with this problem
is for your experts -- not experts, but your real witnesses,
instead of these hired gun experts, your real witnesses will
come in and say, I was there at the time. We didn't even think
about this. We could have designed around all of these
problems.
MR. PURCELL: And there will be testimony to that
effect.
THE COURT: The e-mail that Mr. Reinhold came up with
will be quite important, at that point.
MR. PURCELL: Well, remember that was in 2010. That
was many years later. That was after a lock-in.
THE COURT: All right. You can maybe make that point
71
to the jury.
MR. PURCELL: We will. We will.
The other thing that I should mention regarding the
2006 time frame it -- well, actually, Your Honor, unless you
have any specific questions I think I'll stop there.
THE COURT: All right. What is your next objection
to the report?
MR. PURCELL: Our next --
THE COURT: My court reporter is looking at me like I
need to take a break. It's time for a break so the court
reporter's fingers can rest.
You be thinking about what your next point is going
to be, and then alert the other side so they can be thinking,
too.
We'll take 15 minutes.
MR. NORTON: Your Honor, one question. Your Honor
asked me to respond to particular points in Mr. Purcell's
presentation, but not to its entirety. And he has made
arguments that I have not yet had a chance to respond to.
THE COURT: We're going to move on to the next point
anyway.
MR. NORTON: Thank you, Your Honor.
THE COURT: Save it for the end of the hearing.
(Recess taken from 9:10 to 9:28 a.m.)
THE COURT: Let's go back to work. Please be seated.
72
What, Mr. Purcell, is your next objection?
MR. PURCELL: So given the comprehensiveness of the
briefs, I really think there is only one other issue I want to
raise this morning.
THE COURT: All right. What's that?
MR. PURCELL: Which has to do with the conjoint
survey that Dr. Shugan did.
THE COURT: Okay.
MR. PURCELL: And we've got four objections to that.
Number one is, this is just not a tool for estimating
damages in litigation. There is not case law that approves it.
We pointed that out in our brief.
Oracle responded with a declaration from Dr. Shugan
that's multiple hearsay, where he refers to other industry
professionals who purport to opine that conjoint surveys were
used for some purpose in other cases.
We don't have orders --
THE COURT: But there have been surveys in other
cases.
MR. PURCELL: Consumer surveys, yes.
THE COURT: That's what this is, isn't it?
MR. PURCELL: Well, there's a big difference, Your
Honor, between doing a survey about confusion in the context of
a trademark case or surveys about reliance in the context of a
consumer fraud case, and this sort of conjoint survey, which
73
purports to map consumer preferences about features to
reduction in market share for purposes of estimating damages.
Conjoint surveys are market research tools that are
commonly used when companies are designing new products. They
are not mathematically precise tools that are used to estimate
damages.
And there isn't any example we've been able to find
in the case law of a court saying, yes, you can use conjoint to
map damages in a case like this; or in any case, for that
matter.
And the only point I wanted to make on that is that
Oracle's only response was to say that Dr. Shugan has been
informed by other people that conjoint surveys were used in
other cases. That's completely inchoate. We don't know what
those cases were. We don't have orders. We don't have
transcripts.
The second point I wanted to make is that
Dr. Shugan's methodology disproves itself. It proves that it
is not sound, just looking at the results that he received.
The purpose of a conjoint analysis is to test the
value of individual features by varying those features, and it
relies on an assumption that the respondent, as that one
feature is varying, is able to hold all other features of the
product constant.
So everything else equal, if you reduce the
74
application load time, if you increase the application load
time, how would the consumer preference be affected?
And what Dr. Shugan's results show is that consumers
in his survey weren't holding all other features constant.
They were implying all sorts of other features that varied
depending on what they were looking at.
And the reason we know this is because 24 percent of
respondents in his survey said that they would either prefer or
be ambivalent, be indifferent, between a more expensive phone
as compared to a cheaper phone, all other features constant.
So 24 percent say they would rather buy a $200
smartphone instead of a $100 smartphone, all other features
constant. That obviously makes no sense.
THE COURT: What percentage said that?
MR. PURCELL: 24 percent.
THE COURT: So 24 percent said, we're indifferent as
to spending a hundred dollars more for the same item?
MR. PURCELL: They were either indifferent or they
preferred to spend the extra hundred dollars. So either they
are completely irrational or they were inferring that the more
expensive phone had some additional feature that made it worth
the money. And we don't know. So, that's point two.
Points three and four are illustrated on the fourth
and fifth pages, the last two pages of the handout that we
passed up to Your Honor earlier.
75
The fourth page, next to last page, is a list of all
of the features that the focus group that Dr. Shugan's
assistant conducted prior to the actual conjoint survey
these -- these are all of the features that respondents in that
focus group mentioned as potentially bearing on their decision.
There's 39 of them.
Dr. Shugan selected seven of these. He ignored the
other 32. Among the ones he ignored and didn't test for are
things that are obviously important to consumers, like network.
Do you have Verizon? Do you have AT&T? What's the coverage
like in your area?
THE COURT: If it's highlighted, that means he -- I
don't understand your code here.
MR. PURCELL: Your Honor, could I approach? My copy
actually doesn't have highlighting on it.
THE COURT: Why did you give me one that has
highlighting?
(Laughter)
MR. PURCELL: That's a good question.
THE COURT: Here. Give me the one you want me to
see, and I'll let you just have that copy back.
MR. PURCELL: Oh, no, I'm sorry. It's not
highlighted. It's just for readability, some of the rows are
gray and some are white.
THE COURT: But which ones are -- I thought you were
76
highlighting the ones that got ignored, or something.
MR. PURCELL: No, Your Honor. It's not indicated on
there. These are the totality of the features the focus group
pointed to.
Dr. Shugan selected seven of these. Among them:
Price, screen size, operating system. I believe application
startup time, availability of applications --
THE COURT: Help me -- all right. So I got -- so
this list is the full list of features that focus groups talked
about --
MR. PURCELL: Correct.
THE COURT: -- correct? All right.
MR. PURCELL: Of which --
THE COURT: Of which what?
MR. PURCELL: Of which he selected seven.
THE COURT: Which seven did he select?
MR. PURCELL: He selected application startup time.
THE COURT: Where is that on here?
MR. PURCELL: It's the third one in the first column.
THE COURT: Okay.
MR. PURCELL: Availability of applications.
THE COURT: Yes.
MR. PURCELL: And then if you go to the next column,
he selected multitasking.
THE COURT: Yes.
77
MR. PURCELL: He selected operating system.
THE COURT: Uh-huh.
MR. PURCELL: He selected price.
THE COURT: Yes.
MR. PURCELL: He selected screen size.
THE COURT: Okay.
MR. PURCELL: And there's one other that escapes me
right now.
THE COURT: All right. I don't know either. So --
MR. PURCELL: So there's 32 that he did not select.
And he didn't have any data, that people in the focus group
didn't tell him, well, these seven are more important than the
other 32.
THE COURT: How about availability of wifi?
MR. PURCELL: He didn't select availability of wifi.
He didn't select the brand of the handset itself, whether it's
a Motorola or Samsung, except to the extent that's captured in
operating system, because, obviously, only Apple makes the
iPhone.
He didn't select carrier. As I said, network,
whether it's Verizon or AT&T.
THE COURT: Won't his answer to that be, well, I
can't select those because the patents that are infringed don't
have anything to do with those features?
MR. PURCELL: That may well be his answer. But,
78
again, if you're trying to measure the way consumers actually
view the importance of certain features, you can't narrow it
down. You can't cull it down just to the features that are
important to the litigation.
THE COURT: Well, why not? Why is that?
MR. PURCELL: Well, because you're putting your thumb
on the scale in a very serious way, Your Honor. That renders
the methodology itself unreliable because you are removing
anything that would suggest that these patents are, in fact,
less valuable because consumers don't care as much about them,
and you're including only the features that implicate the
patents, which are going to inflate the value of the patents.
A consumer may very well care about network, may very well care
about, Do I get coverage?
Maybe they live in a rural area, and they may care
less about availability of applications because they just don't
do that. And, yet, they are not given the option in the
conjoint of stating a preference for network as opposed to
availability of applications.
THE COURT: Let me give you an example. Take
something simpler than that. Say somebody had a patent on a
radio, car with radio.
MR. PURCELL: Okay.
THE COURT: Okay. Car with radio. And along comes a
company that builds a car with a radio and infringes the
79
patent. So somebody then does a conjoint study that asks the
question, would you like to have a car with a radio or a car
without a radio? You would be in here criticizing they didn't
ask about what kind of tires are they going to put on the car.
In other words, you're raising issues that don't have
anything to do with what's being litigated.
MR. PURCELL: Well, because without a radio or a car
with a radio, I would imagine a hundred percent of people are
going to say a car with a radio. It's an extra feature.
And this is exactly -- we cited a quote from Judge
Posner, in the Apple-Microsoft case, recently, who rejected a
consumer survey on exactly this ground.
He said, basically, are you telling me that any
consumer is going to say if you have a choice between a product
and a product plus, that they are not going to choose the
product plus? Of course, they will; that's meaningless.
So here --
THE COURT: What if it was a car with a radio, but
car without radio and instead you paid a hundred dollars less?
So, in other words, there is a real choice there.
But my point is, you're asking them to compare
features like what kind of tires are on the car, and trade that
off. And maybe that's what the law requires, but --
MR. PURCELL: I think it's what logic requires in the
sense that, again, the product that Dr. Shugan is testing is a
80
smartphone. A smartphone is a device with a whole bunch of
features. All of those features may have some impact on
whether a consumer buys one phone over another.
Those features include things like network, include
things like availability of wifi, include things like who made
the handset, how attractive is it, what's the styling like?
None of that stuff was tested by Dr. Shugan. He focused only
on a few, narrow features.
THE COURT: What was the respondent survey group
asked to assume with respect to, say, wifi?
MR. PURCELL: They were asked to assume -- nothing in
particular, just that it was constant among all smartphones.
So we don't know what --
THE COURT: Let's -- well, if they were asked to
assume that it was constant among all options, why wasn't that
the right assumption to make, since the -- allegedly, the
infringing features are the ones that they did ask about? You
see what I'm getting at?
MR. PURCELL: I do.
THE COURT: Why do you have to ask about all the
things that are not in controversy?
MR. PURCELL: Well, there's a couple of reasons.
Number one is because you don't know to what extent a
consumer's preference is driven by those other features as
opposed to the features that are tested.
81
Now, I'm not saying and I don't think we would
contend or the law requires them to test every conceivable
feature. What Dr. Shugan could have done is figure out which
features are more important to consumers by doing a different
test, and then testing that set of features. He didn't do
that. He didn't ask any of those questions in the focus group.
The only thing he did was self-select. And, as we
pointed out in our report, actually, some of the features that
he selected were dictated to him by Dr. Cockburn or
Dr. Cockburn's assistant.
But leaving that aside, he selected seven features
with relevance to the litigation. Well, not all of them. I
mean, a couple of them, like price, I'll concede aren't
specifically just relevant to the litigation. But then he
excluded a bunch of other things that are obviously relevant to
a consumer-purchasing decision, without any basis for that.
So he doesn't have to test everything. But he has to
have some basis for selecting the features that he did, other
than they're useful to the litigation, because there may be
things that --
THE COURT: Well, not just useful. They are the
features that are enabled by the patents-in-suit.
MR. PURCELL: I agree that it was appropriate for him
to test those. He absolutely should have.
But what he should also have done was test other
82
features that may be equally important to consumers, or more
so, along with the features that are at issue in the
litigation. Because what he's purporting to demonstrate is
that if these features that are relevant to the case were
disabled or if they were less robust, that consumers would stop
buying Android, and they would stop buying it in certain
percentages.
So, we really don't know, based on his consumer
preference shares, whether he's accurately measuring anything,
because he's excluded a whole bunch of other things that may
actually be more important to consumers than the things he
tested.
THE COURT: Wouldn't they still be more important
under all of these scenarios that they were testing? Let's say
wifi is extremely important. It would be extremely important
under every option. So how would that get somehow traded off
on the respondents' answers?
MR. PURCELL: Well, we also know that the respondents
did not hold all other features constant because of the problem
I explained before, where you have 24 percent, almost a quarter
of the respondents saying they would rather pay an extra
hundred dollars.
THE COURT: That's the problem of maybe the
instructions weren't clear enough, or the way it was
administered it wasn't clear enough. That's a troubling point.
83
I agree with that. But that's different from which features
they were asked to test.
Let's be clear. Go back to the 24 percent. Is that
just some number that you kind of glued together, or is that a
conceded defect?
MR. PURCELL: I believe it is conceded in Dr.
Shugan's reply declaration. Rather, his declaration in support
of Oracle's opposition.
And in Oracle's opposition, Dr. Shugan said that
24 percent of survey respondents were either agnostic between
the hundred dollar and the $200 phone, or they preferred the
$200 phone all other features being equal. I think the number
he gave is actually affirmatively preferring the $200 phone was
8.8 percent.
THE COURT: So it was clear that -- in other words,
if they had followed the instructions, was it clear that the
phone that they were opting for would cost a hundred dollars
more without any more features?
MR. PURCELL: Yes. The price was specified. And
they were instructed to hold all other features constant. So
they were supposed to focus only on the features that were in
dispute, and assume that the phones were identical in all other
respects.
THE COURT: Did anyone do -- I don't even know if
this is possible, but sometimes what they do is they throw out
84
everyone who has an irrational -- let's say you throw out that
24 percent and then did a survey just of the 76 percent that
were left over. Did anyone do that to see how that -- how
those numbers would come out?
MR. PURCELL: I think Dr. Shugan did try to
rehabilitate his analysis, after we criticized it, by doing
that.
It doesn't solve the fundamental problem, which is
that you have a survey that's designed so that 24 percent of
respondents are giving completely irrational answers. How can
the methodology be sound --
THE COURT: That may be, but when he rehabilitated,
did the answers and percentages change or did they stay the
same?
MR. PURCELL: I would have to leave it to Oracle's
counsel to address that.
THE COURT: What's your next criticism of the
conjoint --
MR. PURCELL: The final criticism has to do with a
question Your Honor asked -- and this is addressed on the last
slide in our packet -- which was about confidence intervals.
So, initially, Dr. Shugan did not calculate
confidence intervals. And when I asked him the series of
questions that's on slide 5 of the deposition, he said that it
wasn't possible to do it because he was using a Bayesian
85
estimation, and it was not possible to calculate confidence
intervals using that.
And then in Dr. Leonard's report -- and this is his
initial report, not the recent supplemental report -- at page
116, Dr. Leonard pointed out that, in fact, Dr. Shugan was
wrong; that there was a way of calculating, essentially,
confidence intervals using a Bayesian estimation. This is the
Bernstein-Von Mises theorem, which I don't purport to
understand.
And, lo and behold, in the slides that Oracle
prepared for today and Your Honor has in your smaller binder,
Dr. Shugan has submitted binders that do purport to have
confidence intervals in them.
We haven't had an opportunity to test them. We saw
them for the first time around 5 o'clock last night.
THE COURT: What tab is that?
MR. PURCELL: That I don't know.
I believe it's 8, Your Honor.
THE COURT: Okay. All right. Let's hear from the
other side on this, these points from conjoint only.
MR. NORTON: Thank you, Your Honor.
So, I'll try to address Mr. Purcell's points in the
sequence in which he made them. His first argument is that
conjoint is not appropriate for the calculation of damages.
That is nonsense.
86
First, Google's copyright expert in this case,
Dr. Alan Cox -- and we cite this in our papers -- has twice
written articles in which he says that choice modeling, which
is conjoint, is an appropriate way to calculate damages in
infringement cases.
And when Google tells you that they are aware of no
case in which conjoint analysis has been used to calculate
damages, they can only say that by not reading the articles
that their own expert has published. And those are attached as
exhibits I and J to the Norton declaration in support of the
opposition that we filed on the Daubert.
And what Dr. Cox said, in the article attached to
Exhibit I, is that choice modeling -- this was in 2003 -- is a
relatively new technique, but it was used to good effect in an
infringement suit brought in the mid '90s. And which he then
proceeds to describe the lawsuit in which the plaintiff proved
its damages by using choice modeling which, again, is conjoint
analysis. And he goes on to explain how choice modeling works
and how it presents the consumer with a series of choices and a
study that is, in all material respects, identical to the type
of study that Professor Shugan did.
Exhibit J is the abstract of an article which is
available on the Internet and is cited in our brief. And that
is an article by Dr. Cox, Google's copyright damages expert,
titled, "Survey Techniques for Rigorous Measurement of Damages
87
in Trade Dress Confusion Cases," in which he goes on to explain
that choice modeling is a rigorous measurement of damages in
infringement cases.
Dr. Shugan is a person who is familiar with the
literature, cited a number of cases by name, in which conjoint
analysis has been used. It's used over and over again.
But even if Google were right that it had not been
used specifically to calculate damages, Lucent tells us that
it's appropriate, for the purpose of doing a hypothetical
negotiation, to look at survey data that would have been
available to the parties at the time of the negotiation.
And this is the type of survey that they could have
done. And it's used in the way -- and the way in which
Professor Cockburn uses it is similar to the way in which
parties in a negotiation could have used it. That is, how
important is it to our smartphone that applications load
quickly? How important is it to have lots of applications
available? Which things are more important to consumers?
Where should we really spend our time?
And the conjoint analysis tells you which features
consumers think are important, and it's appropriate to consider
that under Lucent.
THE COURT: Which of the seven things measured
copyright as opposed to measured the hardware part?
MR. NORTON: It is the availability of applications.
88
And the argument is thus: The reason why it was important to
Google to have the API specifications, the API packages that
they infringed, was that they are familiar to carriers. But,
also, they are familiar to programmers.
And if you wanted to be able to quickly get a large
number of applications available for your smartphone platform,
it would be very important to have an established developer
base that is already familiar with your programming language
and the APIs that it requires.
And so Google --
THE COURT: Not the programming language, because
that's -- you've already conceded that away in this case. But
the APIs, is what you're claiming.
MR. NORTON: That's right. Because Dr. Astrachan,
who is Google's copyright expert, says that the APIs are
essential to -- practically required, he says -- if you're
going to be writing applications in the JAVA language.
The JAVA language is free to use, but the APIs are
copyrighted. And when Google incorporates the APIs into the
Android platform, it infringes. It wants to do that because
the programmers are looking for those APIs. They don't want
different APIs. They don't want to be confused with some new
set of APIs, even if they are written in the JAVA language.
So Google decided, we're going to use this collection
of specifications, this hierarchy of specifications, these
89
interrelationships between APIs, because these are the ones
that the millions of JAVA developers expect to see.
And they did that because if they didn't do that, if
they went with some other language -- which they thought about
and rejected -- if they went with some other language they
wouldn't have a developer base.
THE COURT: All right. So in the study that was done
here, how did the availability of applications feature rank
among consumers?
MR. NORTON: So, Professor Shugan calculates how many
consumers would switch in the event that the number of
applications available on the phone were lower.
So what he used was around the time of -- I believe
he used 2010 as the example, the number of applications that
were available on the Palm platform, the Blackberry platform
and the iPhone platform as the choice sets.
So 6,000 -- when you're presented with the phone
choices, there's a phone that has available to it 6,000
applications, 40,000 applications, or a hundred thousand
applications. And how does that affect consumers' choices?
And then he's able to determine the relative
importance of an application universe to consumers' choices on
a smartphone relative to all the other features. And then he's
able to calculate the effect on Android market share as a
result of the inability to provide a robust application
90
universe like that. And that --
THE COURT: The way in which you use this is one step
removed from all of that. Somehow it's used to allocate the
original 2006 bundle, by this assumption or conclusion that the
copyright API part was worth one half of the patent part,
right?
MR. NORTON: That is one of the ways in which it is
used. Let me explain that.
THE COURT: All right.
MR. NORTON: It's -- it seems like a fairly simple
syllogism to me. One of the key virtues or benefits of the
copyrighted APIs that are infringed is that they provide
applications, as I just discussed. One of the key benefits of
the patents-in-suit is that they provide speed, memory,
performance benefits on the smartphone.
And those speed benefits have been measured by the
technical experts. And so we know from the work done by the
technical experts that if you -- by virtue of Google using
these patents, it has improved the performance of the phones by
a certain amount.
When Dr. Shugan did his conjoint study, he used that
incremental benefit. So how important is that incremental
benefit that we attribute to the patents, the performance
benefit? How importance is that performance benefit to the
consumers relative to the value of the applications to
91
consumers?
And so what we know then is consumers value the
incremental speed at about twice what they value the
incremental applications.
So that tells us that the patents-in-suit, the speed
provided by the patents-in-suit, is twice as helpful to your
smartphone as to the applications enabled by the copyrights in
suit.
So by valuing those two attributes relative to one
another, you can apportion the 2006 bundle. So that's the way
in which it's used there, is that we know the relative
importance of these two things. An observation which is true
even if every argument that Mr. Purcell made were correct about
the failure to select features.
Now, Dr. Shugan has explained why that argument is
completely wrong, and I'm more than ready to address it. But
even if that were true, if all you want to know is the relative
value of those two features, then this conjoint would be more
than adequate to do that, even taking into consideration, even
crediting Google's objections. So that's how it's used for the
allocation process.
Professor Cockburn uses it for a second purpose,
which is that the conjoint survey establishes that the speed
benefits that are provided by these particular patents, and the
application benefits provided by these particular copyrights,
92
are important to consumers.
And Dr. Shugan has actually measured, in terms of
market share, just how important they are. But what
Dr. Cockburn is able to do from that is he can say, well, in
addition to all the other evidence I have considered, I've
considered the results of the conjoint, which tell us the
unsurprising fact -- but still verified by data -- that making
the phone perform more quickly makes a difference in terms of
sales, and having lots more applications makes a difference in
terms of sales. And the conjoint confirms that.
That is something that, for example, Judge Rader, in
the Cornell-HP case, said was an important thing for the
plaintiff to put forward. You just don't come in and say,
well, you know, it's important. Consumers like it. You know,
everybody always assumed that consumers liked it.
Where are the demand curves? Where is the consumer
survey? Where is something that tells me that people really do
value it?
So Professor Cockburn has that additional data point
that tells him that not only did Google think it was important,
so did consumers. And that's what the parties would have
thought at the time of their negotiation.
So those are the two ways in which the conjoint is
used to inform Professor Cockburn's calculations in this case.
And that is entirely appropriate, both as an apportionment
93
exercise -- it's a good way to apportion between the value of
the copyrights and the value of the asserted patents. It's
also a good way to determine the consumer demand for these
particular attributes.
So turning to chart 4, that Google submitted today,
these are the features, of course -- this is Professor Shugan's
own slide. It's his own exhibit. And these are the things
that in the focus group people mentioned.
Now, in a conjoint -- I don't believe this is
contested in any way. In a conjoint -- the beauty of a
conjoint analysis is that you do not need to test every
feature. If you had to test every feature, it would be very
burdensome, and it wouldn't be particularly useful. And, in
fact, that's not required. And that's why companies use it,
the government uses, it's used in litigation, because you can
isolate a few features that you care about and only test those.
And, so, in his declaration, submitted in support of
our opposition brief, at paragraph 25, paragraph 25,
Professor Shugan responds directly to this criticism. He says,
"Importantly, it is not necessary in conjoint analysis to test
every feature that may matter to consumers because conjoint
analysis assesses relative" -- in bold italics by
Professor Shugan -- "relative importance."
Nonetheless, he explains, he did include features in
the survey that are not at issue in the litigation, and he
94
explains why. Quote, I took this additional step to estimate a
well-specified model.
Goes on to say, quote, This approach is
methodologically sound, as I included the two most critical
features that generally derive the greatest value in any
estimation and that captured the benefits of Google's
reputation." Those additional two are the operating system,
which is called Android, and so it incorporates Google's brand
value, and price.
Now, what's important in conjoint analysis is not
that you test every feature, but that you make sure to include
among the most important features as Professor Shugan has
explained. So he's got price and he's got the operating
system. And in his study those are the most important features
by a substantial margin.
The operating system is considerably more important
in consumers' decision than the number of apps. It's more
important than the incremental speed benefits that we claim are
provided by the patents. Price is also more important. But
he's included those. And he doesn't need to test everything.
That's not how you do a conjoint analysis. He says so in his
declaration.
There is no declaration from any Google witness that
says you're supposed to include 36 features, because you're
not.
95
THE COURT: Well, but for purposes of your second
use, I wonder if that's correct. You said that the second way
in which Dr. Cockburn uses this is to say that the features
enabled by the patents-in-suit are important to consumers. But
if you only test those features that are enabled by those
patents, and ignore the other features, like wifi and camera,
maybe you don't get the very good benchmark for comparison.
MR. NORTON: Well, that's -- if you only tested the
features that are in dispute, then that would cause some
complication.
THE COURT: So what, other than operating system and
price -- that is true that that's all he tested.
MR. NORTON: No. Screen size is also not in dispute.
But Professor Cockburn has testified and stated in
his declaration that he doesn't need to test the others in
order to be able to calculate the effect on market share, so
long as he's got a significant -- another factor is significant
to consumers, brand and price, that he will be able to estimate
the effect on market share. This is how conjoint is done.
THE COURT: That's not the two times part. That's a
different -- market share, that's separate from the ratio of
two to one.
MR. NORTON: That is correct.
THE COURT: How does he use market share then?
MR. NORTON: So what he does is by using the conjoint
96
analysis he's able to determine how many consumers would choose
a different phone.
They would not choose the Android phone if the
Android phone lacked the attributes that are provided by the
patents and copyrights in suit. Those are called "preference
shares" in his analysis. And there's a very close correlation
between preference shares and market shares in the analysis.
So what he's able to do is by presenting -- the way
the survey works, just to give a little more context, the
consumers are shown each of the features. So this is the thing
that we are going to ask you about: Launch time. Screen size.
They're given a picture of a phone next to a Coca-Cola can so
they can actually get a sense of perspective. They see all
these things.
And then they are actually given choice sets, four
phones with different permutations of features. And they say,
I want the hundred dollar phone with the 4 second startup and
the 40,000 applications, that's an Android phone. They pick
that one.
Then they are given a new set of choices, and then
another set of choices. And the computer software is able to
refine the choices as they make their selections, to really
hone in on which variables are the ones that are most important
to this consumer.
And, as a result of that, Professor Shugan is able to
97
do a statistical analysis that tells him if the phone lacked
the attributes that our technical experts say the infringement
provides, how many consumers would not have bought their
Android phone, they would have switched to something else?
And we see a substantial effect on market share as a
result of not being able to infringe.
Now, in his September report, Professor Shugan
actually translated that into dollars. If Google's market
share declined by X, how many dollars would it lose? And then
that was his numerator. And his denominator was the value of
Android over the same period. Right. The Court has forbidden
that analysis. He doesn't do it anymore. But that numerator
is still a very important number.
If you didn't have these patents, if you didn't have
these copyrights, would you lose a lot of market share? And
the consumer survey says you would. You would lose a lot.
And Professor Cockburn considers that as one of the
indicators of the value of these particular patents in these
particular copyrights. But it's not necessary -- and, again,
this is unrebutted. It is entirely unrebutted. It is not
necessary, in performing a conjoint analysis, to include every
feature.
In paragraph 25 of Professor Shugan's declaration, it
is -- he testified that way in his deposition. It's just not
how conjoint works.
98
And an argument that shows you 36 features and says
the conjoint survey is methodologically flawed because it
doesn't have all 36 of these or 18 of these is just a lawyer
argument that fails to comprehend what Dr. Shugan's
uncontroverted testimony establishes.
THE COURT: What do you say to the 24 percent error
rate, that 24 percent gave irrational answers?
MR. NORTON: Okay. Well, there is what I say and
there is what Professor Shugan says and they are saying.
Your Honor asked, is this a number that Google
created, or is it actually something Dr. Shugan concedes? And,
inexplicably, that was described as a concession.
So Dr. Shugan, in his declaration in support of our
opposition -- this is docket 740 -- explains in great detail
why that's wrong. And he did it back in September when we
moved to strike the opinion of Dr. Leonard. He submitted a
declaration then.
THE COURT: What do you then say is the correct error
rate, or however you want to characterize irrational answers?
What do you think is the correct rate of error?
MR. NORTON: What we are talking about is not a rate
of error. So, the first mistake is that Google says that they
have identified --
THE COURT: Irrational answer there.
MR. NORTON: There are --
99
THE COURT: People don't want to pay a hundred
dollars more for nothing.
MR. NORTON: Sure.
THE COURT: So are you saying there's zero of those
instances?
MR. NORTON: No. So, 8.8 percent of the responses
indicate that individual consumers said, I will take the $200
phone over the hundred-dollar phone that is otherwise
identical.
So the question is, does that -- what does that
8.8 percent mean? We do see that. But the problem is that, in
Google's critique, is that Professor Shugan -- I'm going to --
Mr. Purcell said that there's a mathematical theorem that he's
unable to explain. I'm going to find myself in the same
position, shortly, but I can make a little headway.
Dr. Shugan uses something called a hierarchal
Bayesian approach to probability. It is -- Bayes was a
mathematician in the 18- --
THE COURT: I know what Bayesian probabilities are.
MR. NORTON: Good. All right. So it's Bayesian.
It's Bayesian approach.
And the Bayesian approach, as Dr. Shugan has
explained, I believe, three times now in submissions -- in
testimonies and submission to the Court, you don't test a
hierarchal Bayesian model by looking at individual responses
100
and saying these appear to be irrational.
There are different checks that you use to assess the
robustness of the hierarchical Bayesian model. And he did
that. There is a U squared. There is a hit rate. Those are
disclosed in his original report. And those indicate that his
survey is very robust and very good at predicting aggregate
market share changes.
It is incoherent to critique Dr. Shugan's analysis by
isolating individual choices and saying some set of individuals
gave responses that we would not expect to see. You cannot
critique a Bayesian analysis in that way.
There is some useful information about individual
choices. But if you want to know whether this is a good study,
this is the wrong way to do it. If it were what's called a
frequentest approach, that might be a more useful critique.
So Dr. Shugan, in his declaration, in paragraphs
35 -- I should say 34 through 44, deals directly with this
criticism and explains why you cannot draw the conclusion that
Google insists on drawing, the 8.8 -- it is not 24 percent.
The number that they are trying to calculate is 8.8 percent.
And the 8.8 percent does not mean what they say it means.
And where is the testimony that establishes that
Professor Shugan is wrong? Where is the analysis that says
that Professor Shugan, who has a Ph.D. in statistics and does
this over and over again, doesn't understand Bayesian
101
probability? It doesn't exist because it's just not true.
So in paragraph 42 he explains precisely why this
approach does not -- the approach being the 8.8 percent --
doesn't mean what Google says it means.
In paragraph 40, he points out that what -- what
Google's critique is they say, well, consumers can't be
irrational. If you do a survey you expect all the consumers to
give rational responses.
And we know from research, which Dr. Shugan quotes,
that consumers don't always make rational choices. In any
survey you get some responses that wouldn't make sense.
And so Google says, well, those show that the
survey's irrational. But that's not true. It shows that
consumers gave answers that you might not expect.
Now, is there a reason why you might think consumers
would, in fact, sometimes say they prefer the more expensive
phone? Again, Dr. Shugan says, yeah, that happens.
And there's literature on that, as well. And that is
that people sometimes associate a higher price with prestige or
durability. And so they're going to see that higher price and
say, well, that's a phone -- I might prefer that phone because
I want the more expensive phone.
And we all know people who prefer to buy the more
expensive thing because they like the status that it confers.
And Dr. Shugan addresses that directly.
102
So the fact some people say they would pay for a more
expensive phone doesn't mean that the survey is irrational. It
means that some people don't behave the way Google claims
people ought to behave.
THE COURT: All right. I need to bring your part to
a close. I'll give you one more point.
MR. NORTON: The Court asked whether Dr. Shugan had
recalculated his results to address this particular argument.
And he did. It is the -- the recalculation was in his
declaration submitted, again, docket 740. It's at footnote 44
of his declaration.
And what it says is that if you were to accept this
argument and recalculate the numbers by throwing out those
responses, the Android sales, but for the feature enhancements
enabled by the patents-in-suit and copyrights in suit, would
have been 7.6 percent, at least 7.6 percent lower, as opposed
to his earlier calculation, 7.9 percent lower. So it does not
have a substantial effect on the result if it were, in fact, a
meritorious criticism.
THE COURT: All right. Let's hear rebuttal on these
points. Not all the points, but just your main rebuttal.
MR. PURCELL: Thank you, Your Honor.
So I understand, of course, why Oracle likes to cite
Dr. Cox's articles to us. But what Dr. Cox has opined as an
economist on this issue, it's not the finding of a court. It's
103
not the finding of either the district court, much less the
federal circuit, much less any other appeals court, that
conjoint analyses are appropriate for estimating damages in a
case like this.
Oracle hasn't cited any cases, they haven't provided
any evidence that there, in fact, is any precedent for using
this sort of analysis for this purpose.
The next point, just briefly, Oracle is attacking a
straw man. I made very clear in my presentation to you that I
was not saying that Dr. Shugan's methodology was reliable
because he didn't include everything under the sun, he didn't
test 39 features. That's not what we're talking about.
Mr. Norton just admitted that it is important to
test, in addition to the features at issue in the litigation,
the other features that are most important to consumers.
Dr. Shugan didn't do anything to figure out which of
the omitted features was, in fact, important to consumers. He
didn't include price. And, obviously, price is important. He
didn't include operating system. Obviously, that's important.
He didn't include a whole laundry list of other things. Didn't
make any attempt in the focus group to ask respondents what
they valued so he could even have an opinion about that
question. He just completely ignored it.
And the third point, Mr. Norton harped on the fact
that Google's expert supposedly didn't contravene what
104
Dr. Shugan did. Dr. Leonard did that, in fact, in a
declaration that he submitted last year when Oracle moved to
strike his rebuttal opinion.
But I'd just like to focus you on paragraph 39 of the
Shugan declaration that Oracle itself submitted regarding the
24 percent number. This is the declaration Oracle filed on the
24th of February.
And Dr. Shugan writes, "Second, for the majority of
the 24 percent of respondents the estimates for the two
prices" -- that is the hundred dollar and $200 prices -- "are
so close that a diligent statistician would consider the
difference to be zero, rather than representing some form of
rationality. As I explained in my reply report, I excluded
respondents with utilities associated with a hundred and $200
that are within one standard deviation of the difference in
utilities between levels as a sensitivity analysis. When these
respondents with such utility comparisons are excluded from the
analysis only 8.8 percent, not 24 percent, prefer a price of
$200 over a price of $100."
So what he's saying there is that 24 percent are
either agnostic between the hundred dollar and the $200 phone,
or they prefer the $200 phone. And that's exactly what we're
saying.
So when they say we don't have any evidence, we have
an admission. We have it in Dr. Shugan's own words. That's
105
exactly what he is saying here. And it affects the entire
analysis. It's a problem with the methodology.
That's all I've got.
THE COURT: All right. What's your next criticism?
MR. PURCELL: Your Honor, I'm content to rest on the
briefs with respect to the other criticisms. We raised the
issue about copyright apportionment, about failure to perform a
claim-by-claim analysis.
THE COURT: I have a couple of questions.
What happens -- I need to preface this by saying
Dr. Cockburn's analysis is like an algebra problem. He has,
the copyright part is one half of X. X is the patents-in-suit.
Copyright API is one half X. And then everything else that is
a patent in the 2006 bundle is Y, I guess. But it is a ratio
of the -- you have to go through all of those. That's why I
asked you to check my math because I must have done it wrong
somewhere.
Now, this formula does not appear anywhere in the
history of the universe. I understand that. But that doesn't
mean you can't use algebra. Algebra we know about. And as
long as his assumptions are valid, I guess that's okay.
Anyway, he uses that formula in the one half of the
value of the patents-in-suit equals API value, in getting his
apportionment.
All right. Here's my question. I have trouble
106
figuring out what happens if the federal Patent Office knocks
out all but one patent, or the jury knocks out all but one
patent. What then happens to this analysis?
So while you're standing there, have you worked your
way through that analysis? I don't want you to guess at it.
What happens to the analysis in that algebraic
formula, and where does that lead us if -- in other words, the
conjoint study is based upon a set of answers that assumes
certain features, and those were selected based on what
infringed or didn't infringe.
But if it turns out those patents are not valid in
the first place, then maybe the conjoint falls away, and the
one half falls away, and the whole formula falls away, and we
have no way to apportion. That's what I'm worried about.
Do you know the answer to that?
MR. PURCELL: The answer to that lies in the
distinction that Your Honor discussed with Mr. Norton about the
two separate uses of the conjoint.
One is to measure the relative preference as between
number of applications and application speed. And I think
that's a different use of the conjoint than if you're asking to
determine a bottom line market share reduction number if a
feature doesn't perform as well.
And so I think our challenge, our Daubert challenge,
is directed at the second problem more than the first problem.
107
We were not -- at least, it's not in our brief that
we're attacking the conjoint as an idea of measure of relative
preference.
I think what Dr. Cockburn has done is calculated the
value of the five patents-in-suit now, assuming that they are
all infringed. He has calculated subcategories for what each
of the individual patents are worth.
THE COURT: Where is that? Where can I find that?
MR. PURCELL: I believe that is at the very end of
his exhibits. It might be Exhibit 37. Exhibit 37 has
percentages.
I think Oracle may have prepared a slide for this,
that I can direct you to. It's slide 3, Your Honor, in the
smaller binder Oracle handed up.
THE COURT: All right.
MR. PURCELL: So if that analysis is sustained -- and
we haven't challenged it, other than to the extent it relies on
things we have challenged, but we haven't challenged that
specific apportionment. If that stands, then that would allow
for a patent-by-patent liability determination.
THE COURT: All right.
MR. PURCELL: With the copyright, still, I think
being half of what the broader number would have been had all
those patents been invalid and infringed.
THE COURT: What does the other side say to that
108
question? Do you agree with what counsel just said?
MR. NORTON: It's very close.
The value of the copyrights is equal to half the
value of speed, as measured by the conjoint. The value of
speed, based on the 2006 deal between the parties that was
never consummated, is calculated by Professor Cockburn.
So even if the patents are not valid and infringed,
we know what a group of patents that gave you an incremental
benefit, in terms of performance, would be. And we know the
patents -- the copyrights would be worth half that.
So the number is still a meaningful number, even if
no infringement were ever proven as to those patents. So you
don't need to prove patent infringement to have a copyright --
THE COURT: So let's say the jury determines all of
them are invalid, patent-wise, but then your position is that
the number for copyright on the apportionment of the original
2006, that that copyright API number remains the same
regardless, regardless of which patents are valid or not.
MR. NORTON: Right. Because the value of the
copyrights is not actually dependent on the value of the
patents.
We know there is a relationship between the value of
the benefits the patents provide and -- let me put it this way.
Let's say that the patents were not valid, but they were
infringed. So Google really did do those things. They really
109
did use the inventions, and they got the benefits from the
inventions, and they got the value. In that event, we would
fail to prove our patent infringement case.
But all the evidence that we have that shows that the
benefits of speed are precisely what we claim they are, all
that would still apply.
So the copyright's value doesn't actually depend upon
the value of the patents. We've used the value of the patents
as a tool to help us measure the value of the copyrights,
assuming the patents are valid and infringed. And even if that
assumption were false, the measurement of value would still be
useful.
THE COURT: Thank you.
I have a question for Mr. John Cooper. That is, if
we set a trial for April 16th, which is right now where I've
told the lawyers to be ready for, is the expert going to be
ready to go before that?
In other words, I would like to have his report done
and his deposition taken, and all of that, before the trial
date starts.
MR. COOPER: Yes, Your Honor.
THE COURT: And is his report going to include a
critique of considerations that were used by both sides'
experts?
MR. COOPER: Yes.
110
THE COURT: All right. On the subject of the --
thank you. Have a seat.
On the subject of the Daubert, if anyone has one last
word you want to put in, now is the time to do it by way of a
parting last statement.
MR. NORTON: I have certainly endeavored to answer
all the Court's questions.
If there is anything that the Court remains uncertain
about with respect to what Professor Cockburn did, or his
reasons for why he did them, he is, of course, here today and
is prepared to testify and answer questions directly from the
Court or from counsel about why he used the studies, why the
studies are appropriate, or anything else that may be of
concern to the Court.
THE COURT: Let me ask Google, do you want to take
this opportunity to cross-examine Dr. Cockburn? I'll put him
on the stand and let you have a free shot at him, if that's
what you want.
MR. PURCELL: No, Your Honor.
THE COURT: All right. Then I'm going to pass on
the -- I'm not going to ask him questions.
MR. NORTON: Thank you, Your Honor.
THE COURT: Okay. Do you have anything on your side?
MR. PURCELL: No, Your Honor. That's all.
THE COURT: All right. We will adjourn for now.
111
And, in due course, I will get an order out.
I want our 706 expert to be prepared to do your
report quickly. But I want it to be, of course, your best shot
at it. This is now early March. It may be that you'll be --
by the end of this month, you may be in depositions defending
your own work.
All right. We're adjourned. Thank you.
(Counsel thank the Court.)
(At 10:24 a.m. the proceedings were adjourned.)
- - - -
CERTIFICATE OF REPORTER
I certify that the foregoing is a correct transcript
from the record of proceedings in the above-entitled matter.
DATE: Wednesday, March 14, 2012
s/b Katherine Powell Sullivan
_________________________________
Katherine Powell Sullivan, CSR #5812, RPR, CRR
112
|