Apple Inc. v. Samsung Electronics Co. Ltd. et al

Filing 2130

Declaration Reply Declaration Of John R. Hauser In Support Of Apples Motion For A Permanent Injunction filed byApple Inc.. (Attachments: # 1 Exhibit A, # 2 Exhibit B, # 3 Exhibit C, # 4 Exhibit D, # 5 Exhibit E, # 6 Exhibit F, # 7 Exhibit G)(Jacobs, Michael) (Filed on 11/9/2012)

Exhibit D

Highly Confidential - Attorneys' Eyes Only

Page 1
1

2
3
4

UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
SAN JOSE DIVISION
--------------------------------x
APPLE INC., a California
corporation,

5

Plaintiff,
6

Case No.
11-CV-01846-LHK

vs.
7

8

9

10

SAMSUNG ELECTRONICS CO., LTD, a
Korean business entity; SAMSUNG
ELECTRONICS AMERICA, INC., a New
York corporation; SAMSUNG
TELECOMMUNICATIONS AMERICA, LLC,
a Delaware limited liability
company,

11

Defendants.
12

--------------------------------x
13
14
15
16
17
18
19
20
21
22
23
24

HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
VIDEOTAPED DEPOSITION OF JOHN HAUSER, a
witness called by the Defendants, taken
pursuant to the applicable provisions of the
Federal Rules of Civil Procedure, before James
A. Scally, RMR, CRR, a Notary Public in and
for the Commonwealth of Massachusetts, at the
offices of WilmerHale, 60 State Street,
Boston, Massachusetts, on Friday, April 27,
2012, commencing at 9:31 a.m.
TSG Job # 48803

25

TSG Reporting - Worldwide

877-702-9580

Highly Confidential - Attorneys' Eyes Only

Page 233
1

So normally when you hear these surveys on

16:47:09

2

television, they say, you know, 55 percent for candidate A,

16:47:13

3

45 percent for candidate B, and then they give a plus or

16:47:22

4

minus.

That plus or minus is actually two times the

16:47:27

5

standard deviation -- roughly two times the standard error.

16:47:30

6

So that's that sort of the precision with which I'm -- I'm

16:47:34

7

estimating this parameter.

16:47:37

8

candidate, it's, you know, the votes they're going to get

16:47:42

9

or the people favoring them or whatever.

16:47:45

10

Okay.

In the case of the political

In my estimates,

it's the estimate of a particular partworth.

11

16:47:50

Now, just, again, for the record, kind of mixing a

16:47:55

12

little bit of philosophy here between frequentist and

16:48:01

13

Bayesian statistics, in terms of the Bayesian, it's the

16:48:06

14

posterior Bayesian confidence interval; they've got fancy

16:48:10

15

words, but it's usually best to think of it as the

16:48:15

16

confidence interval.

16:48:17

17

my population, I can give you both the estimate and in

16:48:21

18

Exhibit -- I thought it was here.

16:48:28

19

So for every one of the parameters in

MR. ILLOVSKY:

20

12, K in your report?

21

THE WITNESS:

22

MR. ILLOVSKY:

23

THE WITNESS:

24
25

Are you looking for

A.

16:49:17
16:49:18

Yeah.

I can't find 12.

Use that.

16:49:21

Okay.

16:49:22

Exhibit 12, which is K, you can see the standard

error of the market level mean.

TSG Reporting - Worldwide

16:49:20

16:49:23
16:49:27

877-702-9580

Highly Confidential - Attorneys' Eyes Only

Page 234
1

So to interpret this, they say table K1.

2

the first feature, which is "Reliable Touch, Auto-Switch (1

16:49:38

3

to 2 Fingers, Rubberband, Tap and Re-center the Zoom," the

16:49:42

4

average market level mean is 64.5, and we also have medians

16:49:45

5

in there, and I'll explain the difference in a moment, and

16:49:51

6

the standard error is 2.1.

16:49:54

7

that's pretty good precision.

8
9

Q.

We have

So you can see it's, you know,

16:49:30

16:49:58

Does this mean that the willingness to pay for any

given individual is measured with high precision?

16:50:00
16:50:05

10

A.

Oh, no.

11

Q.

Okay.

16:50:13

12

A.

You know, and I actually give an example of

16:50:25

Absolutely not.

16:50:10

13

flipping coins to try and motivate this concept.

You can

16:50:27

14

have high precision at the population level, but not high

16:50:31

15

precision at the level of each and every individual.

16:50:36

16
17

Q.

Okay.

Let's go to page 48 of Exhibit 1.

Okay.

So do you see -- do you see table 3A in your report?

16:50:41
16:51:00

18

A.

Yes, I do.

16:51:05

19

Q.

And then if you flip the page, you'll see table

16:51:06

20

3B?

16:51:08

21

A.

Yes, I do.

16:51:09

22

Q.

Okay.

16:51:10

23
24
25

So these are the -- here's the -- the means

and the standard deviations; right?
A.

No.

16:51:15

Let's be clear what we have here.

Okay.

There is an estimate of the mean and the market

TSG Reporting - Worldwide

877-702-9580

16:51:17
16:51:21

Highly Confidential - Attorneys' Eyes Only

Page 235
1

heterogeneity, which is the distribution, standard

16:51:24

2

distribution, of these partworths across the population.

16:51:27

3
4
5
6
7
8
9
10

Q.

This is the population of the respondents, or is

this the population of -- of the 10,000?
A.

16:51:36

I -- I have -- I mean I think you're confusing

draw as from a posterior distribution with the population.
Q.

What are you -- okay.

So the pop -- what do you

mean by the population?
A.

Okay.

16:51:32

16:51:45
16:51:49
16:51:56
16:51:58

So I have four hundred and -- this is

smartphones -- 455 respondents.

16:51:59
16:52:02

11

Q.

Uh-huh.

16:52:05

12

A.

Okay.

16:52:06

And what I'm doing is I'm estimating for

13

the target population, now, the mean level of the partworth

16:52:12

14

and how much it varies.

16:52:18

15

different partworth than I do, and Dr. Sukumar may have yet

16:52:24

16

another partworth, and Eugene may have a different

16:52:28

17

partworth, et cetera.

I can't -- I can't actually -- I can

16:52:30

18

get an estimate of each one of those partworths, but -- but

16:52:34

19

not with a lot of high precision.

But I can estimate how

16:52:37

20

these vary across the population.

And I can do that with

16:52:40

21

high precision.

22

So think of it as you may have a

16:52:44

So it's kind of like saying if I drive in the

16:52:47

23

Southeast Expressway, it's going to take me 20 minutes plus

16:52:52

24

or minus 10 minutes.

16:52:55

25

on average it takes me 20 minutes, of course that's

I can get a pretty good estimate that

TSG Reporting - Worldwide

877-702-9580

16:52:58

Highly Confidential - Attorneys' Eyes Only

Page 243
1
2
3

A.

Yes.

You shouldn't -- you should not use ACBC if

you have less than five attributes.
Q.

That's what it means.

17:04:03
17:04:05

It doesn't -- it doesn't -- it doesn't recommend

17:04:11

4

using -- so the way you read this is not a recommendation

17:04:14

5

that you use ACBC when you have more than five attributes;

17:04:17

6

you don't read it that way?

17:04:21

7

A.

Did you read the technical manual?

17:04:22

8

Q.

I did not.

17:04:24

9

A.

Yes, I did.

17:04:25

10

Q.

And what does it say?

17:04:26

11

A.

Okay.

17:04:27

12

Q.

No, that's not -- that's not the question.

Did you?

I'm quite happy to explain ACBC to you.
You

17:04:31

13

know, I've been patient today with -- with the responses

17:04:33

14

that are really nonresponsive to a lot of the questions.

17:04:38

15

But we're running out of time, and I am just asking you is

17:04:41

16

your interpretation of a printout from Sawtooth's website

17:04:49

17

where it says "When to Use ACBC" and it indicates "Five or

17:04:53

18

more attributes," it is -- it's your understanding that

17:04:57

19

ACBC should not be used when you have fewer than five

17:05:05

20

attributes; is that your understanding?

17:05:09

21

MR. ILLOVSKY:

Wait.

So hold on.

17:05:11

22

I've got to object to the preface.

23

answers have been perfectly responsive when

17:05:14

24

the questions have been comprehensible,

17:05:16

25

which has not been a large percentage of

17:05:18

TSG Reporting - Worldwide

The

877-702-9580

17:05:12

Highly Confidential - Attorneys' Eyes Only

Page 244
1

the time.

2

to the question.

3

So object to the preface, object

17:05:20
17:05:23

Go ahead.

17:05:26

4

A.

Can you ask -- re-ask the question?

17:05:28

5

Q.

Let's move on.

Look at the -- look at the last

17:05:39

6

paragraph on page 1 of what's been marked as Exhibit 20.

17:05:41

7

It says, "In addition to the standard partworth utilities

17:05:45

8

that are useful for segmentation and market simulation, we

17:05:49

9

captured the specific 'must-have' and 'unacceptable' rules

17:05:51

that respondents expressed during the screening process."

17:05:55

10
11
12

You're familiar with that term "must-have"; right?
A.

17:06:00

Well, if you had read the technical manual, you'll

17:06:03

13

see that they're pretty widely quoting many of my papers.

17:06:05

14

So, yes, I am definitely familiar with this.

17:06:12

15

Q.

Okay.

17:06:14

16

A.

And I'd like to also point out that we did these

17:06:14

17

tests on our data.

18

lexicography.

19

lexicographic when there are a lot of features or when the

17:06:25

20

choice task is speeded up or other things.

And we did do

17:06:29

21

lexicography tests in our data, which are entirely doable

17:06:32

22

with all the information we gave you.

17:06:37

23

And what these refer to is

And they're -- in many cases consumers are

And I can tell you that out of seven features

17:06:17
17:06:20

17:06:39

24

times roughly, you know, 800 respondents, it's like 5600

17:06:44

25

possible tasks, exactly one was lexicographic, and it was

17:06:53

TSG Reporting - Worldwide

877-702-9580

Highly Confidential - Attorneys' Eyes Only

Page 245
1

lexicographic on price.

2

whether or not we should have must-have features in there,

17:07:02

3

which can be done after the fact, and it fully essentially

17:07:05

4

confirms that there is no lexicography; because there's no

17:07:10

5

lexicography, we don't need ACBC.

17:07:13

6

So we did do the test as to

17:06:59

Now, of course, I only did this after I had the

17:07:16

7

data.

8

experience, from also the qualitative interviews, that we

17:07:23

9

probably did not need to worry about lexicography in this

17:07:26

10

particular study.

17:07:31

11

to have one out of like 5600, that can almost even be by

17:07:35

12

chance.

17:07:39

13
14
15

Q.

So I made a judgment up front, again, from my

17:07:20

And that turned out to be correct.

So

Why didn't you -- why didn't you mention in your

report that you tested for must-have features?
A.

17:07:40
17:07:43

Oh, I -- I only did this test after I read Dr.

17:07:47

16

Sukumar's results, because I didn't -- I mean we did not

17:07:51

17

have any indications that there should be any lexicography.

17:07:55

18

Also, you know, I know Sawtooth says "five or more" here,

17:07:59

19

but you're really -- it's going to be very rare that you

17:08:03

20

have lexicography for as little as five attributes.

17:08:07

21

going to be upwards of 20 or so before you start having it.

17:08:10

22

I did not expect it.

17:08:13

23

Sukumar raised it, and given that I could test it with the

17:08:17

24

data that you had been given, I tested it.

17:08:21

25

Q.

It's

But given that he raised it, Dr.

Part of your opinion is -- is the value placed on

TSG Reporting - Worldwide

877-702-9580

17:08:28

Highly Confidential - Attorneys' Eyes Only

Page 246
1

the '915, the '161, and the '381 patents combined.

2

able to break out the individual value of those patents?

3
4

A.

Let me just add that you've been given all

the code that we ran for the lexicographic tests.

5
6

Okay.

Are you

17:08:32
17:08:39
17:08:42
17:08:44

So can you -- can you re-ask that question?
Q.

17:08:49

Part of your opinion is the value placed on the

17:08:52

7

'915, the '161, and the '381 patents combined.

8

able to break out the individual value of those patents to

17:09:01

9

Samsung consumers?

17:09:04

10

A.

Okay.

11
12

So --

17:08:55

17:09:05

MR. ILLOVSKY:
A.

Are you

Objection to form.

Take a look at the report again.

17:09:05
17:09:07

13

MR. GALVIN:

14

Let's take a break, please.

17:09:45

15

THE VIDEOGRAPHER:

17:09:46

16

record.

I withdraw the question.

Going off the

The time is 5:09.

17:09:23

17:09:47

17

(Recess.)

17:09:49

18

(Exhibit 21, DVD labeled "Hauser

17:20:54

19

Survey Data Files," marked.)

17:20:58

20

THE VIDEOGRAPHER:

17:20:59

21
22
23
24
25

record.

We're back on the

The time is 5:20.

17:21:00

BY MR. GALVIN:
Q.

17:21:11

Dr. Hauser, for each of your respondents, how many

records are there?
A.

17:21:12
17:21:15

I am not sure I understand the --

TSG Reporting - Worldwide

17:21:21

877-702-9580

Highly Confidential - Attorneys' Eyes Only

Page 253
1

at the willingness to pay estimates at some point?

2

individual level, did you examine all --

3

A.

On an

17:30:13
17:30:22

I, in fact, not only didn't I do it, but I -- I

17:30:25

4

gave the example with the head -- with the coin-flipping

17:30:28

5

example, and, again, at least nine places in the report, it

17:30:30

6

was in two of the footnotes, I'm very explicit as to why

17:30:35

7

that would -- that is not what one should do, and that that

17:30:38

8

would be a naive thing to do.

17:30:42

9

with hierarchical Bayes to know these issues are in there.

17:30:46

10

And, you know, when I read Dr. Sukumar's results,

17:30:50

11

and he's getting numbers like he modified the code, and he

17:30:52

12

got numbers that are absurd.

17:30:55

13

and say, well, gee, they are absurd.

14

absurd?

I've got enough experience

And you should look at those
So why are they

17:30:59

Well, because the code was modified.

17:31:03

15

Q.

Well, did you get results that were absurd?

17:31:05

16

A.

No.

17:31:07

17

Q.

But you didn't look on an individual level at

18

the -- at the calculations, did you?

19
20

I did the calculations correctly.

MR. ILLOVSKY:
A.

17:31:09
17:31:12

Objection to form.

17:31:16

You know, I -- I wrote a set of procedures; I

17:31:17

21

wrote the set of procedures that are correct.

22

talking about 10,000 draws for 800 consumers times -- four

17:31:23

23

times -- times 28 partworths.

17:31:31

24

billions?

25

numbers.

So what is that?

We're now

Millions,

No, I didn't look at a billion different
I looked at the appropriate output of

TSG Reporting - Worldwide

877-702-9580

17:31:20

17:31:35
17:31:38

Highly Confidential - Attorneys' Eyes Only

Page 254
1

calculations based on whatever it is, a billion different

17:31:40

2

numbers.

17:31:43

3

Q.

Did you look at the median of those calculations?

17:31:44

4

A.

I looked at -- I did the median appropriately.

17:31:48

5

Q.

So what was the -- what was the median?

17:31:51

6

A.

The median, as described, it's -- we can look it

17:31:55

7

up, you know.

17:31:58

8

Q.

Do you remember?

17:31:59

9

A.

Is it a memory test?

17:32:00

10

Q.

No.

17:32:02

11

It's -- I'm asking you if you remember or

not.

17:32:04

12

A.

Do I remember the exact number?

17:32:05

13

Q.

Uh-huh.

17:32:06

14

A.

The exact number's in my report.

15

We can look it

up.

17:32:07
17:32:10

16

Q.

Okay.

17

A.

(Pause.)

Let's go -- let's go there.

17:32:10

Well, it's basically footnote 72, 73,

17:33:15

18

and you'll note that in 72 it says, "For each of these

17:33:22

19

samples, I computed a median willingness to pay for the

17:33:30

20

market.

17:33:34

21

willingness to pay by taking the median of the 10,000

17:33:37

22

sample medians," okay?

17:33:41

23

in the earlier coin-flipping examples, reporting a

17:33:44

24

willingness to pay for an individual respondent would not

17:33:47

25

be sufficiently precise; however, the overall market level

17:33:50

I then computed an overall market level

And then I cautioned, "As explained

TSG Reporting - Worldwide

877-702-9580

Highly Confidential - Attorneys' Eyes Only

Page 255
1

willingness to pay is sufficiently precise."

2

definitely cautioned that.

3

So I

17:33:54
17:33:58

Now, in -- in paragraph 73, doing those median

17:34:00

4

calculations, so doing the medians within the sampler,

17:34:04

5

okay, so in the -- getting the posterior distribution of

17:34:08

6

the medians, I then, having gotten that posterior

17:34:11

7

distribution of the medians, we can now say something like

17:34:17

8

the willingness to pay estimates at a base price of 199,

17:34:20

9

customers would be willing to pay $40 more for a smartphone

17:34:25

10

that has the functionality associated with patent '915.

17:34:28

11

And then it goes on from there.

17:34:32

12

And you'll note that how this, then, is used is up

17:34:35

13

in paragraph 104, and it says, "The median willing --

17:34:39

14

consumer willingness to pay calculation leads price premium

17:34:44

15

estimates that are similar to what I estimate using the

17:34:48

16

market simulation."

17:34:51

17

check.

18

Q.

So I'm using it for a convergent

17:34:54
So if we look at -- if we look at any one

17:34:57

19

respondent's draws, that doesn't really -- that doesn't

17:35:11

20

really tell us their willingness to pay for -- for any of

17:35:17

21

the features?

17:35:20

22

A.

Again, let's go back to the coin-flipping example.

17:35:23

23

If I end up with two heads, my estimate, you know, for that

17:35:26

24

particular respondent, you know, in fact, my maximum

17:35:31

25

likelihood estimate is 100 percent.

17:35:35

You know, so I'm not

TSG Reporting - Worldwide

877-702-9580

Highly Confidential - Attorneys' Eyes Only

Page 256
1

going to be very precise.

17:35:38

2

Q.

Okay.

17:35:40

3

A.

So in -- when I look at any individual, I have 48

17:35:41

4

constraints, plus I have a number of monotonicity

17:35:47

5

constraints, and I think if we count up the number of

17:35:53

6

features, it's something like 7 times 3.

17:35:55

7

because there's -- not everything's monotone.

8

know, I basically have, what, maybe 60 constraints for 20,

17:36:04

9

21 features.

17:36:10

Not completely,
So, you

I can't expect that to be precise.

17:36:00

However,

10

when I get up to 20,000 constraints, which is what I have

17:36:13

11

for the population, I can expect that to be fairly precise.

17:36:18

12

So, no, you should not look at it at the

17:36:23

13

individual level, and as in the documents that you don't

17:36:26

14

like to refer to, you say, well, some of these appear to be

17:36:29

15

negative, again, just doing the arithmetic calculation,

17:36:32

16

almost none of those are sufficiently precise to -- to say

17:36:36

17

they're negative.

17:36:40

18

What we can say is that for roughly -- and also

17:36:42

19

they're conflated.

20

and say that for 94 percent of the people, they have

17:36:49

21

positive partworths for one of the patents.

17:36:52

22

want to conflate it either way.

23

We can look at it the other way around

But I don't

17:36:55

The key thing is none of those are significant --

24

I think one of those are significant out of all those

25

tests.

And there are people who -- who don't value.

TSG Reporting - Worldwide

17:36:45

877-702-9580

17:36:59
17:37:02

I

17:37:04