Disney Enterprises, Inc. et al v. Hotfile Corp. et al

Filing 426

NOTICE by Columbia Pictures Industries, Inc., Disney Enterprises, Inc., Twentieth Century Fox Film Corporation, Universal City Studios Productions LLLP, Warner Bros. Entertainment Inc. re 425 Reply to Response to Motion,, >>PLAINTIFFS' NOTICE OF FILING PUBLIC REDACTED VERSIONS OF REPLY DECLARATIONS IN SUPPORT OF PLAINTIFFS' MOTION FOR SUMMARY JUDGMENT<< (Attachments: # 1 Affidavit Reply Dweclaration of Jennifer V. Yeh in Support of Plaintiffs' Motion for Summary Judgment (public redacted version), # 2 Exhibit 151 to Declaration of J. Yeh, # 3 Exhibit 152 to declaration of J. Yeh, # 4 Exhibit 153 to Declaration of J. Yeh, # 5 Exhibit 154 to declaration of J. Yeh, # 6 Exhibit 155 to Declaration of J. Yeh, # 7 Exhibit 156 to declaration of J. Yeh, # 8 Affidavit Reply Declaration of Dr. Ian Foster in Support of Plaintiffs' Motion for Summary Judgment (public redacted version), # 9 Exhibit A to Declaration of I. Foster, # 10 Exhibit B to Declaration of I. Foster, # 11 Affidavit Reply Declaration of Dr. Richard Waterman in Support of Plaintiffs' MSJ (public redacted version))(Stetson, Karen)

UNITED STATES DISTRICT COURT
SOUTHERN DISTRICT OF FLORIDA
CASE NO. 11-20427-WILLIAMS/TURNOFF

DISNEY ENTERPRISES, INC.,
TWENTIETH CENTURY FOX FILM CORPORATION,
UNIVERSAL CITY STUDIOS PRODUCTIONS LLLP,
COLUMBIA PICTURES INDUSTRIES, INC., and
WARNER BROS. ENTERTAINMENT INC.,
Plaintiffs,
v.
HOTFILE CORP., ANTON TITOV, and
DOES 1-10.
Defendants.
/
HOTFILE CORP.,
Counterclaimant,
v.
WARNER BROS. ENTERTAINMENT INC.,
Counterdefendant.
/
REPLY DECLARATION OF DR. RICHARD WATERMAN IN SUPPORT OF
PLAINTIFFS’ MOTION FOR SUMMARY JUDGMENT AGAINST DEFENDANTS
HOTFILE CORP. AND ANTON TITOV

PUBLIC VERSION

I, Richard Waterman, hereby declare as follows:
1.

I am an Adjunct Professor of Statistics at The Wharton School at the University of

Pennsylvania, and the President and Co-Founder of Analytic Business Services, Inc., a
consultancy focused on providing expert advice and opinions in the field of statistical analysis. I
have been retained by the plaintiffs Disney Enterprises, Inc., Twentieth Century Fox Film
Corporation, Universal City Studios Productions LLLP, Columbia Pictures Industries, Inc., and
Warner Bros. Entertainment Inc. ("plaintiffs") to provide analysis related to the above-captioned
case. The statements made in this declaration are based on my personal knowledge or
application of my specialized knowledge to facts or data of which I am aware. If called to
testify, I would testify based on the best of my knowledge, information, and belief, as follows:
2.

I received my Ph.D. in Statistics from the Pennsylvania State University in 1993.

I have substantial experience designing and reviewing sampling protocols for various large
organizations, such as the United States Postal Service, for whom I designed and analyzed a
national multi-stage sample for the estimation of operational characteristics. Further explanation
of my experience, qualifications, publications, previous testimony, and compensation is included
in the declaration I previously submitted to the Court in support of plaintiffs' motion for
summary judgment, filed February 17, 2012.
3.

I have reviewed the Declaration of Professor James Boyle in Support of

Defendants' Opposition to Plaintiffs' Motion for Summary Judgment and its attached Exhibit 2,
which is a copy of his Rebuttal Report dated January 6, 2012. In paragraph 53 of his rebuttal
report, he calculates certain statistics based on the results of the statistical analysis that I designed
and about which I testified in my previous declaration. According to his report, he claims to
have added up the total number of "paidfor" counts and total number of downloads (from

"dailydownload") for each file within certain of Mr. Zebrak's categories - "Confirmed
Infringing," "Highly Likely Infringing," "Noninfringing," and "Unknowable." He then
calculated a purported "conversion rate" for each category, consisting of the sum of the "paidfor"
counts (which reflect users "converting" to Premium status) over the sum of downloads.
4.

The resulting statistics - the "conversion rates" for each category - do not

statistically support the conclusions that Prof. Boyle makes about non-infringing files on Hotfile
being more likely to convert users to Premium than infringing files. First, Prof. Boyle calculated
these quantities without consideration of the underlying sampling design and they are therefore
incorrect. Properly calculated, the data does not show any statistically significant difference
between the conversion rate for infringing and non-infringing files. Second, Prof. Boyle did not
calculate the margin of error on the conversion rate for the non-infringing files. However,
because the sub-sample of non-infringing files in the study is so small, and the characteristics of
the files in that sub-sample are skewed, the margin of error on the conversion rate for noninfringing files is relatively high.
5.

First, Prof. Boyle attempts to reach conclusions about the characteristics of

certain files on Hotfile rather than the characteristics of certain downloads. In this case, I
developed a protocol in order to select a sample to draw conclusions about overall daily
downloads on Hotfile (whether the downloads were infringing). However, Prof. Boyle repurposes that data and draws conclusions about categories of different files (the "conversion
rate"). The protocol that was implemented in this case was not designed, without a suitable
weighting adjustment, for that purpose. In order to draw general conclusion about files on
Hotfile from the sample he selected, he must perform additional calculations and analyses. It is

not statistically valid to extrapolate the "conversion rates" he calculated to apply to files on
Hotfile in general.
6.

If Prof. Boyle had attempted to weight the sample to adjust for the probability that

a file was selected into the sample, and had properly taken into account the statistical uncertainty
in the calculations, he would have found that there was no statistically significant difference
between the conversion rates on the infringing and non-infringing categories. In order to weight
the sample correctly, he would have needed to take into account the number of downloads of a
file from the date in which the file was selected, so that the calculations reflected the probability
of the file being selected on that day. Performing that calculation produces conversion rates for
the infringing and non-infringing categories that are different from those reported by Prof. Boyle
but not statistically significantly different from each other.1 In other words, the observed
differences in conversion rates are accounted for by random variation in the data, and from a
statistical point of view the rates are indistinguishable.
7.

Second, independent of whether he used the right calculations, Prof. Boyle did not

calculate the margin of error on the conversion rate for the non-infringing files. However, when
Prof. Boyle seeks to examine the characteristics of only the non-infringing files, he is examining
a sample size of 87 files. The small size of this sub-sample contributes to a relatively large
margin of error on the conversion rate for non-infringing files. Further, even a cursory look at
the sub-sample of non-infringing files shows that three specific files are extreme outliers in terms
of their popularity in downloads and conversions - iREB, snOwbreeze, and Jdownloader. As a
result, the characteristics of the files in the sub-sample are highly skewed. This also contributes
to the relatively large margin of error. Overall, the margin of error tells us the precision of the

This is determined using a 0.05 level of significance, which is equivalent to 95% confidence.

estimate that is calculated. The true conversion rate could be (with 95% confidence) within the
range of the estimated rate, plus or minus the margin of error.
8.

If Prof. Boyle had properly calculated the conversion rate as I explain above in

paragraph 6, the margin of error would be more than 57% of the value of the "conversion rate"
estimate itself.3 This indicates that the calculated "conversion rate" lacks precision. Most
sampling protocols in the real world would be designed to produce a substantially lower
coefficient of variation (a statistic calculated from the margin of error) than the very large
coefficient obtained on the conversion rate for non-infringing files. Further, even looking at the
flawed methodology that Prof. Boyle used to calculate the conversion rate for non-infringing
files in his rebuttal report, the margin of error on his conversion rate is approximately 33% of the
estimated value. Properly calculated, the margin of error shows that there is no statistically
significant difference between the conversion rate on infringing and non-infringing files.
9.

Finally, I also note that if one looks at the number of premium conversions per

category of infringing versus non-infringing content in Prof. Boyle's sample described in
paragraph 53 of his rebuttal report, there are many more conversions on infringing rather than
non-infringing files as a category. Moreover, as I noted above, three specific files are extreme
outliers in terms of their popularity in downloads and conversions. Even if the calculated
conversion rate had any statistical significance (and it does not), a reasonable conclusion would
be that these files, not non-infringing files generally, drive a higher rate of conversions.
2

The 95% refers to the repeated application of the sampling design, in the sense that if the entire
sampling design were repeated many times, then one would expect that 95% of the confidence
intervals to cover the true but unknown population level conversion rate.

3

1 used the widely accepted and reliable "bootstrap" technique to calculate the margin of error.
See Bradley Efron, (1987), The Jackknife, the Bootstrap, and Other Resampling Plans.
4

A typical goal for coefficient of variation would be of the order of 5%, whereas the coefficient
of variation here was approximately 28.5%, which is substantially larger.

I declare under penalty of perjury that the foregoing is true and correct.

Executed on March// 2012, at Philadelphia, PA.

Dr. Richard Waterman