Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 1 of 94
`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 1 of 94
`
`
`
`
`
`
`EXHIBIT A
`EXHIBIT A
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 2 of 94
`
`UNITED STATES DISTRICT COURT
`SOUTHERN DISTRICT OF NEW YORK
`
` Case No. 14-cv-2396
`
`Case No. 14-cv-9558
`
`
`
`
`
`NETWORK-1 TECHNOLOGIES, INC.,
`
`
`Plaintiff,
`
`
`
`vs.
`
`
`GOOGLE LLC and YOUTUBE, LLC,
`
`
`Defendants.
`
`
`
`
`
`
`SUPPLEMENTAL EXPERT REPORT OF MICHAEL MITZENMACHER, PH.D.
`
`
`
`
`
`
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 3 of 94
`
`CONFIDENTIAL OUTSIDE COUNSEL ONLY –
`PROSECUTION/ACQUISITION BAR MATERIALS
`
`
`
`
`
`
`
`
`
`TABLE OF CONTENTS
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Page
`
`1. INTRODUCTION .....................................................................................................................1
`
`1.1. Retention ......................................................................................................................1
`
`1.2. Qualifications ...............................................................................................................1
`
`1.3. The Asserted Patents ....................................................................................................2
`
`1.4. Materials Considered ....................................................................................................3
`
`1.5. Legal Principles ............................................................................................................3
`
`1.6. Level of Ordinary Skill .................................................................................................3
`
`2. SUPPLEMENTAL OPINIONS CONCERNING DEFENDANTS’ INFRINGEMENT BY
`THE SIBERIA VERSION OF THE CONTENT ID ACCUSED INSTRUMENTALITIES..........3
`
`3. SUPPLEMENTAL OPINIONS ON THE ALLEGED NON-INFRINGING
`“OFFSHORING” ALTERNATIVE ................................................................................................6
`
`3.1. Defendants Have Not Shown that “Offshoring” Was a Viable Non-Infringing
`Alternative in the Relevant Timeframes.......................................................................6
`
`3.2. Defendants “Offshoring” Alternative Is Not Acceptable, Is Not Available, and Is
`Not Non-Infringing Because It Infringes Other Network-1 Patents ..........................11
`
`3.2.1. ’216 patent claim 1....................................................................................... 13
`
`3.2.1.1. ’216 patent claim 1 preamble ...................................................... 13
`
`3.2.1.2. ’216 patent claim 1(a) .................................................................. 13
`
`3.2.1.3. ’216 patent claim 1(b) ................................................................. 14
`
`3.2.1.4. ’216 patent claim 1(c)(1)(i) ......................................................... 14
`
`3.2.1.5. ’216 patent claim 1(c)(1)(ii) ........................................................ 28
`
`3.2.1.6. ’216 patent claim 1(c)(2) ............................................................. 29
`
`3.2.1.7. ’216 patent claim 1(c)(3) ............................................................. 31
`
`3.2.1.8. ’216 patent claim 1(c)(4) ............................................................. 67
`
`3.2.1.9. ’216 patent claim 1(c)(5) ............................................................. 78
`
`4. CONCLUSION ........................................................................................................................79
`
`
`
`i
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 4 of 94
`
`CONFIDENTIAL OUTSIDE COUNSEL ONLY –
`PROSECUTION/ACQUISITION BAR MATERIALS
`
`1.
`
`INTRODUCTION
`
`1.1. Retention
`
`1.
`I have been retained as an independent expert witness by the law firm of Russ August &
`Kabat on behalf of Network-1 Technologies, Inc. to testify as a technical expert in the
`following lawsuits concerning U.S. Patent Nos. 8,010,988 (“the ’988 patent”); 8,205,237 (“the
`’237 patent”); and 8,904,464 (“the ’464 patent”) (collectively, the “Asserted Patents”):
`
`Network-1 Technologies, Inc. v. Google LLC and YouTube, LLC, 14-cv-2396 (S.D.N.Y)
`Network-1 Technologies, Inc. v. Google LLC and YouTube, LLC, 14-cv-9558 (S.D.N.Y)
`
` I
`
` refer to Google LLC and YouTube, LLC as “Defendants” or “Google” in this report.
`
`
`2.
`In this expert report, I provide opinions regarding the Asserted Patents, and opinions
`relating to Defendants’ infringement of the currently asserted claims of the Asserted Patents. I
`expect to testify at trial on these issues, as set forth in this report, my prior December 20, 2019
`and February 14, 2020 reports in these cases, and in any other supplemental reports or
`declarations that I may prepare for this litigation in the future. I also expect to testify at trial
`with respect to the matters addressed by any expert testifying on behalf of Defendants, if asked
`about these matters by the Court or by the parties’ counsel. I may also testify on other matters
`relevant to this case, if asked by the Court or by the parties’ counsel.
`
`3.
`To ensure that my opinions are complete and accurate, I reserve the right to supplement
`or amend this report if additional facts and information that affect my opinions become
`available. Such information may include, for example, materials produced in this litigation, and
`information and documents relevant to this case that Defendants has not yet disclosed. I may
`also supplement or amend my report or opinions in response to additional discovery or other
`events, and may rebut expert reports submitted by Defendants.
`
`4. My work in this case is being billed at my standard rate of $850 per hour, with
`reimbursement for actual expenses. My payment is not contingent upon my testimony or the
`outcome of the case. I have no personal interest in the outcome of the case.
`
`1.2. Qualifications
`
`5. My Curriculum Vitae, attached as Exhibit B, is a true and accurate listing of my
`qualifications. I summarize some of these qualifications below.
`
`6.
`I am currently employed as a Professor of Computer Science at Harvard University.
`Specifically, I am the Thomas J. Watson, Sr. Professor of Computer Science in the School of
`Engineering and Applied Sciences. I joined the faculty of Harvard as an Assistant Professor in
`January 1999. I was promoted to Associate Professor in 2002 and to Professor in 2005. In
`2010, I began a three-year term as Area Dean, which is essentially equivalent to what other
`schools call Department Chair, of Computer Science, and held that position through June 2013.
`I served as Area Co-Chair of Computer Science for the 2018-2019 academic year. My work
`address is 150 Western Avenue, Sci&Eng 3.310, Boston, MA 02134. My primary research
`
`
`
`1
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 5 of 94
`
`CONFIDENTIAL OUTSIDE COUNSEL ONLY –
`PROSECUTION/ACQUISITION BAR MATERIALS
`
`interests include design and analysis of algorithms, networks and data transmission, and
`information theory.
`
`7.
`I received my undergraduate degree in Mathematics and Computer Science from Harvard
`College in 1991. I received a Certificate of Advanced Study in Mathematics from Cambridge
`University in 1992. I received a Ph.D. in Computer Science from the University of California
`at Berkeley in 1996. From August 1996 to January 1999, I was employed as a Research
`Scientist at Digital Systems Research Center, where my work included projects on algorithms
`for the Internet.
`
`8.
`I am listed as an inventor or co-inventor on 19 issued patents, and am the co-author of a
`textbook entitled “Probability and Computing” published by Cambridge University Press. I am
`a Fellow of the Association for Computing Machinery (ACM).
`
`9.
`I regularly serve on program committees for conferences in networking, algorithms, and
`communication. For example, I have served on the program committee multiple times for the
`SIGCOMM conference, which is the flagship annual conference of the ACM Special Interest
`Group on Data Communication (SIGCOMM). I have also served on numerous program
`committees related to algorithms, including the ACM Symposium on the Theory of
`Computing, the International Colloquium on Automata, Languages, and Programming, and the
`International Conference on Web Search and Data Mining.
`
`10. The field of endeavor at issue in this case is identification of electronic content (such as
`video or audio content) using algorithmic search techniques. I have published over 200
`research papers1 in computer science and engineering conferences and journals, many of which
`have explored algorithms and data structures for algorithmic search techniques, including both
`mathematical analysis and applications.
`
`1.3. The Asserted Patents
`
`11.
`I described the Asserted Patents in detail in Section 1.3 of my December 20, 2019 expert
`report (“Infringement Report”). That description is equally relevant here, and is incorporated
`by reference.
`
`12.
`
`I understand that the following are the Asserted Claims:
`
`• U.S. Pat. No. 8,010,988 (“the ’988 patent”), claim 17;
`
`• U.S. Pat. No. 8,205,237 (“the ’237 patent”), claims 33-35; and
`
`• U.S. Patent No. 8,904,464 (“the ’464 patent), claims 1, 8, 10, 16, 18, 25, 27, and 33.
`
`
`1 I note that in several comments in the source code Google produced in this case related to the
` I describe in detail below, there is reference to one of my publication on this
`topic. See, e.g., GOOG-NETWORK-SC-00000564; GOOG-NETWORK-SC-00000607.
`
`
`
`2
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 6 of 94
`
`CONFIDENTIAL OUTSIDE COUNSEL ONLY –
`PROSECUTION/ACQUISITION BAR MATERIALS
`
`1.4. Materials Considered
`
`13.
`In preparation for this report and for expert testimony that I may be called upon to
`provide, I have considered and may rely on documents identified in this report or those
`referenced in the exhibits attached to this report. This includes among other materials the
`Asserted Patents and their prosecution histories, Network-1’s infringement contentions, the
`Court’s claim construction, discovery and publicly available information regarding the
`patented subject matter and the accused systems, third-party information, deposition testimony
`and deposition exhibits, other discovery responses, my interaction with the accused
`instrumentalities, and all materials cited in my Infringement Report. In addition to the materials
`explicitly reference in my report, I have also considered the materials listed in Exhibit A to
`this report. My opinions are based on these sources of information, together with my education,
`training, and experience.
`
`14.
`In testifying, I may use some or all of the information referenced above, additional
`information identified in discovery, as well as any materials relied upon by Defendants’
`experts, to support or summarize my opinions. In addition, I may prepare summaries and
`demonstrative exhibits to assist my presentation of testimony to the Court.
`
`1.5. Legal Principles
`
`15.
`I addressed the legal principles relating to patent infringement and claim construction in
`detail in Section 1.5 of my Infringement Report. That description is incorporated here by
`reference.
`
`1.6. Level of Ordinary Skill
`
`16. As stated in my prior December 20, 2019 and February 14, 2020 expert reports, it is my
`understanding that the infringement analysis is to be undertaken from the perspective of a
`person of ordinary skill in the art to which the patents are directed at the time of the invention,
`here in 2000. The Asserted Patents are directed to the field of identification of electronic
`content (such as video or audio content) using algorithmic search techniques. In my opinion, a
`person of ordinary skill in this art would have a Bachelor’s degree in computer science,
`mathematics, or a similar discipline and two to three years of relevant experience, or a graduate
`degree in the same area.
`
`2.
`SUPPLEMENTAL OPINIONS CONCERNING DEFENDANTS’
`INFRINGEMENT BY THE SIBERIA VERSION OF THE CONTENT ID ACCUSED
`INSTRUMENTALITIES
`
`17.
`In my Infringement Report, I explained at length how and why the Content ID Siberia
`Version met the “sublinear” limitations of claim 17 of the ’988 patent and claim element b of
`claim 33 of the ’237 patent. See, e.g., Infringement Report ¶¶148-162, 229-240.
`
`18. For example, in my Infringement Report, I cited to a Google document listing numerous
`“tunable knobs” in the Siberia search algorithm. GOOG-NETWORK-00702308. One of the
` Id. I explained in detail in my Infringement
`“tunable knobs listed is
`
`
`
`3
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 7 of 94
`
`CONFIDENTIAL OUTSIDE COUNSEL ONLY –
`PROSECUTION/ACQUISITION BAR MATERIALS
`
` as the size of the data set size
`Report that increasing the
`increases results in sublinear scaling. See, e.g., Infringement Report ¶¶157-158.
`
`19. Another of the “tunable knobs” listed in the Google document I reference above is
` GOOG-NETWORK-00702308.
`
`
`
`20. As of the date of my Infringement Report, Content ID Siberia Version’s ScaM index
`lookup utilized a data structure with
` with each of
` See, e.g., Infringement Report ¶158. As I
`the
`previously explained, the ScaM index lookup first compares the unknown work to
`
`
`and then compares the unknown work to
`
`my Infringement Report, this given number
`
` See Infringement Report ¶158.
`
` As of the date of
`
`
`
`
`21.
`
`I understand that since the Infringement Report, Defendants have reduced the
`
` GOOGLE-NETWORK-00812409
`Konrad 2021 Depo. at 30:12-31:23 (“And the next bullet refers to searching --
`it looks like it originally said
`
`
`Do you see that? A. I can see that. . . . That talks about only
` when doing a search. Q. . . . Am I assuming correctly that these are changes that
`were implemented sometime, I think you said, in the first part of 2020? . . . I believe that these
`changes were implemented sometime in 2020, probably before August. Yes. . . . Q. Okay. And
`-- and I think you indicated -- and if we go back to the summary table – these are all changes
`that were intended to reduce the -- the resource usage of these different functionalities? A. Yes.
`These were changes making the resource costs lower.”).
`
`
`
`22. Reducing the
` to subject to further
`consideration) as the size of the data set increases results in sublinear scaling. As I explained in
`my Infringement Report, there are two main steps in ScaM’s index lookup, in terms of the
`computational work required: (1) comparing the unknown work to
` and (2)
`comparing the unknown work to
`
` Defendants have reduced the
`amount of work required for the second part by reducing the
`
`
`
`
` Reducing the
` as the size of the data set increases also results in a
`search that scales sublinearly with the size of the data set.
`
`23. As explained in my Infringement Report, the amount of work (W) can be represented by
`the following equation, where a and b are fixed constants, S is the
`, T is the
` and Q is the
`
` W = aP + bSTQ. As I also explained in my Infringement
`Report, Q can be represented as the total number of
` (R) divided by the
`product of the
` or Q = R/SP. And
`substitution for Q in the prior equation yields W = aP + bRT/P. Thus, the amount of work W
`done depends on the term RT, the product of R and T. In my Infringement Report, I explained
`
`
`
`4
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 8 of 94
`
`CONFIDENTIAL OUTSIDE COUNSEL ONLY –
`PROSECUTION/ACQUISITION BAR MATERIALS
`
`that this equation would lead one of skill in the art to understand that the work would be
`sublinear in R, since if T was held fixed one would choose P to be proportional to the square
`root of R in optimizing the work.
`
`24. Similar analysis applies if one allows T to vary instead. If the
` does not
`change, we can represent T by the equation T = S/(f(R)), where f(R) is some function of R. If T
`is reduced, but S and P remain constant (as is the case in the changes described by Mr. Konrad)
`as R increases, then f(R) is an increasing function of R. Substituting T in the equation above
`yields W = aP + bRS/(P*f(R)). The search scales sublinearly because the work grows
`proportionally to R/f(R), which means it is sublinear in R. As an example, if f(R) was equal to
`the function log R, then R/f(R) would be R/(log R), which is sublinear in R. As another
`example, if f(R) was equal to the function √𝑅, then R/f(R) would be √𝑅, which is sublinear in
`R. On the other hand, if the search were linear in R, the amount of work or computing power
`required would grow proportionally to R itself.
`
`25.
`I understand that in their summary judgment briefing Defendants argued that the Content
`ID Siberia Version does not meet the sublinear limitations because “the Siberia version
`searches a fixed percentage of the index” because “
`
`
`” Dkt. No. 224
`at 17. I also understand that Dr. Bhattacharjee made similar arguments in his rebuttal expert
`report dated February 14, 2020. See, e.g., Bhattacharjee Rebuttal Report ¶302. But this is not
`true. The
` is not fixed as Defendants and Dr. Bhattacharjee
`previously claimed, but rather it can be (and has been) altered. In other words, Defendants’
`own changes to Content ID Siberia Version reveal that the
` but rather a changeable parameter that they have in fact changed for
`performance reasons. See, e.g., Konrad 2021 Depo. at 30:12-31:23.
`
`
`
`26. Finally, I understand that Defendants argued in their summary judgment briefing that
`tuning Content ID Siberia Version as the size of the data set increased to avoid linear scaling
`was merely “a hypothetical change that Google might make to its system.” Dkt. No. 227 at 4.
`But the changes described by Mr. Konrad at his deposition make clear that such changes are
`not hypothetical, as Google did in fact take advantage of the Content ID system design to avoid
`linear scaling (by
`). This is
`further evidence that Google designed its search algorithm to be sublinear by allowing for
`precisely the types of adjustments described in its “tunable knobs” document (GOOG-
`NETWORK-00702308), described in my Infringement Report, and described by Mr. Konrad at
`his deposition. Indeed, Google’s own use of the Content ID Siberia Version demonstrates that
`Defendants’ argument in their summary judgment brief fails, as their statement that the
`
`
`
` As
`evidenced by Defendants’ documents, this is exactly the type of tuning contemplated by them.
`See, e.g., GOOG-NETWORK-00701295
`
`; see also GOOG-NETWORK-00704303.
`
`
`
`
`
`
`
`5
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 9 of 94
`
`CONFIDENTIAL OUTSIDE COUNSEL ONLY –
`PROSECUTION/ACQUISITION BAR MATERIALS
`
`3.
`SUPPLEMENTAL OPINIONS ON THE ALLEGED NON-INFRINGING
`“OFFSHORING” ALTERNATIVE
`
`3.1. Defendants Have Not Shown that “Offshoring” Was a Viable Non-Infringing
`Alternative in the Relevant Timeframes
`
`27. As explained in my Infringement Report, I understand that Defendants are contending
`that there are certain activities that would represent “non-infringing alternatives.” See
`Infringement Report, Section 5. I understand that a non-infringing alternative is a modification
`to an accused instrumentality that, if implemented, would render the accused product non-
`infringing. The first proposed alternative that I addressed in the Infringement Report is
`“geographically locating the servers running to Accused Instrumentalities, or a portion of the
`Accused Instrumentalities, outside the United States.” Id. ¶505 (quoting Defendants’ Third
`Supplemental Response to Network-1’s Interrogatory No. 13 (dated May 14, 2015)). I
`explained in my Infringement Report that I have not seen any evidence that Defendants had
`evaluated whether or not this proposed “alternative” would have been technically feasible or
`not, that Defendants had not quantified the amount of work this “alternative” would have
`required, and that Defendants had not evaluated how numerous other YouTube and Google
`functionalities that rely on Content ID to function would be impacted. See id.
`
`28.
`I further understand that in the 2020-2021 timeframe, Defendants “relocated the
`production instances of the Content ID Match System portion of the Accused Instrumentalities
`that had previously been located within the United States to servers exclusively outside the
`United States, primarily in Europe.” Defendants’ Fifth Supplemental Response to Network-1’s
`Interrogatory No. 13 (dated February 19, 2021). According to Defendants, “[t]he relocation
`was complete by no later than January 28, 2021.” Id. (“Defendants relocated to outside the
`United States all instances of
`
`
`
`
`
`
`
`); see
`also Malina Depo. at 84:16-100:8; GOOG-NETWORK-00811793; Ryjkov Depo. at 52:19-21,
`53:24-60:3, 63:5-9, 66:1-68:21. I will refer to this “offshoring” or “relocation” as the 2021
`Migration.
`
`29. As I explained in my Infringement Report, I have been informed that a non-infringing
`alternative must be feasible (i.e., technically feasible and economically feasible) and
`commercially acceptable. I also understand that feasibility and acceptability are evaluated at
`the time infringement began, which I understand in this case to be the issue date of each of the
`Asserted Patents. Thus, for the ’988 patent, the relevant date for this analysis is August 30,
`2011; for the ’237 patent, the relevant date is June 19, 2012; and for the ’464 patent, the
`relevant date is December 2, 2014.
`
`30.
`In my opinion, Defendants have not shown that the 2021 Migration would have been
`technically feasible, economically feasible, or commercially acceptable in any of the
`timeframes relevant here—2011, 2012, and 2014, or at any other time prior to 2021. In a
`general sense, as is true with almost all technology, showing that something can be done today
`
`
`
`6
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 10 of 94
`
`CONFIDENTIAL OUTSIDE COUNSEL ONLY –
`PROSECUTION/ACQUISITION BAR MATERIALS
`
`(or in 2021) does not automatically mean that the same thing could have been done a decade
`earlier. Numerous technological achievements that seem routine today were prohibitively
`difficult or expensive to implement just ten years ago.
`
`31. At the relevant times, Defendants were using a different version of Content ID than in
`2021. As explained in my Infringement Report, I understand that there have been two main
`versions of Content ID during the relevant time period—the version that used locality sensitive
`hashing (“LSH”) for indexing and searching (“Content ID LSH Version”) and the version that
`is known as “Siberia” (“Content ID Siberia Version”). In each of 2011, 2012, and 2014,
`Content ID LSH Version was being used, and in 2021, Content ID Siberia Version had
`replaced the Content ID LSH Version. See, e.g., Pasula 2019 Depo. at 97:13-18, 98:19-101:3;
`Konrad 2019 Depo. at 146:23-147:19, 166:13-168:8. As I describe in detail in my Infringement
`Report, there are a large number of technical differences between the two versions, including
`. See Infringement Report,
`Section 4. I have not seen any evidence that Defendants have evaluated whether the Content ID
`LSH Version (the version in place in the relevant timeframes) could have the MatchSystem
`portion of it migrated “offshore,” whether that migration was technically feasible in the
`relevant timeframes, what that migration would have cost in the relevant timeframes, and
`whether or not that migration would have resulted in a commercially acceptable version of
`Content ID.
`
`32.
`In addition to using an entirely different version of Content ID, in those relevant
`timeframes the
`the time of the 2021 Migration, the
`understand that there are certain technical aspects of
`
`, but at
`. I
`
`
`
`. Moor Depo. at
`123:13-125:15 (“Q. Have you considered whether it would be more difficult, or, you know,
`?
`less difficult to move
`A. . . . So when you say ‘difficulty’ here, I’m assuming you’re just talking about the bits that
`I’m responsible for, which is the provisioning bits, right, like the resource management aspect
`of it. And in that case, there are
`
`
`
`don’t have to request, for instance, for the engineers to move the data.
` And so in some aspects, it’s actually quite easier to do a migration
`
`) (emphasis added); see also Konrad 2021 Depo. at 35:17-37:21 (“Q. And do
`you know why YouTube, in general, was
`
`? A. I believe the main reason was that
`
`
`
` We
`
`centers with
`
` Q. . . . So there would be less operational costs in moving among data
`? A. So with
`
`
`
`
`
`
`
`); see also Malina Depo. at 18:12-20:9; Ryjkov Depo. at 58:21-59:20,
`66:7-68:21, 122:18-123:9.
`
`33. Setting aside the issues discussed above concerning a different version of Content ID and
`a different database management system being implemented, Defendants have also not shown
`that they had sufficient computing resources in any of the relevant timeframes such that the
`
`
`
`7
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 11 of 94
`
`CONFIDENTIAL OUTSIDE COUNSEL ONLY –
`PROSECUTION/ACQUISITION BAR MATERIALS
`
`2021 Migration would have been feasible at those times. In particular, I have reviewed a
`spreadsheet produced by Defendants with the Bates number GOOG-NETWORK-00812739 as
`well as testimony from Christine Moor concerning this spreadsheet (Moor Depo. at 54:3-87:9,
`88:11-20, 89:13-90:4, 91:8-92:7). This spreadsheet does not demonstrate that the 2021
`Migration would have been feasible in any of the 2011, 2012, or 2014 timeframes (or in any
`timeframe) because it does not contain any information concerning (1) how much additional
`computing and storage capacity outside the United States would have been required for
`migration of the MatchSystem outside the United States at any time or (2) how much excess
`computing or storage capacity was available at non-US clusters at any time. Here, excess
`capacity in non-US clusters is the relevant inquiry because some amount of additional
`computing and data storage would have been needed in those clusters to migrate the matching
`portion of Content ID to them.
`
`a. As a preliminary matter, this spreadsheet does not contain any data from the 2011
`timeframe at all.
`
`b.
`
` Tabs:
`
`i. Considering the tab entitled
`
` Id. at 49:14-50:4. I also understand that this tab “
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
` Id. at 57:25-58:19; see also
`
`generally id. at 57:5-68:10, 73:19-75:7.
`
`ii. I further understand the
`
`
` tabs contain
`similar information, but for various forms of memory (random access
`memory, persistent disk storage, and flash memory) rather than computing
`power. Id. at 76:24-78:10.
`
`iii. These four
` tabs do not provide any information
`about what excess computing or storage capacity that YouTube had in non-US
`clusters in 2011, 2012, or 2014 for at least four reasons: (1) there is no data
`from 2011; (2) there is no data from prior to 2015 for any non-US clusters; (3)
`the data on the tabs only reflect total allocated capacity for YouTube, not
`excess capacity that would be available for any migration of any portion of
`Content ID; and (4) the data on the tabs only include clusters that ran the
`match servers in 2021—the list of clusters running the match servers in 2012
`or 2014 (or any time other than when the data was pulled in 2021) would
`likely have been different. See id. at 68:18-69:25. There is likewise no data in
`any of these tabs about the amount of additional computing power or data
`
`
`
`8
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 12 of 94
`
`CONFIDENTIAL OUTSIDE COUNSEL ONLY –
`PROSECUTION/ACQUISITION BAR MATERIALS
`
`storage that would have been required in non-US clusters for a migration of
`the MatchSystem to non-US clusters in any timeframe.
`
`c.
`
` Tabs:
`
`i. Considering the tab entitled
` I understand
`that this tab “gives you YouTube’s total capacity in non-U.S. clusters . . . [i]n
`2012 and all dates through 2021.” Id. at 71:3-72:13 (emphasis added); see also
`generally id. at 72:14-73:18, 75:8-76:23, 85:15-87:1.
`
`ii. I further understand the
`
`
` tabs
`contain similar information, but for various forms of memory (random access
`memory, persistent disk storage, and flash memory) rather than computing
`power. Id. at 78:19-80:10, 85:15-87:1.
`
`iii. These four
` tabs do not provide any information
`about what excess computing or storage capacity that YouTube had in 2011,
`2012, or 2014 for two reasons: (1) there is no data from 2011; and (2) the data
`on the tabs only reflect total allocated capacity, not excess capacity that would
`be available for any migration. See id. at 89:18-90:4 (“Looking at the tabs that
`refer to
`
`
` A. Yes. Q. Is there any way to tell from the
`data in this sheet how much excess capacity there was in a given cell? . . .
`THE WITNESS: Not from the data in this sheet.”); see also id. at 91:8-92:7
`(“Q. If you were -- if you were going to perform that analysis, how would you
`do that, if you know? A. If I were going to perform that analysis for a given
`cluster, I would need to know both how much capacity is available and how
`much capacity is in use. . . . For all of the tabs entitled
` I
`don’t have all the information I need to determine how much excess capacity
`there is.”). There is likewise no data in any of these tabs about the amount of
`additional computing power or data storage that would have been required in
`non-US clusters for a migration of the MatchSystem to non-US clusters in any
`timeframe.
`
`d.
`
` Tabs:
`
`i. Considering the tabs entitled
`
` I understand that these tabs reflect the total capacity
`allocated to the MatchSystem at various points in time. Id. at 84:11-85:14; see
`also id. at 80:14-83:6.
`
`ii. These two
` tabs do not provide any information about what
`excess computing or storage capacity that YouTube had in 2011, 2012, or
`2014 for two reasons: (1) there is no data from 2011; and (2) the data on the
`tabs only reflect total allocated capacity for the MatchSystem, not excess
`capacity that would be available for any migration. There is likewise no data
`
`
`
`9
`
`

`

`Case 1:14-cv-02396-PGG-SN Document 269-1 Filed 09/07/22 Page 13 of 94
`
`CONFIDENTIAL OUTSIDE COUNSEL ONLY –
`PROSECUTION/ACQUISITION BAR MATERIALS
`
`in any of these tabs about the amount of additional computing power or data
`storage that would have been required in non-US clusters for a migration of
`the MatchSystem to non-US clusters in any timeframe.
`
`34. And more generally, I have not seen any evidence that Defendants have evaluated
`whether the Content ID system in 2011, 2012, or 2014 could have had the MatchSystem
`portion of it migrated “offshore,” whether that migration was technically feasible, what that
`migration would have cost, and whether or not that migration would have resulted in a
`commercially acceptable version of Content ID. In fact, Defendants’ witnesses confirmed at
`their depositions that Defendants have not conducted such an analysis. See, e.g., Malina Depo.
`at 124:25-125:10 (“Did you perform any assessment of what would have been required to
`migrate any aspect of Google’s Match system at any time other than 2020 to ’21? . . . THE
`WITNESS: No. Q. Did anyone perform that analysis to your awareness? . . . THE WITNESS:
`I’m not sure.”); Ryjkov Depo. at 119:20-120:4 (“Did you consider -- consider or perform any
`analysis on whether there would have bene enough resources in Europe to complete the
`migration at any other time besides in the 2020 to 2021 time frame? A. No, I haven’t. Q. Do
`you know if anyone else at Google has done that analysis? A. I don’t know.”); Konrad 2021
`Depo. at 47:19-49:15 (“Q. Did you perform any analysis of the costs associated with making a
`shift from having the Content ID system located in the data centers that it was to different data
`centers that were all outside the United States? A. I haven’t personally performed any such
`analysis.”); see also Malina Depo. at 125:20-126:8; Moor Depo. at 87:8-88:9, 115:20-119:3.
`
`35.
`I have also not seen any evidence that Defendants have evaluated whether additional
`hardware (such as additional computers, servers, and trans-Atlantic cabling) would have been
`needed to be purchased to meet any additional computing or storage demands caused by the
`migration of the MatchSystem outside the United States in any of 2011, 2012, and 2014. Such
`costs could have been quite substantial. The only information I have seen is concerning the
`cost associated with such a migration is from the 2021 timeframe. See Malina Depo. at 104:19-
`105:4; GOOG