`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`______________________________________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`______________________________________________
`
`
`
`GOOGLE LLC,
`
`Petitioner
`
`v.
`
`WAG ACQUISITION, LLC
`
`Patent Owner
`
`U.S. Pat. No. 9,762,636
`
`
`
`_______________________________________
`
`Inter Partes Review Case No. IPR2022-01413
`
`_______________________________________
`
`
`
`PATENT OWNER’S SUR-REPLY
`
`
`
`
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`TABLE OF CONTENTS
`
`
`
`INTRODUCTION ..................................................................................................... 1
`
`CLAIM CONSTRUCTION ....................................................................................... 3
`
`A. Preambles .......................................................................................................... 3
`
`B. “Requests” Limitations ..................................................................................... 4
`
`C. “Data Rate of the Data Connection” Limitation ............................................... 6
`
`D. “Without Depending on the Server System Maintaining a Record” Limitation
`
`8
`
`CARMEL GROUNDS (1-4) ...................................................................................... 8
`
`A. Alleged Carmel Disclosure of “Client Control” ............................................... 8
`
`1. Carmel’s alleged “express disclosure” Contention that Carmel discloses
`“Pull” ................................................................................................................10
`
`B. Google’s Arguments as to Specific Limitations .............................................20
`
`1.
`
`2.
`
`“Supplying” Limitation ............................................................................20
`
`“Requests” Limitations ............................................................................21
`
`3. Connection Data Rate Limitation ............................................................22
`
`4. Dependency on record of the last media data element sent .....................25
`
`C. Grounds 2 and 3 ..............................................................................................25
`
`CONCLUSION ........................................................................................................26
`
`
`
`
`
`
`
`
`
`
`
`–ii–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`LIST OF PATENT OWNER’S EXHIBITS
`
`Exhibit
`
`Description
`
`2001
`
`WAG Acquisition, LLC v. WebPower, Inc., 781 F. App’x 1007
`(Fed. Cir. 2019)
`
`2002
`
`Declaration of W. Leo Hoarty
`
`2003
`
`Declaration of Henry Houh (Ex. 1002 of IPR2022-01228)
`
`2004
`
`May 23, 2023, Deposition of Dr. Kevin Jeffay
`
`2005
`
`IETF RFC 2068
`
`2006
`
`May 25, 2023, Deposition of Dr. Nathaniel Polish
`
`2007
`
`2008
`
`2009
`
`2010
`
`2011
`
`Declaration of Dr. Nathaniel Polish, Emblaze Ltd. v. Apple Inc.,
`case no. 11-CV-01079 (N.D. Ca. Feb. 14, 2014)
`
`In re Certain Fitness Devices, Streaming Components Thereof,
`and System Containing Same, Inv. No. 337-TA-1265, Initial
`Determination (ITC, Sept. 9, 2022) (CALJ Clark S. Cheney)
`
`In re Certain Fitness Devices, Streaming Components Thereof,
`and System Containing Same, Inv. No. 337-TA-1265, Evidentiary
`Hearing – Volume III (ITC, March 14, 2022)
`
`Final Written Decision, WebPower v. WAG Acquisition, LLC,
`IPR2016-01238, Paper No. 22 (Dec. 26, 2017)
`
`Final Written Decision on Remand, WebPower v. WAG
`Acquisition, LLC, IPR2016-01238, Paper No. 28 (July 16, 2020)
`
`2012
`
`Microsoft Computer Dictionary, Fifth ed. (excerpts)
`
`2013
`
`Redline comparison of claims of ’824 and ’636 patents
`
`2014
`
`Claim term concordance table
`
`2015
`
`IETF RFC 1945
`
`2016
`
`October 4, 2023, Deposition of Dr. Nathaniel Polish
`
`–iii–
`
`
`
`
`
`
`INTRODUCTION
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`Goggle’s Reply (as did the Petition) focuses nearly entirely on Carmel. The
`
`evidence reflects that Carmel works by a different mechanism than asserted by
`
`Google. The facts are inconsistent with Google’s theories throughout (including
`
`improper new theories as well as the original theories).
`
`Pivoting, Google now seeks to turn Carmel on its head. Google refers several
`
`times to language near the end of Carmel referencing upload from source computer
`
`34 to server 36, stating “similar methods are applicable, mutatis mutandis, to the
`
`method of downloading the files from server 36 to clients 30, as shown in FIG.
`
`6A.” EX1003-13:32-35. Google, at Reply-10-11, relies on its expert, EX2011
`
`¶¶ 74-78, pointing to source computer 34’s control of slices it puts on respective
`
`links for upload to server 36 and seeking to project those operations onto the
`
`download process. But this clearly cuts against Google. In fact, it was Patent
`
`Owner’s argument (POR-56) that server 36, in its transmissions to clients 30, was
`
`doing the same thing—what source 34 did via ftp (concededly a push), server 36
`
`did relative to client 30 (i.e., a push).
`
`At Reply 24, however, Google takes this further, with an entirely new theory,
`
`which not only projects the source 34-server 36 interaction “mutatis mutandis”
`
`onto the server 36-client 30 interaction, but further, based on no written disclosure
`
`anywhere, reverses the interaction as well, imagining it as a client 30-server 36
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`interaction, with the client now in control. But this is still not even an attempt to
`
`explain how a disclosed continuous push transmission of successive elements can
`
`be transformed into individually “as required” requests to send identified elements.
`
`The Board cannot adopt an interpretation of the prior art based on such inversions
`
`and then reinversions of the evidence, which fail to complete the required
`
`connections between the elements.
`
`(Reply-2) Google’s Reply tellingly leads not on the merits but rather by
`
`appealing to a prior decision addressing different claim terms. Neither the Petition
`
`nor the Reply use the words “collateral estoppel” (and it isn’t). The prior decision
`
`is also not evidence. The Board must base its decision on the evidence in this case.
`
`(Reply-2, “Client-side control”) The Petition did not invoke “client-side
`
`control” and did not mention the Federal Circuit decision referenced here. Yet
`
`Google’s mantra throughout its Reply is that “Carmel emphasizes client-control.”
`
`Carmel does not use the words “client-side control,” and what the client does
`
`control in Carmel is limited to bulk interactions, such as where streaming begins,
`
`the number of links to the server, and compression levels. Google would imply
`
`from this that Carmel also discloses individual element requests, but this does not
`
`follow. The words “client-side control” are being used by Goole to gloss over lack
`
`of actual evidence for individual element requests.
`
`–2–
`
`
`
`
`
`
`CLAIM CONSTRUCTION
`
`A.
`
`Preambles
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`(Reply-3-4, addressing Patent Owner’s argument that additional
`
`limitations make clear that the preambles and claims address a continuous
`
`process that continues for the duration of playback) Google’s arguments
`
`concerning the preambles, and indeed each of the claims as a whole, falter when
`
`viewed together with limitations that apply by their terms to “all” of the media data
`
`elements or “each” sending. In view of the preamble’s reference to a method (or
`
`system, or recorded medium) “for distributing a live audio or video program,” and
`
`in view of the specification’s disclosure that the user computer “continues with
`
`additional data requests for the duration of playing the audio/video material”—and
`
`that these are the same “requests” referenced in the claim as being correspondingly
`
`received by the server over the same duration—it is clear that the claim addresses a
`
`process structured to be sustainable over the duration of the corresponding client’s
`
`playback, and that the server must therefore be implemented so that the limitations
`
`will be met for a transmission of that duration.
`
`Google argues that the fact that a user could stop watching a video in mid-
`
`stream limits what “the program” means in the preamble. All this argument does is
`
`highlight the continuing nature of the claim limitations, without disputing that the
`
`limitation requirements continue but simply quibbling over how long they
`
`–3–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`continue. Google does not point to any structural operational difference that results
`
`from the fact that a user may leave early.
`
`(Reply-5, Mr. Hoarty’s testimony) Additionally, Google omits testimony from
`
`Mr. Hoarty it does not like: “I would read that as the duration that the user was
`
`viewing the program.” EX1103-25:10-12. Contrary to Google’s assertions, Mr.
`
`Hoarty’s position is consistent with Patent Owner’s.
`
`(Reply-5, streaming of live programs) Google’s argument with regard to a
`
`live program is even more strained. Users inherently join and leave ongoing live
`
`streams, for example a live streaming of a radio or TV station. Google points to
`
`nothing in the structural characteristics and operational capabilities of a server that
`
`can receive a live feed from a source and redistribute it over the internet, that
`
`would change as a consequence of when users enter or leave the transmission
`
`stream.
`
`B.
`
`“Requests” Limitations
`
`(Reply-6-7, that “PO’s sole evidence [for individually identified requests] is
`
`a sentence in the specification”) Google’s assertion here is incorrect, as Patent
`
`Owner first argued the claim limitations must be read together. POR-13 argued that
`
`“[t]he plain and ordinary meaning of [limitation d] is that if the request responded
`
`to is for more than one element, then each of the elements in the response must
`
`have been specified as part of the request by its respective serial ID.” Patent
`
`–4–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`Owner argued that “read in light of limitation d … the only reasonable
`
`interpretation of limitations c-c(ii)” is that “even if the ‘received request’ is for
`
`more than one element, each such requested element is identified in the request by
`
`the respective serial identifier of that media data element.” POR-11-12. The Reply
`
`does not address or dispute Patent Owner’s arguments that the claims read as a
`
`whole mandate Patent Owner’s construction. Google ducks the issue, focusing only
`
`on the specification.
`
`When it turns to the specification (falsely characterizing it as “PO’s sole
`
`evidence”), Googe gets that wrong as well, bowling over the words:
`
`the user computer transmits a request to the server to send one or
`
`more data elements, specifying the serial numbers of the data
`
`elements.
`
`EX1001-14:51-53 (emphasis added). This disclosure is unmistakable and contrary
`
`to Google’s argument that the specification does not recite requests specifying each
`
`serial number of the requested elements. Google argues for a construction contrary
`
`to any disclosed embodiment. Even Google’s expert agrees, contrary to the Reply,
`
`that this specification language “would not include a data element that is not
`
`–5–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`identified by—identified in the request by a serial number.” See EX2016-42:2-
`
`43:11.
`
`C.
`
`“Data Rate of the Data Connection” Limitation
`
`(Reply-7-8, that “data rate” is only tied to an unspecified number of
`
`elements, implicitly as little as a single element) Google’s Reply on this point
`
`does not address Patent Owner’s actual argument. The connection data rate
`
`limitation is not an instantaneous rate, e.g., transfer rate for a single element (POR-
`
`15) (EX2016 45:4-46:9 (a “single observation wouldn’t really be necessarily
`
`dispositive”)), nor is it a rate over some unspecified number of elements that could
`
`be as little as a single element. Furthermore, Google’s dichotomy, of a single
`
`element vs. each and every element (“must always”), is a false one.
`
`Google’s false dichotomy is: “Under PO’s interpretation, each-and-every
`
`element must be transferred faster than the playback rate, but a ‘live’ program
`
`cannot be continuously transferred at a rate faster than the playback rate since it is
`
`impossible to get ahead for a ‘live’ program.” Reply-8 (Emphasis added). This
`
`Google argument reveals a failure to appreciate the disclosure.
`
`It is the transmission that causes media data to be collected ahead of the
`
`playback, by virtue of filling a client-side store or buffer. The requested serially
`
`identified elements exist on the server when they are requested. They are each
`
`“transmitted to the user computer as fast as the data connection between the user
`
`–6–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`computer and the server will allow” (EX1001-14:61-62), necessarily implying that
`
`there is extra time between the elements. The user computer “incorporate[es] a user
`
`buffer and comprises means for receiving and storing a predetermined number of
`
`media data elements.” EX1001-15:10-12. The player thus starts playback from the
`
`buffer, which stores elements ahead of what is needed for continued playback, and
`
`it stays ahead, because “as data is played out, the next sequential data elements are
`
`requested from the server in such a fashion as to approximately maintain the
`
`predetermined number of data elements in the user’s buffer.” EX1001-15:15-18.
`
`Contrary to Google’s assertion, this does not necessarily require that “each-and-
`
`every element must be transferred faster than the playback rate.” But it does
`
`require a connection having a data rate such that, when measured over at least “the
`
`predetermined number” of sequential elements, the transmission time of those
`
`elements will be less than their back-to-back playback duration. That is the
`
`minimum required to render the combined limitations, in accordance with the
`
`literal claim language, operable, so as to maintain approximately a predetermined
`
`number of media data elements in the user buffer as data is constantly being played
`
`out at the playback rate. EX2002 ¶¶ 26-27, 47.
`
`(Reply-8-9, comparing prior Federal Circuit decision) Patent Owner agrees
`
`that the Federal Circuit decision concerned a transmission rate for individual
`
`elements.
`
`–7–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`D.
`“Without Depending on the Server System Maintaining a
`Record” Limitation
`
`(Reply-9, “PO does not advance any specific construction for this limitation
`
`[d(iii)]”) This is a misstatement. POR-19 pointed out that it was “[t]he Petition
`
`[that] does not argue any particular construction for this term” (emphasis added)
`
`and the POR expressly set forth Patent Owner’s interpretation, to which the Reply
`
`has no response, obliquely inserting cross-references to its discussion of other
`
`limitations. Patent Owner cannot discern here an actual argument against its
`
`construction of limitation d(iii).
`
`CARMEL GROUNDS (1-4)
`
`(Reply-9, “All four grounds in the Petition rely on Carmel as a primary
`
`reference”) Correct (and the Petition’s downfall).
`
`A. Alleged Carmel Disclosure of “Client Control”
`
`(Reply-10, “Carmel discloses client control”) As addressed above, Carmel
`
`discloses that the client can control the starting slice from which the server will
`
`push successive elements, can open HTTP connections, and can send commands to
`
`ask the server to change which available encoding level of the streaming data to
`
`send, none of which get to individual element requests. The catch-phrase “client
`
`control” is insufficient.
`
`(Reply-10, “‘[C]lient 30’ selects and downloads a slice”) This is a
`
`manufactured quote and so misleading it should be stricken. Nor does the
`
`–8–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`disclosure tie the “SELECT SLICE” box in any way to “downloads a slice.” The
`
`disclosure as to the word “select” is that “graphic 56 (FIG. 3C) is displayed by the
`
`client, so that a user can decide and indicate at which slice of data stream 40 to
`
`begin downloading. Responsive to a user input, client 30 selects an appropriate
`
`starting slice.” EX1003-10:43-46 (emphasis added).
`
`(Reply-11, “By teaching that ‘client 30’ monitors links, establishes links,
`
`and controls the slices allocated to each link, Carmel teaches a client-driven
`
`and ‘pull’-based approach”) “Pull” does not follow from “monitors links” and
`
`“establishes links.” The only thing Google cites (in footnote 3) for “controls the
`
`slices allocated to each link” is Carmel’s disclosure of what source computer 34
`
`does. Google repeatedly over-relies on imputing details of the upload process from
`
`source 34 to server 36, to apply in every respect to server 36’s transmission to
`
`clients 30. But here, Google actually argues against itself, because computer 34 is
`
`the media source relative to server 36—the reverse of what Google seeks to
`
`establish. Google’s argument here appears to agree with Patent Owner.
`
`(Reply-11, “‘client 30’ decides whether to break a link”) Carmel does not
`
`suggest that any of clients 30 controls whether to retransmit over a broken link.
`
`The cited disclosure from Carmel’s col. 10 once again references source computer
`
`34. The mere words “mutatis mutandis” (from EX1003-13:30-35) are insufficient
`
`to overcome the specific evidence (or substitute for it), and in any case, the
`
`–9–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`“control” in the upload scenario is being applied from the data source side
`
`(computer 34), again, the opposite of what Google seeks to argue. Google again
`
`argues in favor of Patent Owner’s position.
`
`Moreover, the “mutatis mutandis” reasoning raises the issue that upload is
`
`preferably via ftp, which the experts both agree is a push protocol. POR-45;
`
`EX2006- 114:1-18 (Polish); EX2002 ¶ 63 (Hoarty).
`
`1.
`Carmel’s alleged “express disclosure”
`Contention that Carmel discloses “Pull”
`
`(Reply-12, “only ‘client 30’ can decide when to proceed with the next slice
`
`[to DECODE/OUTPUT]”) Nothing here even suggests a “request” to the server,
`
`as opposed to processing slices already received via push. See POR-51-52.
`
`Nothing in Carmel suggests that the checking link function is coordinated with
`
`any “request” for a slice. The consequence of detecting significant lag is to open a
`
`new link. EX1003-10:55-63. There is no basis to read anything more into this.
`
`There is also nothing that says a “request” to the server for a slice is made
`
`anywhere in the loop iteration.
`
`DECODE and OUTPUT DATA are also not definitive since in any formulation,
`
`push or pull, the client will need to decode and output the media data.
`
`(Reply-12, “even in HTTP1.1 each file must be individually requested—
`
`something Patent Owner’s expert admits”) This argument is fraught. First, it
`
`–10–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`was not in the Petition, second, it blatantly misrepresents the underlying testimony,
`
`and third, it is contrary to the HTTP 1.1 specification in evidence.1
`
`First, in a related case, Disney asserted in its (earlier filed) Petition that
`
`individual file requests are inherent in HTTP, and Patent Owner addressed this at
`
`length in that case. Google did not make that assertion in the Petition herein,
`
`raising it now for the first time in its Reply.
`
`Second, Google’s statement that its argument is “something PO’s expert
`
`admits” is a total fabrication. Mr. Hoarty testified that “a POSITA would not
`
`understand Carmel to disclose an individual request (pull) form of streaming”
`
`(EX2002 ¶ 62), and that HTTP 1.1 supported responses “comprising an open-
`
`ended sequence of data elements, so that, in response to a single invocation, the
`
`server could send sequential portions of the content,” citing and explaining
`
`referenced detailed testimony by Dr. Jeffay (Amazon’s expert). EX2002 ¶ 53-54.
`
`Testimony that Mr. Hoarty cites (EX2009-637:10-639-7) obliterates any
`
`“inherency” assertion:
`
`
`1 Patent Owner recognizes these are harsh words, but intends them, as it believes
`
`the Reply here and in several other places crosses well over the line of zealous
`
`advocacy.
`
`–11–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`• Disagrees that, “if Carmel … is transmitting a request via HTTP, then that
`
`request necessarily identifies the requested file stored by the video server.”
`
`EX2009-637:10-16.
`
`• Explains that, “HTTP is famously the best example of a giant Swiss Army
`
`knife protocol that we have on the Internet.… There are many ways to get
`
`content sent to you by HTTP.” EX2009-637:18-22.
`
`• Confirms that mere reference to HTTP doesn’t inherently mean a request for
`
`a file but instead that “there are many other ways that you can get content
`
`via HTTP without making a GET request and without having to request a
`
`file, for example, by name.” EX2009637:24-638:8.
`
`• That HTTP supports “a GET request that does not use a file name.… [but] to
`
`identify a stream…. not request a file; instead request a higher level object
`
`like a stream. And HTTP supports a mode of content delivery that is called
`
`chunked transfer encoding, where … in each HTTP response, you send a
`
`sequential portion of the content that’s being requested. And this is described
`
`in Carmel in column 2.” EX2009:638:12-639:5.
`
`• Further, a POSITA “would understand that a perfectly viable way to
`
`implement that would be … [to] generate a post request to the server,
`
`sending the data that … the user had input via the client app, and with that
`
`data having been received at the server, the server would then start sending
`
`–12–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`media down to the client, for example, using chunked encoding.” EX2009-
`
`639:18-24.
`
`Google grossly misrepresents Mr. Hoarty’s deposition testimony—from another
`
`case. See EX1111 ¶ 68, quoting EX1101 (Disney deposition of Hoarty) 66:11-67:1.
`
`In the Disney deposition, Mr. Hoarty was specifically asked about “a request for
`
`File A” (EX1101-66:11), to which he replied, answering a following question that
`
`was explicitly “based on that scenario” (EX1101-66:17-18), that the server would
`
`not send File B if the request “didn’t ask for it.” EX1101-66:20. The suggestion
`
`that this answer constitutes an “admission” that “in HTTP 1.1 each file must be
`
`individually requested” is not commensurate with the question and is outright
`
`false.
`
`Google’s belated argument would assume an HTTP request has to be for a file.
`
`However, just the same as Mr. Hoarty and Dr. Jeffay, Google’s expert himself
`
`confirmed the contrary in his most recent deposition, that HTTP 1.1 specifies
`
`requests more broadly, for “entities.” EX2016-88:3-6 (“Q. a GET request is for an
`
`entity? A. “Yeah, I guess that’s the way they refer to it.”); EX2016-90:16-25 (Q.
`
`“chunked encoding here it refers to an entity that is generated dynamically?” A.
`
`–13–
`
`
`
`Case IPR2022-01413
`
`
`Patent 9,762,636
`
`
`Yeah … they’re being if you will, dynamically generated.”).2 There is nothing left
`
`to back Google’s incorrect assertion that an HTTP request must be for an
`
`individual file.
`
`POR-33, 35, 50 noted the lack of evidence of any Carmel disclosure of an
`
`HTTP request for an individual file. Google’s response is its misstatement of Mr.
`
`Hoarty’s testimony, and a new (to this case) and likewise manufactured assertion
`
`that anything else would not have been technically possible. At Reply-12, Google
`
`throws in a string of bare cites, purporting to evidence individually identified
`
`requests, but each of those citations falls flat on inspection.
`
`(Reply 12-13, HTTP cannot support sending unrequested files) It is
`
`uncontested that HTTP does not support the sending of unrequested files. Google’s
`
`argument, however, is a strawman assuming the conclusion that Carmel’s requests
`
`are for files, when Carmel does not disclose that, and nothing technically requires
`
`it. In Carmel, there is one request to send a stream (which is comprised from the
`
`
`2 Dr. Polish, at EX1111 ¶ 68, cites Swaminathan (EX1104-3-4), but later
`
`acknowledged that Swaminathan chunked stream fragments (each corresponding
`
`to a Carmel “slice”). EX2016-28:24-30:5. Here, the larger structure, the stream, is
`
`being chunked. That a file itself may be chunked for delivery does not show that
`
`HTTP 1.1 “requires” that HTTP requests be files or slices.
`
`–14–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`contents of sequential slices) starting from a specified slice. No one says the file
`
`contents included in the resulting stream are “unrequested.” But that does not
`
`imply the included content is “individually requested” for each slice. Indeed, as
`
`evidenced above, a POSITA would recognize that it is not.
`
`(Reply-13, “specialized software”) Google still does not respond to POR-26-
`
`27, as to what would take software that implements an open specification outside
`
`the scope of Carmel, which specifically invokes that specification. EX1003-2:22-
`
`28.
`
`(Reply-12, n.4, “‘chunked transfer encoding’ is used when the file size is
`
`undefined or unknown.”) The HTTP 1.1 standard (EX2005) actually says that
`
`“chunked encoding … allows dynamically-produced content to be transferred”
`
`(EX2005-21/140 (emphasis added))—a permissive statement—not that chunked
`
`transfer encoding “is” used “when the file size is undefined or unknown,” as
`
`Google suggests.
`
`Carmel uses a live feed as its base example, a prime example of dynamically-
`
`produced content. While allowing for such content, the RFC states no requirement
`
`that the content must not be complete, requiring only that the “entity-body” consist
`
`of “OCTETS” (bytes). See EX2005-36/140.
`
`(Reply-13, “HTTP1.1 … require[s] that each file must be individually
`
`requested”) Google finesses even its own expert’s testimony, which did not say
`
`–15–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`“individually” requested. Cf. Ex. 1111 ¶ 69 (“specifically” requested). Google’s
`
`Reply argues that something sent as an internal part of a continuous stream, and
`
`therefore “requested” in that sense, morphs into “individually” requested, a
`
`conclusion that does not follow. The expert himself stretches the argument by
`
`saying “specifically” requested, and while the word “specifically” can perhaps be
`
`justified under a very inclusive view of “specifically,” it is not sufficient to meet
`
`the claim limitations. To take this over the line, Google’s Reply takes “specifically”
`
`requested (the best it could get from the evidence) and expressly represents it as if
`
`the evidence had said “individually” requested, which it does not.
`
`(Reply-13, SELECT SLICE is included in every continue loop) Fig. 6A
`
`reflects a loop in which the client repeatedly receives elements via HTTP. EX2002
`
`¶ 68. Google would infer from Carmel’s Fig. 6A that SELECT SLICE is executed
`
`on every loop iteration and represents a request. Google points to nothing in the
`
`written disclosure to support this. Google dodges the issue of why SELECT SLICE
`
`follows rather than precedes HTTP FROM SERVER raised in the POR. POR-34-
`
`35.
`
`Dr. Polish acknowledged that Fig. 6B represented the same underlying
`
`approach as Fig. 6A. EX2006-105:13-106:1 (“So in 6B it’s adjusting the level. In
`
`6A it’s just checking whether the link is broken”). That being the case, Google does
`
`not explain why Fig. 6B says READ SLICE and not SELECT SLICE as in Fig.
`
`–16–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`6A. The simple explanation for this is because SELECT SLICE only means select
`
`a slice responsive to human input (POR-34), and that, as reflected in the
`
`corresponding descriptions at EX1003-10:42-47 and EX1003-10:64-11:8
`
`respectively for Figs. 6A and 6B, the graphical user interface feature of Carmel that
`
`supports such selection is only addressed in connection with Fig. 6A.
`
`(Reply-13-14, HTTP FROM SERVER means “PULL”) As to Google’s
`
`contentions that Figs. 6A and 6B of Carmel are cast as steps performed by client 30
`
`and so should be understood as a “pull” from the server, the actual underlying step
`
`in each case that is performed by the client is receiving data from the server. This
`
`does not require “pull.”
`
`Google’s citation at Reply-14 to Mr. Hoarty’s deposition is off-base, as this
`
`testimony concerned practicing the ’594 patent, not Carmel.
`
`(Reply-14, “client … reads an index file to identify, select, and request the
`
`next-needed slice”) Google fails to identify any disclosure that Carmel repeats a
`
`process of reading the index file to select and request each successive slice. Merely
`
`glossing over fifteen lines of disclosure in this way when the disclosure itself
`
`actually says nothing of the sort is insufficient.
`
`(Reply-14, “only ‘client 30’ would know when the “DECODE” and
`
`“OUTPUT DATA” steps are complete, so only ‘client 30’ can decide when to
`
`proceed with the next slice) Regardless of a push or pull implementation, it is
`
`–17–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`obviously always the client that decodes and outputs the media data. This thus has
`
`no bearing on “pull.”
`
`(Reply-14-15, ITC ruling) Regardless of the underlying claims in the ITC
`
`case, the ALJ made various factual findings as to Carmel after trial on the merits
`
`that included testimony from various experts. The Board should be aware of the
`
`ITC ruling and factual conclusions therein regarding Carmel.
`
`(Reply-15, Claim constructions in N.D. Cal.) Google’s citation from N.D.
`
`Cal. is construction of a limitation in Carmel’s claims concerning the rate that a
`
`client “can download the sequence” of uploaded files. The claims in question are
`
`broad, and do not recite requests. It is not a ruling that Carmel discloses successive
`
`individual requests.
`
`(Reply-15-16, Download implies a request) Carmel uses the word
`
`“download” with respect to both Figs. 6A and 6B. In the context of Fig. 6B,
`
`Carmel recites “downloading” at 10:65, which it relates to the action of “the client
`
`selects one of the available quality levels in the stream. Responsive to the selection,
`
`server 36 begins to transmit data slices at the chosen quality level.” EX1003-11:4-
`
`7 (emphasis added). The “download” here is unmistakably the result of receiving
`
`the successive slices that the server transmits. Google does not explain why
`
`“download” has a narrower meaning when used earlier in col. 10 of Carmel.
`
`–18–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`Patent Owner does not dispute that there is a “request,” insofar as the client
`
`requests transmission of the stream. The client requests the stream and downloads
`
`its elements as they are sent. However, the mere word “download,” under either
`
`side’s preferred definition, does not address whether there are individual requests
`
`for the constituent elements that comprise the stream that the client has requested.
`
`(Reply-16-17, Expert “poll”) In a case of multiple challengers, a headcount of
`
`their experts is not probative. The notable development is that one of the experts
`
`that Google purports to count in this tally actually waffled. Google quibbles about
`
`this (yet seems concerned enough to raise it (in footnote 5)), but the other expert’s
`
`prior testimony that neither the Figs. 6A nor 6B embodiments disclose slices being
`
`independently requestable is a major inconsistency on the challengers’ side.
`
`(Reply-17, “lacking server-side software for similar functions, a POSITA
`
`would overall understand this disclosure to be consistent with ‘pull.’”) It does
`
`not follow that a client is making individual pull requests simply because it asks
`
`the server to switch to a different encoding level. The POR at 32-33 addressed this,
`
`and merely repeating a contrary conclusion does not advance Google’s case.
`
`(Reply-17-18 and n.6, that the “heart” of the Carmel invention is pull, to
`
`which the inventor later added push) If Google is now joining the camp that
`
`distinguishes the underlying principle of the Fig. 6A and 6B embodiments, its
`
`position on the word “download” is even further nullified, because “downloading”
`
`–19–
`
`
`
`
`
`
`
`
`
`Case IPR2022-01413
`Patent 9,762,636
`
`at 10:65 of Carmel would be concededly tied to push. Moreover, Google cannot
`
`take this tack now, as its expert testified to the contrary. EX2006-106:2-7. Finally,
`
`the assertion that the Carmel inventors, without comment, tacked on an
`
`enhancement that required a completely different underlying implementation