Opinion
2022-1203 2022-1204
10-25-2023
NETFLIX, INC., Appellant v. DIVX, LLC, Appellee
Thomas Saunders, Wilmer Cutler Pickering Hale and Dorr LLP, Washington, DC, argued for appellant. Also represented by Robert Stiller; Mark Christopher Fleming, Sarah R. Frazier, Melissa N. Morel, Boston, MA. Michael Gregory Pattillo, Jr., MoloLamken LLP, Washington, DC, argued for appellee. Also represented by Rayiner Hashem, Jeffrey A. Lamken; Jonathan E. Barbee, Nathaniel Rubin, New York, NY; Nathan Nobu Lowenstein, Kenneth J. Weatherwax, Lowenstein & Weatherwax LLP, Santa Monica, CA.
Appeals from the United States Patent and Trademark Office, Patent Trial and Appeal Board in Nos. IPR2020- 00647, IPR2020-00648.
Thomas Saunders, Wilmer Cutler Pickering Hale and Dorr LLP, Washington, DC, argued for appellant. Also represented by Robert Stiller; Mark Christopher Fleming, Sarah R. Frazier, Melissa N. Morel, Boston, MA.
Michael Gregory Pattillo, Jr., MoloLamken LLP, Washington, DC, argued for appellee. Also represented by Rayiner Hashem, Jeffrey A. Lamken; Jonathan E. Barbee, Nathaniel Rubin, New York, NY; Nathan Nobu Lowenstein, Kenneth J. Weatherwax, Lowenstein & Weatherwax LLP, Santa Monica, CA.
Before Dyk, Linn, and chen, Circuit Judges.
OPINION
CHEN, CIRCUIT JUDGE
Netflix, Inc. (Netflix) appeals two final written decisions of the Patent Trial and Appeal Board (Board) determining Netflix did not meet its burden of proving claims 15 and 12 of U.S. Patent No. 9,270,720 ('720 patent) and claims 1-6, 8-10, 13-14, 16-17, and 19 of U.S. Patent No. 9,998,515 ('515 patent) unpatentable under 35 U.S.C. § 103. See Hulu, LLC v. DivX, LLC, No. IPR2020-00647, 2021 WL 4859625 (P.T.A.B. Sept. 27, 2021) ('720 Patent Decision); Netflix, Inc. v. DivX, LLC, No. IPR 2021-00648, 2021 WL 4859708 (P.T.A.B. Sept. 27, 2021) ('515 Patent Decision). Each Board decision in these inter partes reviews (IPR) is lengthy, with the Board closely analyzing the arguments and evidence presented by Netflix and the patent owner DivX, LLC (DivX). Netflix's appeal does not challenge any of the Board's substantive analysis. It instead makes a purely procedural argument, accusing the Board of falling down on the job by failing to address several arguments purportedly raised in Netflix's petitions. After marching through the relevant papers below, we see no error in how the Board understood Netflix's petition arguments. Because all of Netflix's arguments on appeal were not raised below, they are forfeited. We affirm.
BACKGROUND
A. The Challenged Patents
The '720 and '515 patents, assigned to DivX, relate to a method for adaptive bitrate streaming of content on a playback device, such as a mobile phone or personal computer. "Adaptive bit rate streaming involves detecting the present streaming conditions (e.g., the playback device's network bandwidth and video decoding capacity) in real time and adjusting the quality of the streamed media accordingly." '720 patent col 1 ll. 30-34. To accomplish this adjustment, the method creates an index file that points to alternate source media streams (what the claim refers to as "assets"), where each stream contains video and audio data. '720 patent col. 1 ll. 57-60. Each alternate source stream can encode media at different maximum bitrates, and the playback device may switch between the alternate streams. '720 patent col. 7 ll. 29-42.
The '515 patent is a continuation of an application that issued as the '720 patent. The specifications of the '515 patent and '720 patent are thus nearly identical. We refer to the '720 patent unless otherwise noted.
The '720 patent and '515 patent both generally claim a method for receiving a request for content from a playback device, retrieving a list of assets associated with the requested content, filtering the list of assets based on certain criteria, generating an index that describes each asset in the filtered list, and sending the index to the playback device. For the '720 patent, claim 1 is representative (emphases added):
1. A method of generating a top level index file, comprising:
receiving a request from a playback device at a playback server system, where the request (i) identifies a piece of content and (ii) includes a product identifier;
retrieving, using the playback server system, (i) a list of assets associated with the identified piece of content and (ii) at least one device capability based upon the product identifier, wherein
each asset is a different stream associated with the piece of content;
filtering the list of assets using the at least one device capability using the playback server system, wherein the playback server system maintains a database of product identifiers and associated device capabilities;
generating a top level index file describing each asset in the filtered list of assets using the playback server system; and
sending the top level index file to the playback device using the playback server system, wherein the top level index file is used by the playback device to determine which assets to request for playback on the device.
For the '515 patent, claim 1 is representative and closely mirrors the '720 patent's claim 1 but provides more detail for the top level index file (emphasis added):
1. A method for authorizing playback of content, comprising:
receiving a request for content from a playback device at a playback server, where the request includes a product identifier that identifies a device configuration;
identifying, using the playback server, based on the product identifier, a plurality of device capabilities including a device type and a device software version indicating a version number for an adaptive streaming software component implemented on the playback device;
retrieving, using the playback server, a list of assets associated with the identified piece of content, wherein each asset is a different stream associated with the piece of content;
filtering, using the playback server, the list of assets based on the plurality of device capabilities;
generating, using the playback server, a top level index file describing each asset in the filtered list of assets, wherein the top level index file identifies locations and bitrates of a plurality of alternative streams capable of being used to perform adaptive streaming of the content; and
sending the top level index file from the playback server to the playback device.
B. The IPR Petitions and Board Decisions
Netflix filed petitions for inter partes review for several claims of the '720 patent and the '515 patent under 35 U.S.C. § 103, challenging the claims of both patents over: (1) the combination of Pyle and Marusi and (2) the combination of Lewis and Marusi. At issue in this appeal is whether Netflix's petitions contained certain arguments directed to the "filtering the list of assets" limitation of claim 1 of the '720 patent, the "retrieving . . . a list of assets" limitation of claim 1 in both patents, and the "generating . . . a top level index file" limitation of claim 1 of the '515 patent.
U.S. Patent No. 8,782,268.
European Application EP 2180664.
U.S. Patent Publication No. 2012/0047542.
For the "filtering the list of assets" limitation of the '720 patent, Netflix argued that Pyle's selection of a pre-existing manifest met this limitation. Netflix also argued that to the extent Pyle does not teach or render obvious the filtering limitation's "database of product identifiers and associated device capabilities," Marusi's database would. As for the "generating . . . a top level index file" limitation, Netflix argued that Pyle's creation of a new manifest teaches this limitation.
The Board observed a disconnect in how Netflix used Pyle to teach both the "filtering" and "generating" limitations of the '720 patent. The "generating" limitation creates a top level index file that describes "each asset in the filtered list of assets," and so the index file's content is necessarily a byproduct of what the "filtering" step produces. But because Netflix relied on Pyle's selection of a pre-existing manifest to satisfy the "filtering" limitation and relied on Pyle's creation of a new, separate manifest, which is unrelated to Pyle's pre-existing manifests, for the "generating" limitation, Pyle's new manifest could not describe each asset in the filtered list of assets as required by the "filtering" limitation. '720 Patent Decision, 2021 WL 4859625, at *15. The Board similarly found the combination of Pyle and Marusi deficient for also relying on Pyle's selection of a pre-existing manifest to perform the "filtering" limitation. Id.
For the "retrieving . . . a list of assets" limitation in the '720 patent and '515 patent, Netflix argued "[t]he combination of Lewis and Marusi renders this limitation obvious." J.A. 6069; No. IPR 2020-00648, Paper No. 3 at 60 (P.T.A.B. Mar. 11, 2020). Specifically, Netflix argued, "[a] [skilled artisan] would have found it obvious to retrieve a list of assets using the playback server system because Lewis teaches that the manifest file server and rule resolution server generate[] a dynamic file manifest, and Lewis teaches that a dynamic manifest file contains a list of URLs to container files containing content." J.A. 6070; J.A. 7070.
The Board rejected this argument, finding that a result achieved in Lewis-i.e., generating a dynamic manifest file containing a list of URLs-cannot, without further explanation, render obvious an intermediate prior step of "retrieving." '720 Patent Decision, 2021 WL 4859625, at *19.
Finally, for the "generating . . . a top level file index" limitation in the '515 patent, Netflix argued that Pyle's transmission of a manifest (either a pre-existing or a new manifest) into storage teaches this limitation. The Board rejected this argument and found that a skilled artisan would not have understood transmitting a file into memory to correspond to the "generating" limitation. '515 Patent Decision, 2021 WL 4859708, at *14. The Board further found that-contrary to Netflix's characterization of its petition arguments in its Reply brief-the petition did not assert the additional argument that Pyle's creation of a new manifest also met the "generating" limitation. Id. at *14 & n.9.
Netflix timely appealed both final written decisions to our court. We have jurisdiction under 28 U.S.C. § 1295(a)(4)(A).
STANDARD OF REVIEW
We review the Board's judgments concerning what arguments are fairly presented in a petition and other pleadings for abuse of discretion. See Ericsson Inc. v. Intell. Ventures I LLC, 901 F.3d 1374, 1379 (Fed. Cir. 2018); Altaire Pharms., Inc. v. Paragon Bioteck, Inc., 889 F.3d 1274, 1284 (Fed. Cir. 2018), remand order modified by stipulation,738 Fed.Appx. 1017 (Fed. Cir. 2018). Likewise, "[t]he Board's determinations that a party exceeded the scope of a proper reply are reviewed for abuse of discretion." Apple Inc. v. Andrea Elec. Corp., 949 F.3d 697, 705 (Fed. Cir. 2020). "An abuse of discretion is found if the decision: (1) is clearly unreasonable, arbitrary, or fanciful; (2) is based on an erroneous conclusion of law; (3) rests on clearly erroneous fact finding; or (4) involves a record that contains no evidence on which the Board could rationally base its decision." Intelligent Bio-Systems, Inc. v. Illumina Cambridge Ltd., 821 F.3d 1359, 1367 (Fed. Cir. 2016) (citation omitted).
DISCUSSION
This appeal is yet another instance where an IPR petitioner contends that a particular argument was properly raised, but overlooked or otherwise incorrectly interpreted by the Board. We begin by acknowledging the challenge the Board faces when interpreting inter partes review petitions. We have previously reprimanded the Board for failing to address arguments that were presented in a petition. See, e.g., Provisur Techs. Inc. v. Weber, Inc., 50 F.4th 117, 123-24 (Fed. Cir. 2022); Netflix v. DivX, LLC, 2023 WL 3115576, at *5 (Fed. Cir. Apr. 27, 2023); Uniloc 2017 LLC v. Facebook, Inc., 2021 WL 5370480, at *8-9 (Fed. Cir. Nov. 18, 2021); Microsoft Corp. v. Parallel Networks Licensing, LLC, 715 Fed.Appx. 1013, 1021 (Fed. Cir. 2017). We have likewise reprimanded the Board for finding a claim unpatentable based on theories not presented in the petition. See, e.g., In re Magnum Oil Tools Int'l, Ltd., 829 F.3d 1364, 1380-81 (Fed. Cir. 2016); In re NuVasive Inc., 841 F.3d 966, 968 (Fed. Cir. 2016); Oren Techs., LLC v. Proppant Express Invest., LLC, 2021 WL 3120819, at *4-*5 (Fed. Cir. July 23, 2021). Given this backdrop, the Board walks a fine line when interpreting the scope of a petition and determining what arguments have been fairly presented.
Judge Rich once remarked: "The life of a patent solicitor has always been a hard one." In re Ruschig, 379 F.2d 990, 993 (CCPA 1967). That observation certainly applies today to administrative patent judges conducting IPRs.
While the Board should not take an overly mechanistic view of a petition and decline to address an argument because the petitioner did not present it with ideal vigor and clarity, the Board should also not have to decode a petition to locate additional arguments beyond the ones clearly made. A petitioner may not rely on a vague, generic, and/or meandering petition and later fault the Board for failing to understand what the petition really meant. Ultimately, it is the petitioner's burden to present a clear argument. Intelligent Bio-Systems, 821 F.3d at 1367 ("It is of the utmost importance that petitioners in the IPR proceedings adhere to the requirement that the initial petition identify 'with particularity' the 'evidence that supports the grounds for the challenge to each claim.'") (quoting 35 U.S.C. § 312(a)(3) (2012)); see also 37 C.F.R. § 42.104(b)(4)-(5) (2012) (requiring a petitioner to identify "[h]ow the construed claim is unpatentable," "specify where each element of the claim is found in the prior art patents or printed publications relied upon," and identify "[t]he exhibit number of the supporting evidence relied upon to support the challenge and the relevance of the evidence to the challenge raised, including identifying specific portions of the evidence that support the challenge").
The Board is entitled to discretion in how it interprets petitions. See Ericsson Inc., 901 F.3d at 1379 (the Board's compliance with its procedures reviewed for abuse of discretion). And we have rejected, many times, post-hoc attempts on appeal to include additional, new arguments not contained in the petition. See, e.g., Microsoft Corp. v. Biscotti, Inc., 878 F.3d 1052, 1074 (Fed. Cir. 2017) ("Microsoft did not present to the Board the full argument that it presents here on appeal . . . Microsoft's brief on appeal is far more detailed and contains substantial new arguments regarding why it believes Kenoyer anticipates this limitation of claim 69."); see also Wasica Fin. GmbH v. Cont'l Auto. Sys., Inc., 853 F.3d 1272, 1286 (Fed. Cir. 2017) ("We also are unpersuaded by Continental's attempts to cure the petition's deficiencies in its subsequent briefing to the Board and to us."); Yita LLC v. MacNeil IP LLC, 69 F.4th 1356, 1366-67 (Fed. Cir. 2023) (disagreeing with Yita that an argument was raised in petition and thus finding no abuse of discretion by the Board in "declining to consider Yita's argument about modifying the teaching of Rabbe . . . because it was presented too late-in a footnote in Yita's reply brief"); Henny Penny Corp. v. Frymaster LLC, 938 F.3d 1324, 1331 (Fed. Cir. 2019) ("[T]he Board did not abuse its discretion by holding HPC to the obviousness theory in its petition"); Ariosa Diagnostics v. Verinata Health, Inc., 805 F.3d 1359, 1367 (Fed. Cir. 2015) ("We see no error in the Board's rejection of Ariosa's reliance, in its Reply submissions, on previously unidentified portions of a prior-art reference to make a meaningfully distinct contention."). Any argument not raised to the Board is forfeited, and we decline to consider it for the first time on appeal. In re Google Tech. Holdings, 980 F.3d 858, 863 (Fed. Cir. 2020).
Netflix contends that the Board ignored several arguments allegedly raised in its petitions. First, for the "filtering" limitation in the '720 patent, Netflix argues that its petition relied on (1) Pyle's new manifest as disclosing this limitation (in addition to relying on Pyle's pre-existing manifest), and alternatively (2) Marusi for performing this filtering assets step. Appellant's Br. 38, 42-43. Second, for the "retrieving" limitation in the '720 patent and '515 patent, Netflix argues the Board misinterpreted Lewis as a § 103 obviousness reference when the petition relied on Lewis as a § 102 anticipatory reference. Id. at 53. Finally, for the "generating" limitation in the '515 patent, Netflix argues that the Board erred in treating the argument that Pyle's new manifest could also satisfy the "generating" limitation as an improper reply argument that had no basis in the original petition. Id. at 45. As explained below, none of Netflix's contentions has merit.
I. "Filtering" Limitation
Netflix's petition plainly explained how the "filtering" limitation corresponded to Pyle's selection of a pre-existing manifest. We see no discussion in the petition's "filtering" section, however, of how Pyle's creation of a new manifest meets this limitation. It was more than reasonable, then, for the Board to understand Netflix as only mapping Pyle's pre-existing manifest embodiment, and not also its new manifest embodiment, to the "filtering" limitation. On appeal, Netflix points to the petition's block quote from Pyle stating that the new manifest "can be optimized" based on certain features. Appellant's Br. 36-37. But nothing in that quoted language itself signals to a reader that an optimized new manifest equates to "filtering the list of assets," let alone that Netflix was advancing such an argument. And contrary to the dissent, the petition's associated statement that "Pyle teaches the use of new manifest files that can be created based upon particular device(s) or capabilities," likewise fails to indicate what it is about Pyle's new manifest that discloses a list of assets that then gets filtered. J.A. 6046; see Dissent at 4.
In stark comparison, the petition made it quite clear Netflix was focused on Pyle's pre-existing manifest as teaching the "filtering the list of assets" limitation. When arguing that Pyle's selection of a pre-existing manifest met this limitation, Netflix specifically stated "Pyle . . . filters lists of assets by selecting from the various manifests an appropriate manifest to send." J.A. 6046-47 (emphasis added). Netflix thus showed it knew how to put forward a clear mapping of Pyle to the filtering limitation, but never connected Pyle's new manifest to that limitation like it did for Pyle's selection of a pre-existing manifest. While it is not strictly necessary for a petition to use the exact words of a limitation when making an argument, in this instance, the Board appropriately understood Netflix's "filtering" section of the petition to be devoted to relying solely on Pyle's pre-existing manifest. Moreover, even in its appellate briefing and at oral argument, Netflix does not explain how the items selected to create Pyle's new manifest could be regarded as "a list of assets" "wherein each asset is a different stream" associated with a piece of content, further underscoring the lack of clarity with Netflix's new manifest argument. Oral Argument at 15:51-17:04. In sum, while the petition explained that Pyle disclosed two embodi-ments-a new manifest and a pre-existing manifest-the petition's explanation for how Pyle taught the filtering limitation was limited to just Pyle's pre-existing manifest.
The dissent cites the Board's institution decision as evidence that the Board understood Netflix's petition as arguing that Pyle's new manifest met the "filtering" limitation. Dissent at 4-5. But the institution decision did nothing more than recognize that the petition described Pyle's two manifest embodiments and focused on only the pre-existing manifest for "filtering." In fact, the Board's institution decision never opined on the persuasiveness of any potentially-raised new manifest argument; its analysis was limited to the petition's pre-existing manifest contention, because that was the only relevant argument from Pyle for it to address. See No. IPR2020-00647, Paper No. 9 at 32 (P.T.A.B. Oct. 1. 2020) (Institution Decision) (noting for "filtering," "[w]e do not agree, on this record, that Petitioner's showing is deficient. Specifically, petitioner relies upon Pyle's teaching that 'composition component 210 can examine data included in request 420 and determine a suitable or optimal [pre-existing] manifest, and select and transmit that manifest to the requesting device.'"). As for the other portion of the institution decision which the dissent relies upon for support, it lays out the Board's analysis of Netflix's separate "generating" limitation arguments, and not the "filtering" limitation in dispute here. See Dissent at 5-6 (citing Institution Decision at 35 relating to "generating" limitation).
The dissent also points to another portion of the petition where it sees Netflix as having made its "new manifest" argument. Dissent at 5. However, Netflix's statement that "Pyle is teaching the generating of new manifest files by filtering the available list of assets . . ." is an argument Netflix made when mapping Pyle to the "generating" limitation, not the "filtering" limitation. J.A. 6052 (emphasis added), see also J.A. 6050 (showing heading of "generating" limitation). While the Board should review a petition holistically, it is not obligated to "cobble together assertions from different sections of a petition or citations of various exhibits in order to infer every possible permutation of a petitioner's argument[]." Microsoft Corp. v. FG SRG, LLC, 860 Fed.Appx. 708, 713 (Fed. Cir., 2021). In sum, we do not fault the Board for holding Netflix to the argument it chose to make in the "filtering" section of its petition, as opposed to reading in an additional potential "filtering" mapping from a different petition section related to a different claim limitation.
Netflix's arguments pointing to what it said in the sections of its petition challenging different claims (claims 2 and 3) fail for the same reasons. Appellant's Br. 37-38.
Turning to Netflix's argument based on Pyle and Marusi, Netflix argues the Board overlooked its alternative argument relying on Marusi (and not Pyle) for the step of filtering assets. We do not agree; the Board could not have overlooked that argument because it was never there. The Board reasonably understood this portion of the petition to rely on Pyle for filtering a list of assets and use Marusi for its database teachings.
To understand how Netflix's petition relied on Marusi, it is helpful to revisit claim 1's filtering limitation (emphasis added):
filtering the list of assets using the at least one device capability using the playback server system, wherein the playback server system maintains a database of product identifiers and associated device capabilities.
Netflix initially relied on just Pyle to meet all the features of this limitation, by arguing that Pyle's selection of a pre-existing manifest was the same thing as "filtering the list of assets," and then contending that it would be "obvious" for Pyle to include a server with a database of product identifiers and device capabilities: "a [skilled artisan] would have found it obvious looking at the teachings of Pyle that the product identifiers and associated device capabilities be maintained by the server, such as in a database." J.A. 6047.
As a backup argument, Netflix then argued that to the extent Pyle "does not satisfy this limitation, it would have been obvious to combine the teachings of Pyle with Ma-rusi." J.A. 6048. The petition then presented a drawn-out pastiche of block quotes from Marusi, but highlighted particular, select language in bold to signal what the reader should focus on: Marusi's "database" which stores "terminal capabilities information." J.A. 6048-49. Netflix then concluded its argument with its sole articulation for which elements it relied on from Pyle and Marusi for the filtering limitation: "A [skilled artisan] would have been motivated to employ a known component (Marusi's database of identifiers and associated assets) in a predictable way (for Pyle to filter assets to obtain a subset in a format compatible with the requesting device)." J.A. 6049. The Board acknowledged that, at one point in the petition, Netflix stated that "Marusi filters from a database," but nevertheless reasonably concluded that, for the Pyle-Marusi combination, Netflix "maintains reliance upon Pyle for filtering," given Netflix's concluding sentence mapping Pyle to filtering and Marusi to the database component, and the bolded language in block quotes highlighting Marusi's database. '720 Patent Decision, 2021 WL 4859625, at *14. Moreover, we note that the petition's lone "Marusi filters" statement never suggests or explains how Marusi filters "a list of assets" as claimed. In sum, we reject Netflix's contention that the Board was "absurd" in how it understood Netflix's argument. See Appellant's Br. at 45.
We do not wish to discourage petitioners from making arguments in the alternative. Rather, we emphasize that it is the petitioner's burden to make clear when alternative arguments are being presented and to sufficiently expound on each one. The Board should not have to work as hard as Netflix wants to identify all arguments fairly presented in a petition.
II. "Retrieving" Limitation
Turning to the "retrieving . . . a list of assets" limitation, Netflix argues that the Board wrongfully viewed Lewis as the basis for a § 103 obviousness challenge that required modification to Lewis, when, in Netflix's view, the petition argued Lewis disclosed this limitation. Appellant's Br. 53-54. Specifically, Netflix argues that the petition's "use of language like 'a [skilled artisan] would have found it obvious to retrieve a list of assets' does not suggest that Lewis requires any modification." Appellant's Br. 61. Patent lawyers would be surprised by this statement. There is no word in patent law that sends a more unmistakable signal that § 103, rather than § 102, is being invoked than the word "obvious." See Connell v. Sears, Roebuck &Co., 722 F.2d 1542, 1548 (Fed. Cir. 1983) ("[T]he need to determine obviousness presumes anticipation is lacking"); see also Jones v. Hardy, 727 F.2d 1524, 1529 (Fed. Cir. 1984) ("[T]hough anticipation is the epitome of obviousness, [they] are separate and distinct concepts."). Nor did the Board err in requiring Netflix to articulate a modification of Lewis. As the Board explained, Netflix only relied on Lewis's disclosure of a result-generating a dynamic manifest file containing a list of URLs-but had failed to explain how Lewis's disclosure of this result necessarily would have rendered obvious an intermediate process step (retrieving a list of assets). '720 Patent Decision, 2021 WL 4859625, at *19.
Netflix contends that the Board misunderstood its argument, asserting the petition contemplated that creating a manifest containing URLs pointing to the selected processed video necessarily involves first retrieving a list of assets that reside in the manifest itself. Appellant's Br. 60. In other words, according to Netflix, creating the manifest file and retrieving a list of assets are one and the same. We conclude that it was eminently reasonable for the Board to interpret Netflix's petition differently because (1) the petition never made this sort of part-and-parcel inherency argument and (2) the petition's use of the word "obvious" suggests that retrieving a list of assets and creating a manifest are not the same. See J.A. 6070 (arguing that "[a] [skilled artisan] would have found it obvious to retrieve a list of assets" because Lewis teaches creation of a dynamic manifest). Netflix's petition never established or suggested that Lewis's creation of a manifest necessarily involves retrieval of a list of assets. '720 Patent Decision, 2021 WL 4859625, at *19.
The dissent finds that Netflix sufficiently presented this argument because "if a method produces or uses a list, then the list needs to be retrieved. A list cannot be used without retrieving it." Dissent at 7. That reasoning could have been helpful to avoid forfeiture if Netflix had offered it in its petition, but it did not. Moreover, in these patent claims, the generated list ("generating a top level index describing each asset") is a different, smaller list than the retrieved list, due to the intervening "filtering the list of assets" step. See '720 Patent at claim 1. Thus, if Netflix intended to argue that Lewis teaches (as opposed to renders obvious) the retrieving limitation, some reasoning had to be presented for why Lewis necessarily retrieves an initially larger list that would then go through a filtering process to arrive at a final, generated list. This absence of needed explanation reinforces our view that Netflix never raised this potential argument in its petition, and the Board correctly declined to assume that such an argument had been raised.
III. "Generating" Limitation
Finally, turning to the "generating . . . a top level index file" limitation in the '515 patent, the Board correctly determined that Netflix's reply improperly argued for the first time that Pyle's creation of a new manifest satisfied the "generating" limitation. Netflix argues the Board overlooked its reliance on Pyle's new manifest, but Netflix resorts to pointing to sections of its petition devoted to its analysis for the "retrieving" and "filtering" limitations. Appellant's Br. 46-47. Given the plain language of the '515 patent petition and the substantially more developed analysis of an analogous limitation in the '720 patent petition, we disagree with Netflix.
The '515 patent petition plainly shows that Netflix argued the "generating" limitation was met by Pyle's transmission of a manifest into storage, which is entirely different from Pyle's creation of a new manifest. Netflix argued: "A [skilled artisan] would have understood that the manifest file sent in response to a request for content is generated because the manifest file must be placed in memory for transmission to a client device." J.A. 7043 (emphasis added); see also J.A. 7042 ("Pyle further teaches transmitting a manifest file to the requesting device in response to the request for content."). We agree with the Board and conclude that Pyle's transmission of a manifest into storage was the only argument presented in the petition. '515 Patent Decision, 2021 WL 4859708 at *13. Netflix points to various parts of the petition-including citations to Pyle's Figure 4 and to passages in Pyle describing creation of a new manifest-that allegedly describe how Pyle's new manifest satisfies the "generating" limitations. Appellant's Br. 48. While these portions of Pyle describe a new manifest, Netflix only referred to these portions in citations to support the statement that "Pyle further teaches transmitting a manifest file to the requesting device in response to a request for content." J.A. 7042. It was reasonable for the Board to consider only the theory encapsulated in the plain words of this statement and decline to discern another theory from the citations alone.
Moreover, in contrast to the '515 patent petition, the '720 patent petition clearly argued that Pyle's new manifest met the '720 patent's corresponding "generating" limitation. In the '720 patent petition, Netflix argued, "Pyle teaches generating manifest files that are optimized . . . using the composition component 210 to create new manifests," J.A. 6051 (emphasis added), and "[a] [skilled artisan] would have found it obvious to generate a top level index file . . . because as explained for limitations 1(c) and 1(d), Pyle teaches . . . creating new manifest files" J.A. 6050 (emphasis added). Comparing these arguments to those of the '515 patent petition, we agree with the Board that, "[Netflix] knew how to raise the argument (that creating a new manifest teaches generating a top level index file) if [Netflix] chose to do so." '515 Patent Decision, 2021 WL 4859708 at *14. Thus, the Board did not abuse its discretion in finding that Netflix improperly argued that Pyle's new manifest met the "generating" limitation for the first time in its reply.
CONCLUSION
For the reasons stated above, all of Netflix's arguments on appeal were not presented below and are thus forfeited. We affirm the Board's decision upholding the challenged claims of the '720 patent and '515 patent.
AFFIRMED
DYK, Circuit Judge, dissenting.
Contrary to the majority, I think that the Board improperly ignored arguments that Netflix raised in its petition. In IPRs we are confronted with situations in which parties without basis argue that arguments were presented to and ignored by the Board, and equally with situations in which the Board ignores arguments that were properly preserved. In my view this case, at least in part, falls in the latter category, and the Board improperly declined to consider arguments that Netflix raised in its petition. I respectfully dissent from the majority's contrary conclusion.
See, e.g., Ericsson Inc. v. Intell. Ventures I LLC, 901 F.3d 1374, 1379-80 (Fed. Cir. 2018) (holding that the Board erred by "parsing [petitioner's] arguments . . . with too fine of a filter"); Donner Tech., LLC v. Pro Stage Gear, LLC, 979 F.3d 1353, 1359 (Fed. Cir. 2020) (remanding because it was "unclear whether the Board meaningfully considered all of [petitioner's] arguments and evidence"); Netflix, Inc. v. DivX, LLC, No. 2022-1043, 2023 WL 3115576, at *5 (Fed. Cir. Apr. 27, 2023) (holding that "the Board erred in not adequately assessing Netflix's arguments regarding limitation 1[c]"); Netflix, Inc. v. DivX, LLC, 80 F.4th 1352, 1358 (Fed. Cir. 2023) (holding that "the Board abused its discretion in requiring Netflix to explicitly identify the field of endeavor using specific language"); Provisur Techs., Inc. v. Weber, Inc., 50 F.4th 117, 123-24 (Fed. Cir. 2022); Microsoft Corp. v. Parallel Networks Licensing, LLC, 715 Fed.Appx. 1013, 1021 (Fed. Cir. Dec. 1, 2017); Uniloc 2017 LLC v. Facebook, Inc., No. 20192159, 2021 WL 5370480, at *8-9 (Fed. Cir. Nov. 18, 2021); AMC Multi-Cinema, Inc. v. Fall Line Pats., LLC, No. 20211051, 2021 WL 4470062, at *7 (Fed. Cir. Sept. 30, 2021).
I
Netflix challenged claims of both the '720 and '515 patents, which are directed at methods for generating "a top level index file" for use with media streaming. In media streaming, content (such as a movie) is typically stored as separate "streams" suitable for different device capabilities. '720 patent col. 1 ll. 35-38. Thus, a playback device must be able to select an appropriate media file from the available streams. The specification of the '720 and '515 patents discusses "methods for automatically generating top level index files," which "describe[] the location and content of container files containing streams of media." '720 patent col. 6 ll. 36-43. Claim 1 of both patents recites a method comprising "receiving" a request for content, "retrieving" a list of assets associated with the content, "filtering" the list of assets using device capability, "generating" "a top level index file describing each asset in the filtered list of assets," and "sending the top level index file" to the playback device.
In the IPRs, Netflix relied on the Pyle and Lewis publications as prior art. Pyle describes the use of a "manifest," which is a file that "describe[s] the locations of various content segments." '720 Patent Decision, 2021 WL 4859625, at *8 (quoting Pyle). Pyle's manifests can be "optimized in connection with . . . particular devices or capabilities thereof." Id. at *9. In one embodiment, a manifest can be "selected . . . from the set of available manifests" and in another a "new manifest" can be composed. Id. Lewis describes an embodiment wherein video content is processed and customized according to particular client device parameters, and "when [a] client device requests video content represented by stored video segments, [a] dynamic manifest file server may provide [a] manifest file." Petition for Inter Partes Review, Hulu, LLC v. DivX, LLC, at 63 (No. IPR2020-00647) ("Petition") (quoting Lewis) (reference numbers omitted).
II
Netflix argues that the Board in three instances erroneously determined that the petitioner had not made arguments that in fact were made in the petition. I agree with the majority as to one of these (though I view it as a close question)-that the Board did not abuse its discretion in finding that Netflix failed to raise an argument that the "generating" limitation of the '515 patent was met by Pyle's "new manifest." But I disagree with the majority that Netflix failed to make arguments in two other respects-regarding the "filtering" limitation of the '720 patent and the "retrieving" limitations of both the '720 and '515 patents. In my view, the Board was required to address the merits of these arguments. A. "Filtering" Limitation of the '720 Patent
In its final written decision, the Board concluded that the "filtering" limitation was not met by the selection of an existing manifest. The Board found it "dispositive" that Netflix relied on Pyle's selection of an existing manifest, and not the creation of a new manifest, to meet the "filtering" limitation. '720 Patent Decision, 2021 WL 4859625, at *7, *15. The majority approves of the Board's conclusion. Maj. Op. 6, 10. But, contrary to the majority, Maj. Op. 10, Netflix's petition plainly presented an alternative argument that Pyle, by generating a new manifest, met the '720 patent's "filtering" limitation, and the Board's institution decision credited this argument.
In the petition, under the heading "filtering the list of assets," petitioner explained that Pyle's "new manifest" teaches filtering. Petition, at 37 (emphasis added). The petition argued that Pyle "teaches filtering the lists of assets using at least one device capability" in one of two ways: (a) "Pyle teaches the use of new manifest files that can be created based upon particular device(s) or capabilities [new manifests]" and (b) "Pyle also teaches keeping sets of different manifest[] files [existing manifests]." Id. at 38 (emphasis added). In the paragraph discussing new manifest files, Netflix quoted from Pyle to show how the new manifest can be optimized based on device capabilities, network conditions, and user preferences. Id. Here, Netflix presented its "new manifest" argument on the very page the majority cites as lacking it. Maj. Op. 11.
Contrary to the majority, Maj. Op. 12, in its institution decision the Board explicitly recognized the argument that Pyle's new manifest meets the "filtering" limitation. Under the heading for "Limitation 1[d]," which "recites 'filtering the list of assets . . .,'" the Board recognized that "[f]irst, Petitioner contends Pyle 'teaches the use of new manifest files that can be created based upon particular device(s) or capabilities,'" and the Board quoted Pyle's "new manifest" disclosure. No. IPR2020-00647, Paper No. 9 at 28-29 (P.T.A.B. Oct. 1. 2020) ("Institution Decision") (emphasis in original).
Moreover, other aspects of the petition and the institution decision reiterate the conclusion that the new manifest teaches "filtering." Under the heading for the "generating" limitation, which follows the "filtering" limitation, Netflix explained at length that "Pyle is teaching the generating of new manifest files by filtering the available list of assets to include only a subset of assets based upon the product identifier information," and cited expert testimony in support. Petition, at 44 (emphasis added). Thus, Netflix's petition further tied the new manifest to the "filtering" limitation. The institution decision similarly demonstrates the Board's understanding of Netflix's position. The Board reasoned that, although DivX responded to the "existing manifest" argument in its preliminary response, the petition also raised a "new manifest" argument that DivX ignored, which was sufficient for institution:
Additionally, Patent Owner's argument ignores Petitioner's contention that one of ordinary skill in the art would understand Pyle to teach generating new manifest files by "filtering the available list of assets to include only a subset of assets based upon the product identifier information, including for example the capabilities of the device." On this record, Petitioner's position is reasonable and supported by the '720 patent and Dr. Reader's testimony.
Second, Patent Owner contends that Petitioner fails to explain how Pyle's manifest file "describ[es] each asset in the filtered list." Specifically, Patent Owner contends that Pyle "teaches a system for streaming content based on manifests that
already exist when the request for content is submitted" and does not teach composing a new manifest in response to or based on a request for content.
Patent Owner's argument, however, fails to detract from Petitioner's position, explained above, regarding filtering the available list of assets and Pyle's teaching regarding optimizing a new manifest based on particular device capabilities.Institution Decision, at 35 (citations and parentheticals omitted) (emphasis in original).
In its final written decision the Board backtracked, finding instead that Netflix's petition presented only one argument-that Pyle teaches filtering by selecting an existing manifest. The Board dismissed Netflix's "new manifest" position, judging that Netflix presented only a "single-paragraph discussion," which "does not (a) expressly assert that Pyle's creation of a new manifest involves 'filtering' or (b) explain how Pyle's new optimized manifest teaches filtering." '720 Patent Decision, 2021 WL 4859625, at *14 (citations omitted) (emphasis in original). The final written decision does not acknowledge the petition's express new manifest argument under the filtering heading or the Board's express statements in its institution decision regarding Pyle's teaching of "filtering" by creating new manifest files. See Institution Decision, at 28-29, 35.
When the Board recognizes two arguments in its institution decision but addresses only one on the merits, a remand is required. Google LLC v. Conversant Wireless Licensing S.A.R.L., 753 Fed.Appx. 890, 895 (Fed. Cir. 2018) ("In its institution decision, the Board clearly and correctly understood that Petitioners made two distinct arguments in their petition Yet, in its final written decision, the Board concluded that Petitioners' arguments failed without even discussing Petitioners' [first] argument."). Here, the Board recognized in the institution decision the alternative argument that Pyle teaches that the "new manifest" "generates top level index files by filtering assets," yet did not address this aspect of the institution decision in the final written decision. Institution Decision, at 35 (quotation omitted); see also id. at 28-29. The Board cannot contradict itself without explanation. The majority does not recognize the evident contradiction between the Board's institution decision and its final written decision or the Board's failure to explain the contradiction in its final written decision.
I would remand for the Board to consider the merits of Netflix's "new manifest" argument for claims 1-5 and 12 of the '720 patent.
B. "Retrieving" Limitations of the '720 and '515 Patents
The "retrieving" limitation of both the '720 and '515 patents recites "retrieving, using the playback server [system,] a list of assets associated with the identified piece of content ...." Netflix relied on Lewis as demonstrating this limitation. The Board concluded that the argument failed because Netflix did not show how it would be obvious to modify Lewis to retrieve a list of assets:
Fundamentally, Petitioner fails to provide any reason in the Petition as to why one of ordinary skill in the art would have found it obvious to retrieve a list of assets based on Lewis. Instead of providing a reason as to why retrieving a list of assets would have been obvious, Petitioner points to a result achieved by Lewis-generating a dynamic manifest file that contains a list of URLs to container files containing content.'720 Patent Decision, 2021 WL 4859625, at *19; '515 Patent Decision, 2021 WL 4859708, at *22 (emphasis in original). The majority approves this decision. Maj. Op. 15. However, if a method produces or uses a list, then the list needs to be retrieved. A list cannot be used without retrieving it.
Netflix adequately raised an argument that the "retrieving" limitation was met without modifying Lewis.
In both petitions, Netflix argued:
A POSITA would have found it obvious to retrieve a list of assets using the playback server system because Lewis teaches that the manifest file server and rule resolution server generate[s] a dynamic file manifest, and Lewis teaches that a dynamic manifest file contains a list of URLs to container files containing content In other words, because a manifest file is a file that contains a list of URLs that point to container files, a POSITA would have found it obvious that the manifest file server and rule resolution server retrieve a list of assets because the manifest file server and rule resolution server ultimately produce a list of assets to the client device in the form of a manifest file.Petition, 62-63; J.A. 7070 (citations omitted) (emphasis added).
Netflix explained how Lewis alone teaches the "retrieving" limitations because "a POSITA would have found it obvious that the manifest file server and rule resolution server retrieve a list of assets." Petition, 62-63; J.A. 7070. Netflix's argument was not that a POSITA would find it obvious to modify Lewis to retrieve a list of assets. Rather, it was that a "POSITA would have found it obvious" that Lewis already did "retrieve a list of assets" as it produced its manifest file to the client. In Netflix's argument, "retrieve" is in the present tense, and there is no hint that Netflix thought Lewis needed to be modified.
III
Contrary to the Board's conclusions, Netflix's petitions adequately raised arguments that the "filtering" limitation in the challenged claims of the '720 patent (based on Pyle) and the "retrieving" limitations in the challenged claims of both patents (based on Lewis) were obvious. I would remand for the Board to consider Netflix's arguments regarding the '720 patent's "filtering" limitation and the "retrieving" limitations with respect to claims 1-6 and 12 of the '720 patent and claims 1-6, 8-10, and 13 of the '515 patent. I respectfully dissent.