[00:00:00] Speaker 02: That brings us to our final case this morning, number 22443, Netflix Inc. [00:00:07] Speaker 02: versus DivX LLC. [00:00:11] Speaker 02: Okay, Mr. Saunders. [00:00:13] Speaker 05: May I please the court? [00:00:15] Speaker 05: The board's reading of Netflix arguments was clearly unreasonable and based on a methodological error that pervaded its opinion. [00:00:23] Speaker 01: Do we review all of your arguments under an abusive discretion standard? [00:00:26] Speaker 05: They are reviewed under an abusive discretion standard, but under that standard we have time and again the board is looking at statements in isolation without considering the surrounding sentences or the relationships between the arguments. [00:00:42] Speaker 05: For example, in its reading of our new manifest filtering argument. [00:00:48] Speaker 05: That argument is the lead argument under the filtering section. [00:00:52] Speaker 05: The sentence that leads into that passage talks about filtering. [00:00:59] Speaker 05: It directly quotes from Pyle's teaching about optimizing the new manifest based on a number of characteristics, the same characteristics we see in the pattern. [00:01:08] Speaker 01: You never connect the creation of a new manifest 422 to the word filtering a list of asset in the same way that you do at the bottom of A6046 when it comes to selecting an appropriate manifest from a set of manifests [00:01:28] Speaker 01: as filtering a list of assets. [00:01:31] Speaker 01: Is that fair to say? [00:01:32] Speaker 05: No, I respectfully disagree. [00:01:35] Speaker 01: Well, let me put a finer point on it then. [00:01:39] Speaker 01: You don't ever say the word filter in connection with the block quote that talks about creation of a new manifest. [00:01:48] Speaker 01: Is that fair to say? [00:01:50] Speaker 05: No, because it's in the sentence right beforehand. [00:01:54] Speaker 05: So the lead into that, 6046, [00:01:57] Speaker 05: says. [00:01:58] Speaker 01: Well, the lead into it says, Pyle teaches the use of new manifest files that can be created based upon particular devices or capabilities. [00:02:07] Speaker 01: Then comes your block quote. [00:02:09] Speaker 05: So right above that. [00:02:10] Speaker 05: Pyle further teaches filtering the list of assets using at least one device capability. [00:02:18] Speaker 05: Pyle teaches the use of new manifest files that can be created based upon particular devices [00:02:24] Speaker 05: or capabilities. [00:02:26] Speaker 01: It's a different paragraph, though, right? [00:02:29] Speaker 01: Sure. [00:02:30] Speaker 01: And then you didn't include it in the joint appendix, but when you actually looked at the prior page of the petition, you're talking about how, as explained for limitations 1B and 1C, Pyle teaches a, quote, manifest component, end quote, configured to, quote, maintain multiple manifests, end quote. [00:02:52] Speaker 01: That is to say, [00:02:53] Speaker 01: We're back to talking about selecting one manifest from multiple manifests, not talking about the creation of a new manifest when it comes to filtering a list of assets. [00:03:04] Speaker 05: No, because the theory here on the multiple manifests is they are serving as the source of information that would then be going into the new manifest. [00:03:15] Speaker 05: And then there's an alternative theory that you could have the databases from Marusi come in and supply that information. [00:03:23] Speaker 01: So here's the problem I have. [00:03:28] Speaker 01: I mean, whether I like it or not, I've been appointed to a life sentence of reviewing IPRs on appeal. [00:03:36] Speaker 01: And so I've looked at over 100. [00:03:38] Speaker 02: That's not the only thing we do. [00:03:42] Speaker 01: Sometimes it feels like it. [00:03:43] Speaker 01: And so I've read well over 100 IPR petitions now, maybe closer to 200. [00:03:50] Speaker 01: And when a petitioner in a petition [00:03:53] Speaker 01: makes alternative arguments about the prior art meeting a given limitation and acclaim. [00:04:00] Speaker 01: They'll always give you a signal. [00:04:03] Speaker 01: Alternatively, reference Y likewise teaches this limitation, or in addition, reference Y teaches this limitation. [00:04:11] Speaker 01: Moreover, reference Y also teaches this limitation. [00:04:15] Speaker 01: I don't see anything like this. [00:04:16] Speaker 01: I see [00:04:18] Speaker 01: it being very reasonable for the board to look at your block quote about the new manifest as an introduction and then continues on to what's the real heart of the argument here at A-6046, which is the idea of selecting an appropriate manifest from a list of manifests as being the quote-unquote filtered list of assets as you appropriately and clearly and cleanly connected at the bottom of A-6046. [00:04:44] Speaker 05: Three quick things, Your Honor. [00:04:48] Speaker 05: then it would make no sense what is the new manifest argument doing in there, because it's not connected. [00:04:54] Speaker 05: Second, the lead-in to the next paragraph is Pyle also teaches keeping sets. [00:05:02] Speaker 05: And then third, what you have to remember is that the board's holding here is that there is a mismatch between your theory on filtering and your theory on generating. [00:05:11] Speaker 05: And if you look in the generating discussion, [00:05:16] Speaker 05: That begins on 6.0.5.0 by referring back to what we just said on filtering. [00:05:24] Speaker 05: And then it says, it's explained for limitations 1C and 1D. [00:05:27] Speaker 05: Pyle teaches both creating new manifest files as well as, it has the separate selection. [00:05:34] Speaker 05: And then it finishes the generating discussion. [00:05:37] Speaker 05: So again, holding is there's a mismatch between generating and filtering because your filtering isn't about new manifest. [00:05:43] Speaker 05: The generating discussion finishes on 6.0.5.2 [00:05:47] Speaker 05: by saying, Pyle is teaching the generating of new manifests by filtering the available list of assets to include only a subset of assets based on the product identifier information. [00:05:59] Speaker 01: But there's another problem with this new theory. [00:06:02] Speaker 01: And that is going back to the prior claim limitation of retrieving a list of assets wherein each asset is a different stream associated with a piece of content. [00:06:14] Speaker 01: And when you go back and look at your petition on the retrieving a list of assets, it does a long song and dance about how there are a whole host of manifests, and then each manifest is an asset, and each manifest is a different stream of content. [00:06:35] Speaker 01: And so you've conditioned the board through this long section on the retrieving list of assets to understand that you're contemplating your reliance on the list of manifests as being the respective list of assets that's being retrieved. [00:06:50] Speaker 01: So then when we get to the filtering of the list of assets, it makes perfect sense to read the text of your section as is, which is to say selecting one of the manifests from that list of manifests as being a filtering of the assets. [00:07:05] Speaker 05: I'm not, I mean, there is, there are two theories. [00:07:08] Speaker 01: Selecting is still the... Right, so I guess what I'm wondering, if I can finally ask the question, where in the retrieving the list of assets section of your petition did you talk about how whatever individual components that make up a new manifest can also be retrieving a list of assets? [00:07:28] Speaker 05: Well, with respect, there's separate limitations. [00:07:32] Speaker 05: So it is talking about an individual manifest or all of the manifests collectively can be the list of assets. [00:07:40] Speaker 05: And then we have a teaching here that's saying, or you can create a new manifest that's optimized based on these characteristics. [00:07:48] Speaker 05: And the theory is, where is that information coming from? [00:07:52] Speaker 05: that information's coming from because you have lists already. [00:07:55] Speaker 01: I guess what I'm highlighting is there's a collision of internal inconsistency in the theory presented anyway. [00:08:04] Speaker 01: Whether we link up the filtering of assets to be consistent with the generating of a top-level index [00:08:12] Speaker 01: you nevertheless still have a problem on the retrieving the list of assets side, because you never continue to thread through that limitation that generating a manifest is somehow included in that as a precursor step of retrieving a list of assets that will eventually make up a new manifest. [00:08:33] Speaker 05: Well first a procedural point and then a substantive one. [00:08:36] Speaker 05: Procedural point, they made both these arguments. [00:08:38] Speaker 05: There's a mismatch between filtering and generating and a mismatch between retrieving and filtering [00:08:43] Speaker 05: board did not rule on that first one. [00:08:45] Speaker 05: So under Chenery, I don't think that is before this court. [00:08:50] Speaker 05: And there isn't a mismatch there, again, because the information for the new manifest [00:08:57] Speaker 05: can be coming from those pre-existing sources. [00:09:02] Speaker 05: That's where you have these lists that are available. [00:09:05] Speaker 05: But also in the alternative, the motivation to combine, we talk about reading this all in context, the whole section begins by talking about Marusi [00:09:14] Speaker 05: and the combination there in terms of its use of databases. [00:09:21] Speaker 05: The theory isn't solely that Pyle's new manifest can be created based on [00:09:29] Speaker 05: the existing information in the existing manifest, there's also an alternative theory that says you'd take Marusi's database teachings and all of that background information. [00:09:39] Speaker 05: We're talking about very, very simple stuff. [00:09:42] Speaker 05: The basic information about these video files would be saved in that database of Marusi and then could be used when creating this new re-optimized manifest. [00:09:54] Speaker 05: And on this issue of whether there are separate arguments, I mean, if there is any doubt in the petition, [00:09:59] Speaker 05: We then come back in the reply brief and say, we're talking about a new manifest or selecting. [00:10:07] Speaker 05: There's a colloquy about this at the oral hearing at 6528 through 6530, in which the board's saying, do I understand you to be having two theories of filtering? [00:10:18] Speaker 05: And the answer is yes, and that is all laid out and explained. [00:10:23] Speaker 05: So to ultimately have a holding that say, we're not going to consider this because your sole theory on filtering is selection, and that's mismatched with your generating limitation, it is a clearly unreasonable reading of the petition. [00:10:38] Speaker 05: All we're asking is to be able to have a ruling on the merits on that and not to have the board based on a mistake about what we're arguing. [00:10:46] Speaker 01: How is creating a new manifest filtering a list of assets? [00:10:53] Speaker 05: Because you're creating a new manifest that is tailored for device capabilities language. [00:11:00] Speaker 01: The claim language says wherein an asset or each asset is a different stream of the video content. [00:11:11] Speaker 01: So if you're going to be doing this filtering, you've got to be doing it on a stream-by-stream basis. [00:11:20] Speaker 01: And I don't see how creating a new manifest is doing that. [00:11:25] Speaker 05: Well, you're filtering, but you're not filtering out every stream. [00:11:30] Speaker 05: You finish with a top-level index file that will have multiple streams so the device can select. [00:11:37] Speaker 05: For example, if it has a bandwidth problem, [00:11:39] Speaker 05: and then it can select. [00:11:41] Speaker 01: Yeah, I'm just trying to understand how does creating a new manifest mean that you are filtering assets when assets are different streams of video content. [00:11:53] Speaker 01: For example, I want to watch The Lion King. [00:11:58] Speaker 01: There's going to be a bunch of different streams for that Lion King. [00:12:03] Speaker 01: And I'm going to get one of them, but it'll be dependent on whether I want to watch it on an iPhone or HD television or standard television, whether I want to watch it in English, Japanese, German, what the network connection is, what form factor we're working with, all of those different things. [00:12:22] Speaker 01: And all of those different possibilities make up the different streams, make up the different manifests, make up the different assets. [00:12:30] Speaker 01: And so I don't understand, I'm completely baffled by how the creation of a new manifest, the creation of a given stream is actually somehow filtering one stream from another. [00:12:44] Speaker 05: So the new manifest is being optimized based on the very characteristics you're talking about. [00:12:52] Speaker 05: So if you have [00:12:53] Speaker 05: your pre-existing store of information, whether it's in the pile manifest. [00:12:58] Speaker 01: Do you agree that an asset is a stream? [00:13:02] Speaker 05: An asset, yes, could be. [00:13:04] Speaker 01: It is a stream. [00:13:05] Speaker 01: It's a stream of content, of demand. [00:13:11] Speaker 05: I think it depends on what you mean by stream. [00:13:13] Speaker 05: In pile, it's broken down even further, because you're looking at additional track levels. [00:13:20] Speaker 05: An asset could be some piece of content. [00:13:22] Speaker 05: whether it's audio, video, in one format, in a different format. [00:13:28] Speaker 01: So how does creating a stream mean you're filtering one stream from the rest of the streams? [00:13:34] Speaker 01: That's different. [00:13:36] Speaker 05: Well, but if you have a list, so imagine that you take the list of cases that you see in the table of authorities [00:13:44] Speaker 05: And then you, on another piece of paper, looking at that list, say, OK, these are the four cases that are most relevant to me, because they're the precedential cases. [00:13:53] Speaker 05: They're not the non-precedential cases. [00:13:57] Speaker 05: That would be an act of filtering. [00:13:58] Speaker 05: You've taken a larger universe of information. [00:14:01] Speaker 05: You've created a new document. [00:14:04] Speaker 05: But it is optimized for your purpose. [00:14:07] Speaker 05: It contains the subset. [00:14:08] Speaker 05: So when it says, create a new manifest that's optimized for things like, is it French language? [00:14:14] Speaker 05: The clear implication is you're not interested in the German language and not interested in the English language. [00:14:21] Speaker 05: You would be filtering it by focusing on that subset. [00:14:26] Speaker 05: And the patent itself is very broad. [00:14:30] Speaker 05: You're not going to see a narrower conception of filtering in the patent. [00:14:35] Speaker 05: It essentially says anything, any technique can be used to filter. [00:14:41] Speaker 05: So having cast a broad net, I don't think we've [00:14:44] Speaker 05: hard to claim construction argument about that not being filtering. [00:14:49] Speaker 03: Mr. Saunders, as I understand it, the new manifest is something that is created by the content provider. [00:14:58] Speaker 03: Am I [00:14:59] Speaker 05: Well, the board disagreed with that. [00:15:01] Speaker 05: It didn't address that in the file written decision, but in the institution decision, 6208 through 6209, the board said, we do not read that portion of Pile as limiting the composition of new manifests to content providers. [00:15:14] Speaker 05: And that was correct. [00:15:16] Speaker 05: It pointed out that that was an exemplary language. [00:15:21] Speaker 05: For example, the content provider may create it, but it's not limited to the content. [00:15:25] Speaker 03: So your position is that even the new manifest is somehow created by retrieving a list of assets identified with some piece of content that the [00:15:44] Speaker 03: at the playback server requested? [00:15:48] Speaker 05: Yes, absolutely. [00:15:49] Speaker 05: And we've relied on the disclosure and figure four in the diagram shows the request coming in and what can come out is either selection. [00:15:56] Speaker 01: What are the assets that are being retrieved in that instance? [00:16:00] Speaker 05: In that instance we have, if it's based on pile alone, the assets that are being retrieved is the information you're taking in that the system already has from its existing manifest. [00:16:12] Speaker 01: I didn't get an actual more detailed answer. [00:16:17] Speaker 01: When you are creating a new manifest in Pyle, what are the list of assets that are being first retrieved and then filtered? [00:16:27] Speaker 05: The assets themselves are the underlying track, the audio and video tracks in Pyle. [00:16:35] Speaker 05: Those assets exist in a list form within the existing manifests, or it would be obvious in combination with Merusi and its database teaching to have that information in a database form. [00:16:49] Speaker 05: In that form, you have sort of the entire universe of what you might need, and you are optimizing that for the particular device by selecting a subset. [00:17:01] Speaker 05: So you've retrieved the list of assets, and you are then filtering it and generating the new manifest. [00:17:10] Speaker 05: Now, all this really collapses into one step. [00:17:12] Speaker 05: We have something that's sort of breaking it down in granular detail, much in the way that you would have if you say, you know, pick up my pen, point it towards the paper, make contact with the paper, produce intelligible writing, you might have a prior art reference that just says, do handwriting on a piece of paper. [00:17:29] Speaker 05: I think a lot of the board's confusion here is trying to say what are the relationships between these and which buckets do we put these in when it is really a very simple step that's coming here in terms of generating, taking in that information and filtering. [00:17:48] Speaker 02: Could you briefly discuss the pile Marusi combination in the board's statement that somehow you weren't relying on Marusi for full time? [00:18:01] Speaker 05: Right, I think the board, this is sort of what I mean by not looking at the context. [00:18:05] Speaker 05: The board is hung up on the fact that it talks about using Marusi's databases, identifiers, and lists for pile to filter the assets. [00:18:16] Speaker 05: The sentence right before the sentence the board's relying on, and this is on 6049, says, Merusi also teaches identifying a subset of multimedia content files having a specific digital format supported by the requesting device. [00:18:34] Speaker 05: Above that on the same page, there's a block quote from Merusi about its matching unit 150. [00:18:40] Speaker 05: And the whole point of this was to be making an alternative argument. [00:18:43] Speaker 05: The board is saying, [00:18:44] Speaker 05: We don't agree with you on pile filtering, so you lose on the combination of Merusi and pile. [00:18:50] Speaker 05: What we're saying is no, in the context of the combination, it's not pile unchanged doing the filtering, it's pile in the context of bringing in these database teachings from Merusi and what it has there. [00:19:06] Speaker 01: I guess why wouldn't it be reasonable to read this whole Merusi discussion at 6048 to 6049 in the context of [00:19:14] Speaker 01: what is said at the bottom half of 6047, where you're talking about how a posita would have found it obvious, looking at the teaching for Pyle, that the product identifiers and associated device capabilities be maintained by the server, such as in a database, block, quote, and then to the extent that Vicks argues that Pyle does not satisfy this limitation, it would be obvious to combine the teachings of Pyle with Marusi. [00:19:42] Speaker 01: And then we finally get to the bottom and then you talk about how the motivated to employ a known component moves the database in a predictable way, which is for pile to filter the assets. [00:19:53] Speaker 01: And so in that sense, this whole discussion is about the database in that context. [00:19:59] Speaker 05: Well, it's a discussion of two responses. [00:20:04] Speaker 05: It's a discussion of Marusi and its database that includes the sentence I read just before, which is how Marusi is using that database, which is to identify the subset of assets. [00:20:14] Speaker 05: There's the disclosure of Marusi's matching unit. [00:20:18] Speaker 01: But even if it were just... I guess on these two pages, are you saying there are two alternative [00:20:24] Speaker 01: Theories of combination of Pyle and Marusi. [00:20:27] Speaker 01: Theory number one would be Pyle's filtering with Marusi's database. [00:20:33] Speaker 01: And then theory two, just Marusi's filtering with Marusi's database. [00:20:41] Speaker 05: When we talk about piles filtering, these are all interrelated, so the question is, what is pile filtering at that point? [00:20:48] Speaker 01: I'm just trying to figure out what the board was supposed to figure out from these pages. [00:20:53] Speaker 01: Was it supposed to figure out that there were two alternative theories in the combination of pile and Marussi? [00:20:59] Speaker 05: Right, they're supposed to think that [00:21:02] Speaker 05: that it is Marousi and its associated teachings that have been brought in here. [00:21:07] Speaker 01: And when it says for pile, OK, so I just need a yes or no answer. [00:21:12] Speaker 01: Are these two pages representing that Marousi teaches both filtering the list of assets and the database? [00:21:19] Speaker 05: The discussion on these two pages has both of those teachings. [00:21:22] Speaker 01: So you don't even need pile, then, for meeting this particular limitation. [00:21:27] Speaker 01: That's what these two pages are communicating. [00:21:29] Speaker 05: You wouldn't need pile, but it can be that it's argued in combination with pile, but it's not pile unmodified. [00:21:37] Speaker 05: It's pile in combination. [00:21:38] Speaker 01: Well, of course, with Marusi's database, that one theory comes through clearly. [00:21:44] Speaker 01: You're now arguing that also lurking in here is the idea that Marusi was relied on for filtering a list of assets. [00:21:55] Speaker 01: And that's what I'm trying to figure out here. [00:21:57] Speaker 01: Because if that is true, then you are tantamount to saying, Merusi teaches this entire claim limitation. [00:22:04] Speaker 05: It could be, but we don't have to go. [00:22:06] Speaker 01: It could be. [00:22:06] Speaker 01: It could not be. [00:22:07] Speaker 01: How do we know? [00:22:08] Speaker 01: It doesn't say what I just said. [00:22:11] Speaker 05: Let's assume it's just Merusi's database. [00:22:17] Speaker 05: What we have then is a teaching that Merusi has this information in the database and pile [00:22:24] Speaker 05: would be generating, creating a new manifest that is optimized for the particular devices. [00:22:30] Speaker 05: So you have a database that has the list of all your assets, some are in this format, some are in this language, some are in this language, and then you create a new manifest optimized based on those particular characteristics. [00:22:44] Speaker 05: What is missing from that? [00:22:45] Speaker 05: The reason the board, but what the board says, it doesn't look at those in combination. [00:22:51] Speaker 05: But you're coming back to pilot. [00:22:53] Speaker 02: I mean, it says on page 40 here, which I don't have the... 6048. [00:23:00] Speaker 01: 6048. [00:23:02] Speaker 02: It says explicitly, as discussed, Merusi filters from a database, which suggests that Merusi is doing the filtering. [00:23:11] Speaker 02: So what does that mean? [00:23:14] Speaker 05: Right. [00:23:14] Speaker 05: So this is what I'm trying to say with Jay Chen. [00:23:16] Speaker 05: I think this is fairly right. [00:23:18] Speaker 05: There are two arguments here. [00:23:19] Speaker 05: It relies on [00:23:20] Speaker 05: It can be either Marousi's filtering or it can be Pyle's filtering. [00:23:27] Speaker 01: And the board is... But it's always Marousi's database in both of those. [00:23:32] Speaker 05: In these combinations, it's always from Marousi's database, yes. [00:23:35] Speaker 01: And so there is no combination when you're relying just on Marousi's filtering. [00:23:42] Speaker 05: When you talk about, there is a combination in the sense that it's combined with pile for some of the other limitations. [00:23:49] Speaker 01: I'm just talking about this limitation, trying to understand what you were trying to tell the board for this one filtering limitation. [00:23:57] Speaker 05: Right, so it talks about Marusi filters from a database. [00:24:00] Speaker 01: Why wouldn't it be reasonable to read this portion of the petition that Judge Dyke just referred to as setting the table for why one would combine piles filtering with Marusi's database and [00:24:16] Speaker 01: the petitioners trying to show us that see, you know, Merusi is kind of in the same space, sort of trying to do the same thing. [00:24:23] Speaker 01: And so that's why it would be obvious to combine Merusi's database with piles filtering. [00:24:30] Speaker 01: That's a typical thing that we see in petitions, again, from the many IPR petitions I've read. [00:24:36] Speaker 01: You try to, you know, explain and identify why the references you're combining are located in the same technical space. [00:24:44] Speaker 01: But you're not relying on everything from each of the references individually to try to match up to the claim limitation. [00:24:50] Speaker 00: No. [00:24:51] Speaker 01: You are relying on one piece from reference A and one piece from reference B to match up with the claim limitation. [00:24:57] Speaker 05: Right. [00:24:58] Speaker 05: But the way this petition is structured, and I do think the board was thrown about this, is it has that earlier discussion. [00:25:04] Speaker 05: At 6.029 is the discussion of the motivation to combine. [00:25:09] Speaker 05: And then at this point, we're talking about the specific [00:25:11] Speaker 05: limitations. [00:25:12] Speaker 05: 6029 says implementing Marusi's technique for organizing multiple representations and choosing among them based upon a device's capabilities would predictably function in Pyle's system. [00:25:29] Speaker 05: I think one of the things we struggle with is because there are so many multiple paths here, [00:25:35] Speaker 01: It's incumbent upon the petition to make it clear. [00:25:40] Speaker 05: But the petition is allowed to argue in the alternative like this, and what the board can't do is when you have these multiple paths laid out, is say, I am doing no separate analysis for the combination of Palamruci. [00:25:55] Speaker 05: That's what the board did. [00:25:56] Speaker 02: Well, that's what the board seems to be saying. [00:26:00] Speaker 02: is even though you said Merusi filters, we're going to not credit that and think what you really meant was that Pyle filters. [00:26:08] Speaker 05: Right. [00:26:09] Speaker 05: They're not addressing it as argued. [00:26:13] Speaker 05: They're addressing it as if we were just saying it's the exact same as Pyle. [00:26:19] Speaker 05: And so there's no separate analysis of the combination. [00:26:21] Speaker 05: They just say, we ruled against you on Pyle. [00:26:23] Speaker 05: We ruled against you on this. [00:26:25] Speaker 05: So we think that this is another example where [00:26:29] Speaker 05: latched on to, for Pyle to filter in isolation, it didn't look at those surrounding teachings based on Marussi. [00:26:38] Speaker 05: I recognize that. [00:26:40] Speaker 02: Okay, thank you, Mr. Sands. [00:26:42] Speaker 02: We'll give you a couple minutes for the bottle. [00:26:44] Speaker 02: Thank you. [00:26:45] Speaker 02: Mr. Petillo? [00:26:47] Speaker 02: May I please the court? [00:26:48] Speaker 02: Could we look at this last point while we have it fresh in our minds about the pile Merusi combination? [00:26:57] Speaker 02: Certainly. [00:26:59] Speaker 02: The petition specifically says Merusi filters from a database. [00:27:04] Speaker 02: How is it that the board can say they didn't argue the Merusi filters, what they really meant, only pile filters? [00:27:11] Speaker 00: Well, I think they're... How could that be? [00:27:14] Speaker 00: Well, because there are several responses to that, Your Honor. [00:27:19] Speaker 00: The first, I think, is, as Judge Ten was alluding to, when they mentioned that Merusi filters from a database, they were trying to show why you would make the combination with Pyle. [00:27:31] Speaker 00: You would say, oh, well, Pyle would filter from a database because Merusi filters something from a database. [00:27:37] Speaker 00: Now, the issue is, well, what were they filtering? [00:27:42] Speaker 00: And the board correctly found that here, although the petition mentioned that Merusi filters from a database, [00:27:48] Speaker 00: It ultimately relies on Pile for filtering in the context of these patents. [00:27:53] Speaker 02: And excuse me. [00:27:56] Speaker 02: When somebody says explicitly that a reference does something, they can say, well, you didn't really mean that. [00:28:02] Speaker 02: So we're going to construe it as meaning something else. [00:28:07] Speaker 02: How can they do that? [00:28:08] Speaker 00: I think we have to look at specifically, what did they actually say in their petition that Merusi filters? [00:28:14] Speaker 00: And it's 6048. [00:28:15] Speaker 00: What they actually say Merusi filters from, they mention filtering a database of product identifiers and device capabilities. [00:28:24] Speaker 00: And that's an entirely different thing than what's going on here. [00:28:27] Speaker 00: That's essentially saying... That's a different argument than the board's. [00:28:32] Speaker 00: Well, no, Your Honor. [00:28:34] Speaker 00: I think that they recognized that they were trying to link up the filter. [00:28:42] Speaker 00: Why would you apply Marussi's database teachings with Pyle? [00:28:45] Speaker 00: Because it actually does a sort of filtering. [00:28:48] Speaker 00: The board said, in this context, we're looking to what did they tell us themselves? [00:28:54] Speaker 00: How did they sum up their arguments? [00:28:56] Speaker 00: At the very end, Netflix explicitly says, we are relying on, quote, Marussi's database [00:29:02] Speaker 00: teachings of identifiers and associated assets and quote, relying on pile to filter the assets and it cannot be an abusive discretion for the boards who have accepted Netflix's own characterization of its own argument at the end of the day. [00:29:18] Speaker 00: They noted that they were relying on Marusi only for the database teachings, and they were relying on Pyle for the filtering, and to the extent they earlier talked about Marusi doing a different sort of filtering, a different sort of function, that was just to show, well, why would you use, why would you combine these two? [00:29:34] Speaker 00: Well, if you're trying to filter things, you might use a database, but that's entirely different, because Marusi... That's not what the board said. [00:29:41] Speaker 02: The board said, we look at this, and we think what they really meant [00:29:46] Speaker 02: not what they said explicitly. [00:29:48] Speaker 02: They said explicitly Marussi filters. [00:29:50] Speaker 02: Why doesn't the board have to address that argument instead of saying, oh, well, they didn't really mean it? [00:29:56] Speaker 00: Well, I think that, respectfully, I think that they did. [00:29:58] Speaker 00: They were talking about they ultimately rely on pile for filtering. [00:30:02] Speaker 00: Filtering in the context of what? [00:30:04] Speaker 00: And here, the filtering limitation is filtering a list of assets where each asset is a different stream. [00:30:11] Speaker 00: And nowhere do they try and show where Marussi involves [00:30:15] Speaker 00: filtering a list of assets. [00:30:17] Speaker 00: Not once did they discuss Marusi filtering a list of assets to ultimately sort them out and produce the manifest files at the end of the day. [00:30:26] Speaker 00: Marusi doesn't even involve manifest files at all. [00:30:31] Speaker 00: And so it was entirely fair for the board to say that even though it mentions filtering, that's not the sort of filtering we're talking about with respect to what the limitations require. [00:30:40] Speaker 00: And so it was fair for the board to look back and say, [00:30:44] Speaker 00: I think that's absolutely fair reading of the boards. [00:30:47] Speaker 02: They just say, oh well, we think they didn't really mean it. [00:30:50] Speaker 02: They were rolling on file. [00:30:52] Speaker 02: They don't say, this is referring to a different kind of filtering. [00:30:55] Speaker 02: They don't make that argument. [00:30:57] Speaker 00: Your Honor, I respectfully think that that is a fair reading of the board's decision, especially when you look at what Netflix was actually arguing. [00:31:05] Speaker 00: Where did Netflix actually argue that it filters a list of assets? [00:31:13] Speaker 00: at all, and they don't point to that, there's not a single statement in the petition. [00:31:19] Speaker 00: To the extent that they talk about, you know, now counsel points to identifying a subset of multimedia content files, they didn't actually make that argument expressly in the petition. [00:31:30] Speaker 00: The only statement in the petition, now, you know, counsel points to a big block quote and they try and, you know, pluck something out of it. [00:31:37] Speaker 00: But the fact is the board's rules require you to say, if you're going to cite evidence, [00:31:42] Speaker 00: you have to explain the relevance of that evidence. [00:31:45] Speaker 00: And to the extent they want to point to identifying a subset of multimedia content files now, they never made that argument below. [00:31:51] Speaker 00: All they said was, yeah, Mersey involves filtering database of product identifiers and device capabilities. [00:32:02] Speaker 00: And so it wasn't even discussing filtering the list of assets that are at issue here. [00:32:07] Speaker 00: So I think it's entirely fair to say, [00:32:09] Speaker 00: as the board found, even though you mentioned it's filtering, that's not the type of filtering we're talking about here. [00:32:15] Speaker 00: You said you were ultimately relying on Pile to filter and Merusi for its database teachings, and I think it's absolutely fair to hold Netflix to how they characterize their own argument. [00:32:27] Speaker 00: And in fact, that's what all of these issues boil down to, is was it fair for the board to hold Netflix to what's in its petition? [00:32:34] Speaker 00: The rules required Netflix to make its arguments with particularity. [00:32:38] Speaker 00: And the narrow question before this court is, was it an abuse of discretion for the board not to have discerned all of the supposed alternative arguments that counsel was discussing here from the very skeletal and very conclusory statements that we have in the petition? [00:32:59] Speaker 00: And I don't think this is an abuse of discretion. [00:33:02] Speaker 00: It's not even the same ballpark as abuse of discretion. [00:33:07] Speaker 00: To go back to Pile itself, that the board overlooked the petition's argument that Pile's new manifest disclosed the filtering list of assets limitation. [00:33:18] Speaker 00: The board didn't overlook that argument. [00:33:20] Speaker 00: It noted, it specifically noted that Netflix had pointed to Pile's discussion of the new manifest in relationship to the filtering limitation. [00:33:28] Speaker 00: But the problem was Netflix's argument was a single sentence and it didn't discuss filtering. [00:33:32] Speaker 00: It said Pile teaches the use of a new manifest [00:33:35] Speaker 00: that can be created based upon particular devices or capabilities. [00:33:39] Speaker 00: And again, that was followed by a block quote to Piles New Manifest 422. [00:33:44] Speaker 00: But it didn't mention filtering. [00:33:46] Speaker 00: It doesn't mention a list of assets. [00:33:48] Speaker 00: And the board rules require you, you can't just throw out... You don't have to use magic words, but you have to provide an explanation. [00:33:56] Speaker 00: And you can't just say, well, this teacher's creating, and then have a block quote, and then say, well, that shows filtering a list of assets. [00:34:03] Speaker 00: when the block quote is not apparent how that involves filtering a list of assets, and in fact the block quote doesn't involve filtering a list of assets, because the only disclosure in Pile of the new Manifest 4.22 is talking about content providers create this, and that means that the studios who are producing the multimedia content [00:34:25] Speaker 00: And what does Pyle say about this? [00:34:27] Speaker 00: Well, if they look at what streams are important, and we notice more people are using the iPhone 12, and we notice more people are looking at this in French, we need to do a version of the new X-Men movie, and this will be the most popular manifest, so let's create a new manifest. [00:34:42] Speaker 00: None of that has anything to do [00:34:44] Speaker 00: with generating a new manifest file dynamically the way DivX's patent claims to. [00:34:51] Speaker 00: It's an entirely separate process that's entirely consistent with how Pyle works, because Pyle doesn't work remotely like DivX's patents do. [00:35:01] Speaker 00: The whole theory behind Pyle is saying, you know what, it would be a great idea if we could identify the universe of [00:35:07] Speaker 00: manifest files that devices are going to need, the most likely ones, we're going to create those in advance, and we're going to have them stored, and we're going to have them ready to go so that when someone needs the French version of Batman Returns for an HDTV, we're going to pluck that off the shelf and provide it. [00:35:24] Speaker 00: And that is why they cannot show that Pile discloses filtering a list of assets to create a new manifest in the same way as our system, because it's an entirely, entirely different process. [00:35:37] Speaker 00: And it was entirely fair for the board to say, you know, you pointed to creation of the new manifest, but you didn't explain how Pyle's new optimized manifest teaches filtering. [00:35:50] Speaker 00: And I don't think we can point to where they mentioned the new manifest 422 in the context of the filtering limitation, but they never provided it [00:36:00] Speaker 00: The specificity that they needed to board the rules require you to explain the relevance of the evidence, and they never explained how that requires filtering, and they can't. [00:36:10] Speaker 00: And part of the reason, too, is it ends up especially clear when you look at what their theories of the list of assets actually are. [00:36:16] Speaker 00: I heard counsel say, and I believe they also said this in their reply brief, that the assets were sort of the underlying things themselves in the manifests. [00:36:25] Speaker 00: And why did they say that? [00:36:25] Speaker 00: Because there's clearly no way that Pile discloses going back and grabbing manifests to create the new manifest. [00:36:33] Speaker 00: But the fact is that that's incorrect. [00:36:35] Speaker 00: The notion that somehow now the underlying assets themselves are the list of assets is just [00:36:40] Speaker 00: You know, the board made a finding that Netflix relied on the manifests as lists of assets, whether taken individually, taken collectively. [00:36:50] Speaker 00: And Netflix's opening brief confirmed that was correct. [00:36:53] Speaker 00: They said, we have two theories. [00:36:55] Speaker 00: There are two ways in which piles manifests are the list of assets. [00:36:58] Speaker 00: It's each manifest individually. [00:37:00] Speaker 00: It's each manifest collectively. [00:37:01] Speaker 00: That's at 35 of their brief. [00:37:03] Speaker 00: They confirmed, so instead of disputing that, they confirmed it's correct. [00:37:07] Speaker 00: And so there's just no way that pile disclosed. [00:37:10] Speaker 03: Excuse me. [00:37:12] Speaker 03: You mentioned that pile generates these new manifests in advance, and they're available as needed. [00:37:21] Speaker 03: And you make the distinction in your brief that the claims require generating files in top level index files on the fly. [00:37:32] Speaker 03: Where in the claim does it talk about generating top index files on the fly? [00:37:40] Speaker 00: Well, it's throughout the entire method. [00:37:44] Speaker 00: It's the method of generating a top-level index file. [00:37:48] Speaker 00: The playback server system receives the request. [00:37:51] Speaker 00: It's the playback server system performing all of these steps in a row. [00:37:55] Speaker 00: And so it receives that request in response to that request, just walking through the limitations. [00:38:01] Speaker 00: The next thing that you do in the method is you retrieve a list of assets, a broad set of assets. [00:38:06] Speaker 00: They're different streams. [00:38:07] Speaker 00: All of this happens [00:38:09] Speaker 03: in response to a request. [00:38:11] Speaker 03: You're characterizing that as sort of generating on the fly. [00:38:16] Speaker 00: Yes, Your Honor. [00:38:17] Speaker 03: I mean, that's... And distinguishing that from Pile. [00:38:20] Speaker 03: We're distinguishing that from Pile, exactly, because... Well, and if you just walk through it... Is the new manifest in Pile not generated on the fly as you're using that expression? [00:38:33] Speaker 00: It is not. [00:38:33] Speaker 00: The only disclosure... Well, the only disclosure anywhere in Pile [00:38:38] Speaker 00: of how the new manifest is created, and this is at appendix 1528, column 1060, is the new manifest is created by the content providers themselves? [00:38:49] Speaker 00: The new manifest is created by the studios who are finding what are the most popular? [00:38:53] Speaker 02: I thought that was an issue that the board didn't reach. [00:38:56] Speaker 02: I heard Mr. Saunders say that the question of whether it had to be generated [00:39:02] Speaker 02: by the content provider was not resolved. [00:39:09] Speaker 02: Is that correct or not? [00:39:11] Speaker 02: Perhaps I'm misunderstanding your question. [00:39:15] Speaker 02: Mr. Saunders said that the board did not resolve the question of whether the content provider had to communicate the list of assets. [00:39:29] Speaker 02: and that the board specifically did not resolve that question. [00:39:34] Speaker 02: And now you're attempting to distinguish pile on the ground that there was not this involvement by the content provider. [00:39:42] Speaker 02: Is it correct that the board didn't resolve that question? [00:39:45] Speaker 00: It is correct that the board found that the example I have been discussing is not limiting. [00:39:50] Speaker 00: That is absolutely correct. [00:39:52] Speaker 00: What the board did find, though, is that the only express disclosure... And how is it that the [00:39:58] Speaker 02: That was the whole basis, as I understood it, for your suggestion that it had to be done on the fly. [00:40:06] Speaker 00: That's not our suggestion of why it has to be done on the fly. [00:40:09] Speaker 00: Our claims are the reason why it has to be done on the fly. [00:40:11] Speaker 02: Our argument is showing why... When you responded to Judge Lynn, you said the claims require it to be done on the fly because of the involvement of the content provider. [00:40:22] Speaker 02: And now you're saying there's some other reason? [00:40:25] Speaker 00: I'm sorry, perhaps I wasn't clear. [00:40:26] Speaker 00: No, our claims require it to be done on the fly. [00:40:31] Speaker 00: What distinguishes our claims from Pile is that Pile is not performing. [00:40:36] Speaker 02: Where does the claim require it be done on the fly? [00:40:39] Speaker 00: I think it's absolutely implicit in the method itself. [00:40:44] Speaker 00: It's the playback server itself that's receiving a request from the playback device. [00:40:49] Speaker 00: The next step [00:40:50] Speaker 00: In response to that, it retrieves the list of assets that are different streams to see which ones do we need, and that's when it filters the list of assets. [00:40:59] Speaker 00: It gets rid of the streams that you don't need if you're in the United States. [00:41:03] Speaker 00: We need an English language version. [00:41:06] Speaker 01: I don't quite follow why you're putting so much time into this on-the-fly business about your generating limitation when, at the moment, I don't see how it connects at all to the question at hand, which is, are any or all of the arguments on appeal by Netflix forfeited by not being fairly presented in the petition? [00:41:27] Speaker 00: Judge Chen, we absolutely agree that the court doesn't need to go there. [00:41:33] Speaker 00: The fact is all of the board's findings that these issues were waived or not raised are entirely fair based on the actual arguments in the petition. [00:41:43] Speaker 00: They failed to show an abuse of discretion. [00:41:45] Speaker 00: We're just discussing the other elements to show why these sort of alternative theories are actually not really there and why they couldn't actually be made based on the references themselves. [00:41:55] Speaker 00: I agree the court does not have to go there. [00:41:58] Speaker 00: It could just look at whether the board fairly characterized the petitions and whether that was an abuse of discretion. [00:42:03] Speaker 00: And the answer is no, it was not. [00:42:05] Speaker 00: And we respectfully ask that the court affirm. [00:42:09] Speaker 00: Thank you. [00:42:11] Speaker 02: Thank you. [00:42:11] Speaker 02: Mr. Saunders, you have a couple of minutes. [00:42:14] Speaker 05: Thank you, Your Honor. [00:42:15] Speaker 05: So as was just addressed, most of what we heard are not the issues that the board reached. [00:42:20] Speaker 05: These are the issues that we want to get back to. [00:42:22] Speaker 05: We want to fight these out in front of the board so it can consider on the merits. [00:42:26] Speaker 05: Instead, what we had was a threshold procedural ruling. [00:42:30] Speaker 05: And on the combination with Marussi, [00:42:33] Speaker 05: We have the statement about filtering from a database used to select the device identifiers and it's used to select the appropriate stream to transmit. [00:42:43] Speaker 05: Next page, entire paragraph there from Marussi about its matching unit adapted to identify a subset of the multimedia content files. [00:42:51] Speaker 01: Why should any of that be understood as a database that somehow contains material for filtering a list of assets? [00:43:02] Speaker 01: when those words don't quite come out in these pages. [00:43:06] Speaker 05: Well, there's not a requirement that the prior art disclosed it in the same words. [00:43:12] Speaker 01: I understand that, but there is a requirement to explain what you're actually pointing to that actually matches the words of the claim. [00:43:21] Speaker 01: The words of the claim are filtering a list of assets where each asset is a different stream of the video content. [00:43:27] Speaker 01: And so the next question is, [00:43:30] Speaker 01: You do use the word filtering at one point, but I don't understand why there's a basis for you to say that filtering from a database that associates terminal identification information with terminal capabilities information is somehow should be understood as the same thing as filtering a list of assets, where each asset is a different stream of content. [00:43:54] Speaker 05: Because the end of that sentence is explaining you're doing that, and then, quote, that is used to select the appropriate stream to transmit. [00:44:03] Speaker 05: And then the block quotes are describing the filtering from the capabilities. [00:44:09] Speaker 05: Then it's describing Marusi's matching unit. [00:44:12] Speaker 05: and then it's summing up at the beginning saying Marussi also teaches identifying a subset of multimedia content files. [00:44:19] Speaker 05: So with all of those teachings laid out, and this is when we talk about the level of detail here, I mean these block quotes are trying to be as transparent as possible, show you exactly what's being relied on as you would see in a claim chart. [00:44:33] Speaker 05: Putting everybody on notice, these are the very passages we're relying on, and the whole point of this argument is to argue in the alternative. [00:44:41] Speaker 02: If you're not with us on Pile, Maroussi fills this in.