[00:00:00] Speaker 01: Our first case for argument today is 22-1529, Juniper Networks vs. Huawei Technologies. [00:00:07] Speaker 01: Mr. Kagan, please proceed. [00:00:09] Speaker 00: Thank you, and may it please the court, Jonathan Kagan of Irelia Manila, representing Appellant and Cross-Appellee Juniper Networks. [00:00:16] Speaker 00: So the Patent, Trial, and Appeals Board erred in refusing to consider certain reply evidence and arguments that Juniper put forth in the case below. [00:00:28] Speaker 00: And I think there are a lot of issues in the case, but I think the easiest way to demonstrate that is we should assume, and we can assume for purpose of this argument, this oral argument, that the positions that Huawei took in their patent owner response were correct regarding the limitations on the prior art. [00:00:48] Speaker 00: What Huawei argued in the Patent Owner Response is that the primary reference, Maschak, was limited to HTML root pages. [00:00:59] Speaker 00: Now, we disagree with that assumption, but let's accept it for purpose of argument today. [00:01:05] Speaker 00: Secondly, Huawei argued that the Gribble reference was limited to looking at HTTP headers only. [00:01:17] Speaker 00: That's what they argued in their Patent Owner Response. [00:01:19] Speaker 00: We disagree. [00:01:20] Speaker 00: We cited broader language in the petition, and we cited that in the briefs. [00:01:25] Speaker 00: But for purposes of this argument, again, we can assume that Gribble is limited to HTTP headers only. [00:01:32] Speaker 00: Because then the question is, what is the argument that Huawei made, given those assumptions? [00:01:37] Speaker 00: And the argument they made, and this was in their Patent Owner Response, and this is on Appendix 451, they say, as would have been known to a Pasita, [00:01:47] Speaker 00: there is no indication in the HTTP response header as to whether the requested HTML page includes active embedded objects or whether the HTML page includes embedded objects at all. [00:02:03] Speaker 00: So Huawei's argument that they put forward in their response is there is no indication in this HTTP response header about whether there's objects [00:02:16] Speaker 00: on the page. [00:02:18] Speaker 03: Is this argument you're making right now really you're focusing, I believe, on whether the board erred in saying that your response argument was entirely new, right? [00:02:27] Speaker 00: Yes, that is correct. [00:02:29] Speaker 03: But even if I agree with you on that, don't you also have an additional hurdle of showing that there isn't substantial evidence to support the board's alternative ruling? [00:02:38] Speaker 03: I mean, it said, even assuming that you hadn't presented new evidence, [00:02:43] Speaker 03: uh... substantial evidence uh... if you had it presented you are innocent argumenters it wasn't you that's a there is evidence that which uh... that they found to be credible expert explaining why that still didn't satisfy the claimant how do you respond to that well so that's actually not a factual determination it's a it's a claim construction argument so what the board did at the end was it said the reason the evidence is insufficient [00:03:10] Speaker 00: is because the content script type, which is what we pointed to, that tag, doesn't necessarily indicate the presence of an executable object. [00:03:24] Speaker 00: So that's a claim construction issue, because we're talking about the judging limitation. [00:03:29] Speaker 00: Does the judging limitation from the 693 pattern require that it necessarily determine [00:03:36] Speaker 00: that there is an executable object. [00:03:40] Speaker 00: And so first of all, it's a de novo issue. [00:03:44] Speaker 00: It's not a substantial evidence issue. [00:03:46] Speaker 00: But secondly, and I was going to get to this, but I'll get to it right now. [00:03:51] Speaker 00: It's actually contrary to what both experts have said in the case. [00:03:55] Speaker 00: There's no requirement in the patent [00:03:59] Speaker 00: that the judging be correct. [00:04:01] Speaker 00: And I'm somewhat embarrassed to tell this court. [00:04:03] Speaker 00: I mean, you review judgments all the time from other judges. [00:04:07] Speaker 00: And sometimes they get it right, and sometimes you reverse them. [00:04:10] Speaker 00: It doesn't mean they're not judging. [00:04:12] Speaker 00: If they get it wrong, it's OK to get things wrong sometimes. [00:04:15] Speaker 03: So you're saying that detecting by the network device whether the file is an executable file doesn't have to actually detect whether it's an executable file? [00:04:24] Speaker 03: It doesn't matter whether it works or not? [00:04:25] Speaker 00: Well, the claim element for judging whether it's an executable device, I assume you're looking at element one. [00:04:36] Speaker 00: Is it judging by the network device whether the file is an executable file according to at least one of the requests in the data stream carrying the file from claim one? [00:04:46] Speaker 03: Is it because of the word judging that it doesn't matter whether it's correct? [00:04:52] Speaker 00: The real reason it doesn't matter is what we're talking about is malware. [00:04:56] Speaker 00: We're talking about something where the hackers and the people that are doing this are not kind enough to let us know, hey, this is a piece of code that I'm trying to infect your computer. [00:05:06] Speaker 00: And so mistakes get made by this all the time. [00:05:08] Speaker 00: And everybody, every expert in the field recognizes this. [00:05:11] Speaker 03: And in fact, it's not judging whether there's malware. [00:05:14] Speaker 03: It's judging whether there is an executable file. [00:05:17] Speaker 00: That's correct. [00:05:18] Speaker 00: I'm just giving the context for this. [00:05:21] Speaker 00: The way this patent works, it assumes if there's an executable file that it might be malware. [00:05:26] Speaker 00: So then what it does is it takes that, and it doesn't allow it to pass through the user computer. [00:05:31] Speaker 00: It instead puts it in a virtual machine, sort of safe environment. [00:05:36] Speaker 02: I want to explore some of the arguments about whether or not the information was new. [00:05:40] Speaker 02: So I think part of the argument you were making was that you had a broader disclosure on Gribble. [00:05:45] Speaker 02: Is that right? [00:05:46] Speaker 00: In Gribble. [00:05:47] Speaker 00: In Gribble. [00:05:47] Speaker 00: Yes, for sure. [00:05:48] Speaker 02: In terms of just giving the citation to those relevant ranges, I'll say of pages or whatever, but not necessarily in line numbers, but not necessarily actually talking about the additional part in that broader disclosure. [00:06:02] Speaker 02: Is that a fair statement in terms of kind of... Yes. [00:06:05] Speaker 00: So what I would say we did is we talk about Gribble's broader disclosure, but when we analyze it, we just analyze one specific example. [00:06:12] Speaker 00: And so we took the example from Gribble, where Gribble is looking at web objects. [00:06:17] Speaker 00: So it's looking at individual objects, not group pages. [00:06:21] Speaker 00: But Gribble, by its terms, if you look at Gribble, it expressly says, look, this is just an example. [00:06:26] Speaker 00: This is not limiting. [00:06:27] Speaker 00: I mean, that language is all in Gribble itself. [00:06:30] Speaker 00: And so in this way, this case is actually very similar to the Apple Viandria case. [00:06:36] Speaker 00: In the Apple Viandria case, the petitioner [00:06:41] Speaker 00: filed a petition and said, there is an algorithm that was being used there. [00:06:46] Speaker 00: And they provided an example. [00:06:48] Speaker 00: They said, look, it had to do with the number of subwindows. [00:06:51] Speaker 00: Said, if there's one subwindow, you combine this with Hirsch, and it has every element of the patent. [00:07:00] Speaker 02: Are you also familiar with our Ariosa and Dell cases? [00:07:03] Speaker 02: Yes. [00:07:04] Speaker 02: How do you distinguish those cases? [00:07:06] Speaker 00: Well, so in Ariosa, [00:07:10] Speaker 00: What actually happened in Ariosa was the court, there was a bit of confusion, I think, about what the board had done in terms of excluding certain evidence. [00:07:21] Speaker 00: But the key to those cases is that we are not changing our theory of patentability. [00:07:28] Speaker 00: So what you have in Ariosa, well, there was [00:07:39] Speaker 00: And Dell, I believe, is they were changing the theory of patentability. [00:07:42] Speaker 00: They were saying, this is different than what we had described earlier. [00:07:46] Speaker 00: So in Dell, for example, it was at oral argument that this new theory was introduced. [00:07:55] Speaker 00: So counsel at oral argument just said, here's a new theory of invalidity. [00:07:59] Speaker 01: I don't think that it matters whether it was in reply or at oral argument. [00:08:02] Speaker 01: I mean, I didn't think that was a significant factor in the Dell case. [00:08:06] Speaker 01: I mean, I think either way, [00:08:08] Speaker 01: You could have a waived argument. [00:08:11] Speaker 00: Agreed. [00:08:12] Speaker 00: Agreed. [00:08:12] Speaker 01: So why isn't it waived when it's raised in the reply? [00:08:16] Speaker 00: Because this is not a new theory of indolence. [00:08:18] Speaker 01: What do you mean by theory of invalidity? [00:08:21] Speaker 01: This is what I'm trying to understand. [00:08:23] Speaker 00: So what we do is we say, if you take the MoshChuck system, which includes getting requests, sending requests to the web, receiving pages back, and if it contains an executable, sending it to a virtual machine for examination. [00:08:41] Speaker 00: Taking that, combining it with Gribble's way of determining whether it has an executable by looking at the file header. [00:08:48] Speaker 00: That's the invention, or that's the claimed combination here. [00:08:53] Speaker 00: That's what we claimed in the petition. [00:08:55] Speaker 00: And as I said, for purpose of oral argument, we'll say, we can argue, let's limit it to the HTTP header. [00:09:02] Speaker 00: So what's happened here, just as in Apple, is Huawei came up with a different example, and they said, no, no, you need to look at the HTTP header. [00:09:12] Speaker 00: So then in reply, we have the right to say, OK, well, even if you look at the HTTP header and the root page, there is a way that- I thought what they said was what you pointed to, what you think was content type in the HTTP header, didn't actually identify where there's an executable file. [00:09:31] Speaker 03: So I must not understand that. [00:09:33] Speaker 00: Well, we never mentioned content type in the petition. [00:09:36] Speaker 00: In the petition. [00:09:37] Speaker 00: We never mentioned content type. [00:09:38] Speaker 00: We just said that HTTP type. [00:09:39] Speaker 00: Right. [00:09:39] Speaker 00: Now, the example from Gribble. [00:09:42] Speaker 00: was a content type header, but that works with web objects. [00:09:45] Speaker 00: It doesn't work with web page. [00:09:46] Speaker 03: It works with root pages. [00:09:47] Speaker 03: And so then on reply, you expand it and said, no, no, look at the whole header or look at something else. [00:09:53] Speaker 00: Right? [00:09:53] Speaker 00: Well, all we did was say, look at the header. [00:09:55] Speaker 00: We say, if you're a person of some, because as I said, Ripple was using a particular example. [00:10:00] Speaker 00: And they're clear in the patent. [00:10:01] Speaker 00: They're saying, this is just an example. [00:10:03] Speaker 00: So if you're looking at it as a header, what their statement is, there's nothing that you can look at in the HTTP header that's going to tell you if it's an executable file. [00:10:11] Speaker 00: That was a statement of Patent on Response, page 451. [00:10:16] Speaker 00: They say there's no way of, there's no indication in the HTTP header in the context of looking at HTML root page, and we said that's not true. [00:10:26] Speaker 00: We said here's the specification, and what it tells you, the HTML specification itself says if you're going to have an executable, which is a script, in a [00:10:38] Speaker 00: in a page, on a web page, you need to specify, you need to say up front in the header what language you're using, because otherwise it will not operate properly. [00:10:49] Speaker 00: So therefore, one is still in the art. [00:10:50] Speaker 03: What line is that? [00:10:52] Speaker 00: I'm sorry, what? [00:10:52] Speaker 03: What line is that in the header? [00:10:55] Speaker 00: That's the content script type line. [00:10:57] Speaker 00: Thank you. [00:10:59] Speaker 00: So, and again, we submitted the evidence from the HTML specification of that. [00:11:03] Speaker 00: We submitted a declaration from our expert indicating that. [00:11:07] Speaker 00: In the HTTP header, exactly as they said, what Huawei then did was in their sir reply. [00:11:14] Speaker 00: That's when they changed and said, oh, now it's looking just at the content type header. [00:11:20] Speaker 00: And the content type header itself won't tell you that. [00:11:24] Speaker 00: But we have never limited. [00:11:25] Speaker 00: I mean, there was not even an argument in the patent on a response that this is somehow limited to the content type header. [00:11:30] Speaker 00: And that would be contrary to what this court ruled in Apple v. Andrea, which is you can use a different example that's within the scope of the patent to demonstrate that's perfectly acceptable. [00:11:43] Speaker 01: During your rebuttal time, do you want to save time? [00:11:45] Speaker 00: I will reserve the rest of the time. [00:11:49] Speaker 01: Thank you. [00:11:50] Speaker 04: Thank you, Chief Judge Moore. [00:11:51] Speaker 04: Robert Courtney, representing Huawei. [00:11:54] Speaker 04: I'll begin with the primary appeal, turn to the cross appeal at the end. [00:11:58] Speaker 04: I want to respond to some of the conversation we just had. [00:12:00] Speaker 04: Chief Judge Moore, you pressed my colleague Mr. Kagan on the theory of invalidity that Juniper was presenting. [00:12:07] Speaker 04: And I think it's telling that Juniper immediately said, well, let's talk about Huawei's submissions to identify the theory of invalidity. [00:12:13] Speaker 04: But that is not the test. [00:12:16] Speaker 04: The rules empower the board to look at the petition to see the theory of invalidity. [00:12:21] Speaker 04: That's at Appendix 187. [00:12:23] Speaker 04: And when we look at Appendix 187, the theory of invalidity [00:12:27] Speaker 04: But if they lay out, we begin with a design principally established in the Moshuk reference. [00:12:33] Speaker 04: Going to make a single important modification to it based on Gribble. [00:12:38] Speaker 04: And appendix 187 says we're going to bring in the part of Gribble where it looks at application octet screen header. [00:12:47] Speaker 04: Now that is the content type header. [00:12:49] Speaker 04: Gribble expressly says that's the content type header. [00:12:52] Speaker 04: So the theory of invalidity they presented at the time of the petition makes no reference of any outside header, indeed. [00:13:00] Speaker 03: What about the fact that when they said application slash architect string, they said EG? [00:13:05] Speaker 03: Absolutely. [00:13:05] Speaker 03: I mean, they said specifically to see if the HTTP header is associated within executable EG. [00:13:12] Speaker 03: So isn't that just an example? [00:13:14] Speaker 04: In Gribble, it is an example. [00:13:16] Speaker 04: But I think this case illustrates that the board looks to the petition. [00:13:19] Speaker 03: Well, this is the language of the petition. [00:13:21] Speaker 03: So how am I? [00:13:22] Speaker 03: I'm just asking, like, is it reasonable to say they say the HTTP header, and then they say EG application slash op-deck string. [00:13:31] Speaker 03: So maybe something else in the header could be included in what they're relying on? [00:13:35] Speaker 04: I think if Juniper wanted to rely on something beyond the example given, the petition would have been an excellent time for them to articulate that. [00:13:42] Speaker 03: How many different things are in the HTTP header? [00:13:47] Speaker 04: There are a bunch. [00:13:48] Speaker 04: There's at least six elements. [00:13:50] Speaker 04: We lay out some of them. [00:13:52] Speaker 04: That's less than helpful. [00:13:54] Speaker 04: Our briefing actually, I think, quotes the line, content script type is not in all HTTP headers. [00:14:00] Speaker 04: It's an optional field. [00:14:02] Speaker 04: But I think the board's decision here was that the burden is squarely on the petitioner, on Juniper. [00:14:07] Speaker 04: If it wants to fill in that EG, it should do that at the petition stage. [00:14:11] Speaker 04: Because what we have instead is a patent owner that looks at the petition, says, oh, they're relying on content type. [00:14:20] Speaker 04: We went and got an expert, consulted with the expert, submitted the expert's opinions. [00:14:24] Speaker 04: The expert said, content type won't do it. [00:14:26] Speaker 04: Here's all the problems with it. [00:14:28] Speaker 02: What about the fact that they cite it to the broader disclosure? [00:14:31] Speaker 02: I think I was talking a little bit with opposing counsel about that. [00:14:34] Speaker 02: Do you feel like then that allowed them to go into the discussion later on the well-known signature, et cetera? [00:14:40] Speaker 04: I don't when you look at the context of that. [00:14:42] Speaker 04: And I think the board, we are under abusive discretion review here. [00:14:46] Speaker 04: It was very reasonable for the board to look at that and say, yes, you have a string site to 12 lines of Gribble. [00:14:53] Speaker 04: But you also explicate that string site at page 187 of the appendix. [00:14:57] Speaker 04: And you say, we're specifically, using that word, specifically modifying our primary reference according to this application octet string. [00:15:05] Speaker 04: And the board said, once you've said that and your adversary has [00:15:09] Speaker 04: built their evidence based on that to then make a pivot and say, well, actually, there was something else hidden in that EG that we didn't mention in the petition. [00:15:19] Speaker 04: You're making that pivot after your adversary, in this case, Huawei, no longer has an opportunity to submit new evidence. [00:15:25] Speaker 04: The board's rules preclude Huawei from presenting new opinions from their own expert on this issue, preclude Huawei from presenting new documents addressing this. [00:15:36] Speaker 04: So this is absolutely an issue that Juniper laid behind the log, and then after the response came out said, ah, you didn't think of content script type, you didn't think of the head element of an HTML header, and now Huawei under the board's rules has no ability to present new evidence. [00:15:55] Speaker 03: Do they rely on both the head element and the content script type? [00:16:00] Speaker 04: Do they have alternatives? [00:16:04] Speaker 04: That may be a better question for Mr. Kagan. [00:16:06] Speaker 04: I understood then representing as kind of an either or. [00:16:10] Speaker 04: Certainly Mr. Kagan's argument only relied on the HTTP header, which would be content script type. [00:16:16] Speaker 04: Also, I think Apple Andrea is very instructive. [00:16:20] Speaker 04: I think my colleague has misread it. [00:16:23] Speaker 04: In Apple Andrea, Apple's contention was that the number of subwindows did not matter for purposes of their assessment of these algorithms. [00:16:31] Speaker 04: So they arbitrarily said, well, if it's one, here's a discussion of how things work if the number of subwindows is one. [00:16:37] Speaker 04: Their adversary said, no, we think it does matter. [00:16:40] Speaker 04: We think there's a different outcome if the number of subwindows is two. [00:16:43] Speaker 04: And Apple came back and said, no, it doesn't matter. [00:16:46] Speaker 04: Here's why you're misreading the reference. [00:16:49] Speaker 04: And even if you were writing it and the number of subwindows was material, it wouldn't change the outcome. [00:16:54] Speaker 04: So that is a petitioner defending its position. [00:16:57] Speaker 04: defending the language in the petition against an attack from its adversary. [00:17:01] Speaker 04: Here we have Juniper abandoning the language in his petition. [00:17:04] Speaker 04: They are not here seeking review of the board's decision that the arguments in the petition [00:17:10] Speaker 04: were insufficient to establish unpatentability. [00:17:13] Speaker 03: Do you want to address the board's alternative ground, which is that even assuming it wasn't a new argument, we still find that the fire doesn't teach the claimant element. [00:17:24] Speaker 03: Could you address that? [00:17:24] Speaker 04: Certainly. [00:17:25] Speaker 04: So as to the content script type header, Huawei was unable to present new evidence, but was able to rely on the documents that Juniper itself had put into the record. [00:17:36] Speaker 04: and pointed out that even in those documents, the content script type header only declares a default language for the script. [00:17:43] Speaker 04: If there is a script, you could declare that element and maybe there's no script at all. [00:17:47] Speaker 04: There is no record on how well, if at all, content script type relates to executability because Huawei had no opportunity to build one. [00:17:55] Speaker 03: What do you think of the board's use of the word necessarily? [00:17:58] Speaker 03: Is there something to be implied from that? [00:18:00] Speaker 04: What I take from that is the board is saying, [00:18:03] Speaker 04: Juniper failed in its proof burden. [00:18:05] Speaker 04: Juniper's obligation is to show that this header performs the judging according to limitation. [00:18:12] Speaker 04: They've pointed to this header, but it doesn't necessarily have anything to do with executability. [00:18:17] Speaker 04: So Juniper has failed in its burden. [00:18:18] Speaker 04: It's not a claim construction issue. [00:18:21] Speaker 04: It is absolutely substantial evidence review in our assessment. [00:18:25] Speaker 03: And what do you think about the word judging? [00:18:26] Speaker 03: Does judging make it so it doesn't have to necessarily be correct all the time? [00:18:31] Speaker 03: Or actually, sometimes it just, I don't know, it's just curious about your thought on that, since it is a claim construction issue now being raised. [00:18:39] Speaker 04: I think this claim requires the act of making a determination as to executability. [00:18:45] Speaker 04: This claim does not, no, I guess I think it does have to be at least useably accurate in some way that the invention is operative. [00:18:53] Speaker 04: If it's making decisions at random, right, that's not judging. [00:18:57] Speaker 04: It has to be some kind of, [00:19:00] Speaker 04: call it evidence-based determination. [00:19:01] Speaker 04: But the claim does not impose specific accuracy requirements. [00:19:06] Speaker 04: Finally, I'd like to direct the court. [00:19:08] Speaker 04: I'm sure the court is already familiar with yesterday's opinion in Medtronic. [00:19:12] Speaker 04: That's appeal 2357. [00:19:13] Speaker 04: We think Medtronic is highly instructive here. [00:19:17] Speaker 04: There as here, the petitioner presents an initial theory. [00:19:20] Speaker 04: The adversary comes back and says, your theory has holes. [00:19:25] Speaker 04: There would need to be a whole lot of additional modifications made to your theory for you to prevail. [00:19:30] Speaker 04: The petitioner comes back and says, oh, right, OK. [00:19:33] Speaker 04: Here's a whole new bunch of modifications. [00:19:34] Speaker 04: And the board declined to consider that. [00:19:37] Speaker 04: The board is empowered to enforce its own rules that put the emphasis on the petition and the burden on the petitioner. [00:19:43] Speaker 04: The petitioner is the lowest cost avoider for these kinds of problems. [00:19:47] Speaker 04: The board has discretion to identify that. [00:19:49] Speaker 04: And here we think the board's decision was clearly within that discretion. [00:19:54] Speaker 04: If I may, I'd like to turn to the cross appeal. [00:19:56] Speaker 04: So we have different claims, different set of references. [00:19:59] Speaker 04: The issue here is the content of the Dubrovsky reference. [00:20:02] Speaker 04: We are again on this judging limitation. [00:20:04] Speaker 04: And for this limitation, the petition is unambiguous. [00:20:08] Speaker 04: It relies exclusively on Dubrovsky. [00:20:11] Speaker 04: For the judging limitations, it's at appendix 214 to 15. [00:20:15] Speaker 04: There is no mention of modifying Dubrovsky. [00:20:18] Speaker 04: There is no mention of anything other than we're going to take Dubrovsky's disclosure, and that will be sufficient to practice the judging limitation. [00:20:29] Speaker 04: However, we also have it undisputed at this stage that Dubrovsky does not actually teach how it makes determinations as to executability. [00:20:39] Speaker 01: At appendix 2... But the claims don't specify that it has to determine how it makes the determination [00:20:45] Speaker 01: for executability, only that it be determined from certain information. [00:20:49] Speaker 01: But the claims themselves don't require what you're saying. [00:20:53] Speaker 04: So respectfully, I read the judging according to limitations, saying the judging according to one or both of the data stream or the URL. [00:21:03] Speaker 04: So the judging has to be according to those. [00:21:05] Speaker 01: Right, according to the data stream or request. [00:21:08] Speaker 04: Yes, that's correct. [00:21:09] Speaker 01: But that's all it says. [00:21:11] Speaker 04: That's correct, but the judging must be driven by one or both of those features. [00:21:15] Speaker 01: Well, why doesn't substantial evidence support the board's finding that Dombrowski determines whether the file is executable according to the data stream? [00:21:23] Speaker 04: Because there's unrebutted evidence in the record, actually agreed evidence in the record, that there are ways of judging executed [00:21:30] Speaker 04: Ways for the drug. [00:21:31] Speaker 01: No, not judging. [00:21:32] Speaker 01: Stop using the word judging, because that's my problem. [00:21:34] Speaker 01: When you use the word judging and the evidence that you're referring to isn't using the word judging in the way that you and I just defined it, which is all the claims require, that it be according to the data stream or according to the request. [00:21:46] Speaker 01: So focus on that. [00:21:47] Speaker 01: What is the evidence that Dombrowski doesn't disclose judging according to the data stream or the request? [00:21:56] Speaker 04: Thank you. [00:21:56] Speaker 04: So we know from Dubrovsky, and this is at appendix 1855, that it is somehow making a determination that a file is or is not executable. [00:22:07] Speaker 04: It doesn't say how. [00:22:09] Speaker 04: Juniper described this situation at appendix 227 as agnostic. [00:22:14] Speaker 04: Dubrovsky is a black box on this issue. [00:22:17] Speaker 04: Huawei presented opinions from Dr. Davis describing how there were ways to practice this, or to implement this. [00:22:25] Speaker 01: Yes, but Dr. Nielsen testified that because the only external inputs to the gateway relate to the requested file are the data packets and URL requests, the skilled artisan would understand that Dobrowski uses URL requests for the data packets to determine whether the file is executable. [00:22:40] Speaker 01: So why isn't that substantial evidence? [00:22:42] Speaker 04: Because at the same time, Dr. Nielsen conceded that there could be other ways of doing this outside of using those two elements. [00:22:50] Speaker 04: This is at appendix 9666. [00:22:52] Speaker 02: But the board can make the credibility determination here, right, in terms of what it's going to believe. [00:22:58] Speaker 04: Certainly the board can make credibility determinations, but those have to be determinations that he witnessed You know it's not telling truth right is is somehow Misdirected in their testimony. [00:23:08] Speaker 04: There's nothing like that here. [00:23:10] Speaker 01: Dr. Davis described how in his opinion there is a substantial evidence question So whether your expert put on contrary evidence is irrelevant [00:23:18] Speaker 01: Dr. Nielsen testified that a skilled artisan would understand the Dabrowski reference uses the URL request or the data package to determine when the file is executable. [00:23:27] Speaker 01: I don't know why that's not just the end of your process, Neil. [00:23:30] Speaker 04: Because he also testified that there could be other ways of doing it. [00:23:33] Speaker 01: There could be other ways, but he testified that a skilled artisan reading this reference would understand these to be the ways to do it. [00:23:41] Speaker 04: So I think when a witness says, so the way I respectfully read that is a witness makes a broad statement in his initial declaration, and then is frankly impeached on that statement during deposition, and then makes a concession. [00:23:56] Speaker 04: And I think when the board sees that initial statement, impeachment, concession, to rely only on the initial statement. [00:24:04] Speaker 01: The board didn't understand you to have impeached him. [00:24:07] Speaker 01: is your problem. [00:24:08] Speaker 01: You are claiming you impeached him, but the board did not understand his testimony as having been impeached. [00:24:12] Speaker 04: And that's why we think the board's decision lacks substantial evidence support, because we think this court reviewing that record and also reviewing that Juniper itself referred to Dabrowski as agnostic on this point. [00:24:21] Speaker 01: I guess I am also a little incredulous about your having impeached him, because doesn't figure one of Dabrowski at page 1848 of the appendix only show URL requests or data streams? [00:24:34] Speaker 04: There's no dispute that the system of Dabrowski has access to those data. [00:24:37] Speaker 01: Not has access. [00:24:38] Speaker 01: Dabrowski itself doesn't disclose anything else. [00:24:41] Speaker 01: You asked him, could there be some other magical way of doing this? [00:24:44] Speaker 01: But Dabrowski doesn't disclose some other magical way of doing it. [00:24:48] Speaker 01: So why wasn't the board, quite frankly, reasonable in relying on his representation that this reference discloses only these two ways? [00:24:56] Speaker 01: And so a skilled artist would understand these two ways to be the way it's done in this reference. [00:25:00] Speaker 01: Of course, it's possible some other way could be [00:25:03] Speaker 01: used, but that's not what's disclosed in this reference. [00:25:06] Speaker 01: Figure 1 only discloses these two ways. [00:25:10] Speaker 04: So I'd like to respond to my rebuttal time. [00:25:13] Speaker 04: I will just say I think Dubrovsky makes no disclosure as to how the executability determination is made. [00:25:20] Speaker 03: I have a question for you. [00:25:22] Speaker 03: How does your patent do that stuff? [00:25:24] Speaker 03: I read the 693 patent, and I'm having a hard time seeing what its specificity is on whether their file is an executable file. [00:25:32] Speaker 03: And I realize that you might say that's not an issue here. [00:25:35] Speaker 03: But thinking about that in terms of the level of specificity required, because we're looking at that same limitation with regard to several pieces of prior argument. [00:25:44] Speaker 03: So I'm trying to understand how yours works, and some of these do not, according to you. [00:25:50] Speaker 04: The way, again, not of issue, as Your Honor probably noted. [00:25:54] Speaker 04: It works exactly as the claim does, which is that the invention is able to look at not, it doesn't require a complete file. [00:26:01] Speaker 04: It can look at the stream as it comes in and look for indicators of executability, or it can look at the URL itself, the link to the data, the address. [00:26:10] Speaker 03: How? [00:26:11] Speaker 03: What specifically is it looking at? [00:26:13] Speaker 04: I mean, I'm going maybe a little beyond, I don't have citations prepared, but my understanding is it can intercept the data stream as it can intercept the request to begin with, so it sees the address. [00:26:23] Speaker 04: And it can intercept the data stream as it's coming through, then analyze that data stream for signifiers of executability. [00:26:31] Speaker 04: And that's described in the specification. [00:26:33] Speaker 04: There hasn't been a challenge on that basis. [00:26:35] Speaker 03: I understand. [00:26:36] Speaker 03: I'm just trying to understand how it works. [00:26:38] Speaker 03: And as I said, because the level of specificity in the reference [00:26:43] Speaker 03: and looking at the part that kind of matters, since we're all looking at the same limitation. [00:26:48] Speaker 03: OK, thank you. [00:26:50] Speaker 01: OK, we'll restore your rebuttal time if the county council addresses it across the field. [00:26:55] Speaker 01: Thank you. [00:26:56] Speaker 01: Mr. Kinken? [00:26:57] Speaker 00: Thank you, Your Honors. [00:26:58] Speaker 00: I think with Dabrowski, the court has it right. [00:27:02] Speaker 00: The disclosure shows there's two inputs, and two inputs only in Figure 1. [00:27:12] Speaker 00: And then it makes a determination based on those inputs. [00:27:16] Speaker 00: And that's what Dr. Gilson says. [00:27:18] Speaker 00: And that is substantial evidence supporting that finding. [00:27:22] Speaker 00: The gateway device in this disclosure, that's what's showing. [00:27:26] Speaker 00: Some alternate disclosure. [00:27:27] Speaker 00: There could be other ways of determining executability. [00:27:30] Speaker 00: But for Dabrowski, those are the two inputs. [00:27:32] Speaker 00: That's the information. [00:27:33] Speaker 00: That's where the gateway device makes the disclosure. [00:27:37] Speaker 00: And then turning to the discussion of Apple, I think [00:27:42] Speaker 00: Somebody's probably misreading that case here. [00:27:44] Speaker 00: But here's my understanding of what happened in the Apple case, the Andrea case. [00:27:47] Speaker 00: And I obviously invite the court to look at it and reach its own determination. [00:27:52] Speaker 00: But what happened was the petitioner in that case cited an example of an algorithm using one subwindow and said, if you have this one subwindow and you combine it with Hirsch, you have all the elements of the patent that were challenging. [00:28:04] Speaker 00: What the patent owner said in response is, you can't use one subwindow. [00:28:08] Speaker 00: If you only use one subwindow, then you won't combine it with Hirsch, which is what you need to do in order to get the combination that invalidates the patent. [00:28:15] Speaker 00: They said you have to use two subwindows. [00:28:17] Speaker 00: And if you use two subwindows, which is a different example, then it doesn't meet all the limitations. [00:28:22] Speaker 00: So then there would be a motivation to combine it with Hirsch, but then you would not meet all the claim limitations. [00:28:28] Speaker 00: What Apple said in response is, no, even under your example, the example that you provided with two sub windows, it meets all of the limitations. [00:28:36] Speaker 00: The board excluded that evidence, said it's a new argument, and this court overturned it. [00:28:40] Speaker 00: It said, no, that's a legitimate reply to the example that Andrea provided. [00:28:47] Speaker 00: And what the court said is, [00:28:50] Speaker 00: It is unreasonable to hold petitioners to such a high standard that if they choose to rely on one example of an algorithm, they must disclose all potential permutations of the variable or risk waiving the opportunity to further discuss other relevant examples in their reply. [00:29:04] Speaker 00: That's exactly what's happening here. [00:29:06] Speaker 00: We provided an example in the petition [00:29:09] Speaker 00: using the exact example that was shown in Gribble, which was when you're looking at web objects and that you can use the file content field, although we didn't limit it to that and we never mentioned that, but we just used that example because that was in the patent. [00:29:25] Speaker 00: Then what Huawei said is, well, if you use that combination, basically it doesn't work with Moschuk because that's a root page. [00:29:33] Speaker 00: And we said, okay, if we're using the example of the root page, [00:29:37] Speaker 00: that here are the other fields in the HTTP header that you will look at. [00:29:40] Speaker 00: And the file content script is an example because we presented evidence from the HTTP specification that in order for that to operate correctly, if you have an executable on a page, you need to have a script definition. [00:29:54] Speaker 00: You need to have that field there. [00:29:57] Speaker 00: to tell the page what to do. [00:29:58] Speaker 00: So this is something, and we submitted evidence that when a skill in the art would know, and therefore they would look to that tag, that is a legitimate reply to their argument on page 451 of the appendix, that there is no indication in the HTTP header of whether there's an executable script on a web page. [00:30:20] Speaker 01: Thank you. [00:30:21] Speaker 01: Mr. Courtney, I'll restore your two minutes of rebuttal time. [00:30:23] Speaker 01: Of course, it's limited to the cross appeal. [00:30:26] Speaker 04: Thank you, your honor. [00:30:27] Speaker 04: And with that time, I'll just direct the court. [00:30:29] Speaker 04: My colleague pointed out, well, Dabrowski only has these two inputs. [00:30:33] Speaker 04: How else could it be doing it? [00:30:35] Speaker 04: The court has addressed those kinds of circumstances. [00:30:37] Speaker 04: Before, we had directed the court to Institute Pasteur, 738 F 1337. [00:30:42] Speaker 04: There, you had prior art references that came awfully close to describing [00:30:48] Speaker 04: cleaving chromosomal DNA, but did not actually describe it. [00:30:52] Speaker 04: We had testimony from experts describing how, oh, we would get very close. [00:30:56] Speaker 04: This is almost there. [00:30:58] Speaker 04: And this court vacated that, said it was not sufficient. [00:31:01] Speaker 04: Even more on point, it's non-precedential. [00:31:04] Speaker 04: It's the Google versus at-home bondholders case. [00:31:07] Speaker 04: It's in the briefs. [00:31:08] Speaker 04: That case, I think, is very apt here. [00:31:10] Speaker 04: It uses the phrase, silence is not a genus. [00:31:13] Speaker 04: The theory was ventured that because it was silent, it somehow disclosed all possible ways of fitting that determination into the blank box that was present. [00:31:22] Speaker 04: This court squarely rejected that, and we think that guidance is persuasive here. [00:31:27] Speaker 01: Thank you, Mr. Horton. [00:31:28] Speaker 01: I thank both counsels. [00:31:29] Speaker 01: This case is taken under submission.