Oral argument — Daedalus Blue LLC v. Vidal

Next case for argument is 23-1312 Daedalus Blue v. Vidal. How do you pronounce the name of your company? Daedalus. Daedalus. Yes. Sorry I messed that up. That's okay. Greek mythology, inventor, all of those things. And Blue, IBM. So, Daedalus Blue. If you see us again, leave a message. Please proceed. Thank you. Good morning, your honors. Denise DeMori here on behalf of Daedalus Blue. So we are here on an appeal from a PTAB decision of a patent that I think the place to start is the original idea of the patent expressed in the specification, which is to address the user's loss of control over scaling the number of virtual machines running an application, which was currently carried out manually by a system administrator. So the patent is all about effectively controlling the number of VMs across which a specific workload can happen. And so it says, the invention says, the claim says, create a template for this VM. That template says what memory I have, what CPUs I use, and what applications are running on it. I create that template, and then I deploy that template. I then monitor that template against certain criteria. If I get to a certain threshold, I deploy, I create another copy of myself. So the flag that exists here, flagging for autonomic scaling, is a flag that tells the system this particular instance of this VM can either be cloned, duplicated, you can add another one of me, or you can terminate me, which hopefully isn't going to happen to me anytime soon. But the context of it is, you have a machine that can be replicated, added, and taken away. And it is a very elegant system for solving the problem of, at the time, not being able to control the number of virtual machines on which a particular workload was running. So, I mean, just to give it an example, in the context of a law firm, I have a machine processing documents that are produced by the other side. I get a dump of documents. I have depositions in three days. The system knows that it can scale up. It can add virtual machines. And when the load is no longer there, I terminate those virtual machines. What's critical here, this is the context of cloud computing. So, of course, you're paying for these resources. They don't pre-exist. They get created in your template. And then once they're no longer needed, they're terminated. And you don't have to pay for them anymore. That is the entire context of the patent. So when we look at the, it's these two features that are sort of deploying an instance of a VM, and terminating, flagging for termination, that are the two claim elements that are critical here. And our challenges here are claim construction issues. So, and I think both instances, the boards contended it was sticking with the plain and ordinary meaning. And what you were advocating diverged from the plain and ordinary meaning. Is that the way you understand it as well? I don't, I mean, I understand there's a hint of that. I don't think that's actually what happened in the board's decision. So if we start with the deploying term, and we go to the board's actual decision, and I believe that is at page, appendix page 17, they actually, the board actually starts with petitioner, in turn contends that the patent owner improperly proposes changing the word deploy to create new into the claims. And so they say the plain meaning of deploy is, and they quote a dictionary. And they say the dictionary is bring into effective action, which is unrelated to the, it's not in the specification, it's a dictionary definition. And ironically, when I reread this, it goes on to say, petitioner further contends that the specification consistently uses the term deploy to describe the process of setting up the VM. So it is creating. I mean, it's setting up the VM. It doesn't exist before you get there. That's the whole point of having a template. There's, you know, you have a general purpose computer, and the template is. I thought when I was reading the patent spec, and looking at the figures, it seemed to suggest there were like a bank of VMs waiting on the sidelines, and then ready to be activated when called upon, which is, therefore, deploying. That's the prior art. Instances of virtual machines. Yeah, that's the prior art. That is, I mean, that is absolutely not what the six months. So you're saying that your patent describes your claimed invention in the sense that once a threshold is passed, the system is going to create a virtual machine from scratch? Yeah, I mean, that's how most cloud computing happens, right? So you have a blade server. Well, I don't know if that's how all these. Yeah, I mean, it's not only what the specs describes. So let's just start with the specification. I mean, we can just go directly. Creating instead of deploying, why didn't you say creating? I mean, those two words don't mean the same thing. So let's actually just look at the claim. I mean, this is something that has not really been the focus here. But when we look at the claim, there's deploying, and then it says, the final element says, and deploying an instance of a VM further comprises. Do you want to tell us where you are from? Claim one, the final element of claim one of the 612 patent, which you can look on the front cover of the blue brief. But the final element says, deploying an instance of a VM further comprises, passing by the self-service portal user specifications for the instance of the VM. So those user specifications are, I want a virtual machine with a certain amount of memory, and we can see this in figure two, which the office relies on. I want it with a certain kind of memory, with certain applications running on it, et cetera. That doesn't exist. It's not out there in the world. There's a template for it, just like there's a template for ordering. If I wanted another copy of my computer, I could go on the Apple site and order the same one again. But it doesn't exist. What exists is computing resources. So passing by the self-service portal user specifications for the instance of a VM to the deployment engine, implementing and passing to the data center administrator by the deployment engine a VM template with the user specifications, and calling by the data center a hypervisor on a cloud computer to install the VM template as an instance of a VM on a cloud computer. That's deployment. That's in the claim. That is create. I mean, whether you use the word creating, installing, it doesn't pre-exist. They're not just out there waiting to be picked up. How do we know that from what you just read us? It doesn't use the word create. It doesn't use the word new or anything like that. How do we know this isn't all just predetermined templates, and you're just calling up an additional version of it? Because it says install the template. And even if we read the director's version of what happened. How would you install inconsistent with having a previous one that you just deployed? Let's look at the template. Let's go look at figure two of the patent itself. So are you now saying there's nothing in the plain language of the claims that we're looking at, so we're going to have to read in separate specifications to determine that there's a difference between the two? No, I'm not saying that at all, because I'm reading the claim here, and it says. Well, you're moving on from the claim language. So I take it you don't have anything further to add in response to how the claim supports your argument. I was answering your question, which is how do we know that the template is not something that. . . I mean, the template exists, and the template is a thing that is not disputed. And the template, I mean, we can read it from the director's brief. We can read it from whatever. The template is depicted in the specification of the patent. It's not something that's ever been subject to dispute. Figure two, there's a VM record. Let me ask you, what the board said about this at 19, I think we're talking about the same issue. Maybe not. It says the specification does not support patent owner's contention that deploying requires creating a new instance of a VM, but rather suggests that an existing VM can be deployed by moving it from one cloud computer on another. Yeah, so, I mean, that moving reference there has nothing to do with what the hypervisor does with installing. There is one reference to moving in the specification, but it's not even the piece in the claim that we just read that actually installs the VM. So this is like the data center, a VM can be moved somewhere in the data center. But that moving quote, there's no relationship to the language in the claim that we were just looking at, which is calling by the data center administrator a hypervisor on a cloud computer. So the hypervisor on the cloud computer installs the VM template as an instance of the VM on the cloud computer. Does the specification use the word create in lieu of deploy? It doesn't. I mean, so it could be install. I mean, the point is that the actual – Well, install doesn't mean create either, right? Well, we're talking about a computer. So it's, right, it's not writing the software for the first time, but it is – I mean, it's a cloud computer system, and it's just a bunch of generic resources, memory, CPUs, et cetera. The template is I want a MacBook, essentially. It doesn't have a screen or anything, but it's I want a certain kind of processor, I want a certain kind of memory, and I want certain applications running on it. That's the template. And so we're talking about we have generic resources, and then we create an instance of the thing Denise wants her application to run on. To create that, we have to install the software on there. We have to say run this operating system, add Denise's eDiscovery tool to the system so that it can share this workload. And we know it's not just existing out there. It says install the VM template as an instance on the cloud computer. How do you understand VM template as opposed to an instance of a VM? I mean, it could well be that the VM template is, like, here's the type of virtual machine instance that the user is looking for, and then the system will go back into the back room and look at all the available VM instances that are in there and say, ah, I found one in aisle 14 off of shelf number 3 that looks just like the requested, you know, specifications for what an instance of a VM the user needs. I think Kigi even helps us with this. I mean, if you read the background of Kigi, it talks about exactly this, which is you. Well, I guess we're just a little bit lost here in the patent because you're asking us to, you know, like the board was confronted with the same problem, which is the claim says deploy. You're interpreting deploy to mean create. And now the board is trying to figure out, okay, what is it in the specification that could make us feel good about the idea that deploying a VM instance is really all about creating from scratch a VM instance? I mean, the specification. I don't see it in figure one, that kind of a feeling. In figure one. Let's just look at, let's go ahead and look at the figures then. And, I mean, even if you look at the director's description of it, but I think it's more, so it's figure two is the one that the director uses in her brief, actually. What page are you on? Appendix 779. So you start and you receive through the user interface exposed by a self-service portal user specifications of a VM. And user specifications are 174, which is the template. And then if we go to, we look at how, I mean, the director walks through figure two and says exactly the words that I've just pointed you to on the claim. In order to set up, this is on page six of the red brief. In order to set up a VM, this is the director describing how it works. A user provides a VM specification to a deployment engine, which selects an existing VM template from a catalog and installs the VM in accordance with the template. And when an additional instance needs to be deployed, the system installs this additional instance. I mean, I really think we're talking about a semantic difference between creating and, I mean, installing, making it happen. I mean, there are. Installing could mean activating, i.e. deploying. I mean, I don't see anything in that description from the director's brief that suggests, OK, we've got to go into the kitchen and figure out how to cook up a brand new virtual machine that is tailored to exactly what the user wants. We've got to put in a little of this. We've got to put a little of that. We've got to mix it together. We've got to bake it. And now we've got a new virtual machine that's tailored, customized for the user. I mean, this figure doesn't say that. It says the user wants what the user wants. That gets plugged into the system. And then BAMO, the system, deploys an instance of the VM. By installing what the user wants on it. So, like, if the user wants Word. So if the user wants four CPUs and 32 megabytes of memory on a Blade server, the system has to go and allocate particular CPUs, allocate particular memory, segregate it from everybody else that's using the resources on that thing, and receive the user specifications, and then install Windows or Mac OS, install whatever programs the user wants on there. I mean, I don't think this is actually in dispute from the director's perspective. So then you're saying the director's conceding that your claim construction is correct? No, I'm saying that the director concedes by this description that there is a template. The template includes the resources that the user wants, plus the software that the user wants running on those resources. I mean, it sounded to me like you're saying this is all going to be customized, and each user is going to have very specific requirements, but does this exclude standardized templates where the user just gets to pick and say, I want this version A, this version B, this version C? Well, I mean, that's not what the patent describes. I mean, I don't think this aspect of the patent. So redeploy as creates. Does the specification ever describe this as creating something? I mean, I think the claim itself says the steps that are required. Pass? You can keep saying the steps, but you want us to use a different word or a different definition than the claims actually use. So does the specification somewhere suggest that deploy was actually meant to be defined as create? I think everything. Wait a minute, stop. Is the word create ever used in the specification to describe deploying something? The word is not. It's a summary word to describe the process. The word create is not used. The process is past specifications. So there's a front end where the user has the specifications, they put it in. Pass it to somebody, implement it in the data center, so allocate resources for it, call the hypervisor, and then install all the stuff that the user wants, that the user has specified on it. That is the process of deploying that is set forth specifically in the claim. I am way over time. Yes, you are. Thank you. Let's hear from the other side and what we saw. Thank you. Thank you, Your Honors. Peter Ers on behalf of the United States Patent and Trademark Office. I'd like to go first to Judge Post's point because I think that's really the salient one here, and that is that the board described these claims, their plain and ordinary meaning, and the patent owner in this case wants to argue for some kind of idiosyncratic definition. Well, except can you be a little more specific? I mean, she pointed to at least two things. One, explicitly the claim language. Yes. And two, figure two. So why do those not individually or collectively lead us to some sort of morphing, deploying into creating? Certainly, Your Honor, and the reason is that even if we credit all that we've heard here today about how the 612 patent describes this invention, that is merely the preferred embodiment. We see that on APPX 783, column two, in describing these figures. It says, figure one sets forth the functional block diagram of example apparatus. Figures two through five, which was the focus of the argument, set forth flow charts illustrating example methods of autonomic scaling. And then go down further in the detailed description of the exemplary embodiments, column two, starting at line 46. Example methods, apparatus, and products for automating, autonomic scaling of virtual machines are described in those figures. And the question is, should we read the limitations of the preferred embodiments as shown in those figures into the claim? And the answer is no, because there's no clear reason to do so. And ironically, the definition that patent owner ascribes to deploying doesn't even do the most important thing that deploying should do, and that is putting it to work, making it do something. They just want it to be created, and they stop there. Are you saying that there's something in the spec that is a preferred embodiment that in fact is creating? No. I mean, I understood your argument to be, oh, well, this passage is just a preferred embodiment. That's not included in the claim. And I couldn't tell if what you were essentially implying is, okay, this passage, it may in fact be about creating a virtual machine instance. I was crediting their argument, giving it its full, strong support. But the patent itself does not actually use creating. It talks about passing a template, installing all the things that were discussed. Right, and so that takes us to this where-in clause at the very end of the claim. That's right. of this various steps, and this feels a little bit like a late plot with a twist in the show because I don't remember this being that much of a focus in the briefing. But can you walk us through this final set of claim language? Yes. And explain why in your view this either individually or collectively- Certainly, Your Honor. doesn't amount to the actual act of grabbing various existing resources to essentially manufacture a new virtual machine from them? Certainly, Your Honor. I think it's important. There's been some discussion about the file history, and if you go back and look at the claims as they existed as they were on appeal, they did not include the where-in limitation. That was added by examiner's amendment after the appeal. All of this was in a dependent claim, further narrowing what deployment meant. And imagine if this case had been presented to this court with that claim as a dependent claim. We'd be in here arguing that this is claim differentiation, and that's essentially what we have here. We don't dispute that deploying has a number of different sub-steps to it. We see that in the flow chart. We see that in- Can we walk through these actual sub-steps? Certainly. Yes, Your Honor. That's really what I'm interested in understanding as a technical matter. Yes. What is going on with this self-service portal? What is going on with this data center administration center? What is going on with this hypervisor? Right. Well, we don't dispute that these VMs are tailored to the individual user, right? If you want a virtual machine that runs macOS, you need to specify, okay, I need macOS, right? So those are the user specifications, and then there is a template that's associated with that. You might need an Intel processor. You might need one gigabyte of memory. You might need a certain IO input-output in order to access a certain legacy system or something like that. We don't dispute that you have to make those specifications, and they have apparently prefabbed templates that you can use to sort of construct your template. And I don't think there's any dispute that at some point these virtual machines have to be assembled and created in some form. Okay. So that's the buzzword, created. Are you saying that what this claim language here at the end of the claim amounts to is assembling a virtual machine, creating a virtual machine? These are sub-steps of deploying a virtual machine. And do the sub-steps include actually creating the virtual machine? Your Honor, with all due respect, we think that that's a red herring because the claim does not use creating, and we also know that in the prior art, those virtual machines have to be created at some point as well, right? They don't just appear out of thin air. The virtual machines also were created in Miller and in KD at some point. Your only question is, is part of this claimed deployment, must they be an essential part of this step, or could they have been created somewhere in the past and just be sitting there waiting to be deployed, not unlike an army that's being deployed, you know, just in its ordinary use? Right, but, you know, I'm waiting for you to talk to me about this passing sub-step, this implementing sub-step, this calling sub-step at the very end of claim one and walk through it and give us your view of what's the right understanding of any of that and why those steps, either individually or collectively, still do not amount to the action of creating a VM instance because in your view, this is really about other activity, I guess, dealing with an already existing virtual machine? Or not. I mean, if you want to tell us this is part and parcel to actually creating the virtual machine, then tell us that. We would concede that, Bob. We don't care, and the claims don't care, whether or not it previously existed. I'm getting worried because this claim, you know, earlier in the middle of the claim, it says, of course, deploying an additional instance of the VM, and then later calls back to that limitation, says we're in blah, blah, blah, deploying an instance of a VM further comprises colon, and then all these further steps. So whatever these steps are, if you're saying that they encompass the creation of that virtual machine instance, that's necessarily part of the act of deploying the instance of the virtual machine. And so, therefore, deploying now has embedded inside of it, right inside of the claim language, the notion of creating said instance of virtual machine. We don't dispute that these sub-steps are part of that deploying step, and the Patent Owners case conceded that these steps were satisfied by the prior art. Right? They did not dispute that those elements were satisfied, and I could point you to the board's final written decision where they didn't dispute. Well, I guess I'm getting more and more confused. I mean, that's a different answer. It could be that the prior art had creating as well, and that's why this case goes down, but that's not the way I thought this was being argued. Right. The question is really did these have to – could these virtual machines have pre-existed the deployment or not? And the answer is yes, that the claims are broad enough to deploy pre-existing virtual machines. They don't have to be created at the time. Well, the difference – and I didn't hear you respond to Judge Chen, but I thought the difference, and I'm not sure how much of a difference it was, but you were saying bring into effective action. So I guess the opposite of that was they weren't effectively in action beforehand, and the board said and the petitioner said that's different than bringing it into existence. That's right, Your Honor. Which is creating. That's right, and there's no dispute that in the Miller reference that those virtual machines existed, and the only question was – Well, how do we know, though, that that bringing into effective action is what's going on here in the claims versus bringing it into existence? How do we know that that's what's – the difference between the two of those was going on? Let me be very clear. I mean, it seemed – I thought I heard you say that these final sub-steps do encompass creating a virtual machine at the time that the user is requesting resources in order to do some processing. And if you admit and agree that these final sub-steps do encompass the act of creating a virtual machine, then you've given the game away. Then I can't do anything except accept the idea that deploying necessarily includes creating. So did I misunderstand you? I believe you did, Your Honor. I certainly didn't give the game away. What I said was that deploying includes, as we can see from the claim, these additional subsidiaries. Whether that means creating as that they have used it in the term, to me it's sort of beside the point because that's not what the claim says. It admittedly requires these additional steps as part of the deploying. And, in fact, as I said, they can see that the prior art performs these additional steps, so those are not in dispute. And so presumably that could be done with these existing virtual machines, whether they're created in the first instance or not. We still – I still haven't heard you explain what these sub-steps are. Are you familiar with these sub-steps? I don't know if they've really been highlighted adequately through the briefing and ventilated, but in order for me to affirm the board's claim construction, I have to reach a conclusion that these sub-steps don't encompass and are not directed to creating a new virtual machine instance at the time a user requests resources to do some workload processing. So you are telling me whatever's going on in these sub-steps is not creating a new virtual machine at that time. So if it's not doing creating a new virtual machine, what are these actions doing? They could be fulfilled by existing virtual machines. They don't dispute that. Okay, so by definition that is – Yes, they do dispute that. Their argument is no. When you are taking in all these user specifications, requesting a virtual machine instance, and then you are, you know, doing something at the deployment engine with a VM template, and then you're calling a hypervisor to install the VM template as an instance of a VM on the cloud computer. Well, if I – Can you translate that for me? I mean, I can do my best to explain it, Your Honor, but before I do, I would just point the court to APPX 50, where this whole where-in clause is discussed, and then we go to APPX 51, and it says, patent owner does not directly dispute petitioner's contention. The only thing that they're contending is whether or not deploy in the first instance requires creating it out of whole cloth, but it doesn't dispute that these virtual machines can satisfy these later, or at least as described in Miller, satisfy these later elements. But I can describe this further, but I think it is helpful to understand, to step back for a moment and think, what if these steps were in the dependent claim still as they were presented? We wouldn't be limiting, even if that does describe creating, okay? We wouldn't read those limitations into deploying, right? And that's effectively what they're arguing here, is that you should narrow deploying to creating, but what they're saying is that these more specific steps are what implement creating. Certainly we wouldn't read those into – we wouldn't read create in if those limitations appeared in the dependent claim. That's how they were presented. I guess what I'm trying to figure out is, in the end, when it says install the VM template as an instance of a VM, why aren't you just saying install doesn't mean create? Again, I don't know what they mean by create. Apparently they mean by create what's recited in these steps. And perhaps more. I don't know. And that's the problem with reading the limitations from the specification. There might be other steps that are also described in the specification that are part and parcel of deploying a virtual machine in this scheme. And yet, where do we stop? And that's why we have the rule that we don't read limitations from the specification into the claim, absent some compelling reason to do so, and they've given you none. They said this is what the preferred embodiment describes. That's not enough to read those limitations into the claim. Okay, can I – in your – I'm looking back at your brief now, because I recall that at least for the first issue, for the flagging issue, you made an alternative argument that even if you accept patent owner's claims instruction, the prior art still reads on it. Did you make a similar argument – I mean, today, in responding to this deploying thing, you've been making comments about how all this stuff was in the prior art. Is there an alternative argument presented that even accepting the patent owner's construction, or not differentiating between creating and deploying, whatever it is, it still reads on the prior art? There is an argument – well, as I understand it, there were two theories presented. One was Kege alone describes adding and terminating because it describes scaling up the resources of the virtual M to sort of an unlimited extent, which would include down to zero resulting in termination. I don't know that that directly answers your question, your honor, about whether we presented an alternative that goes directly to this deploying. But I think that the board's claim constructions are correct, both flagging and deploying. And we haven't really talked about In re Smith, and I think that that really sort of shows the extent to which they're trying to reach in this case to claim something that the claim language itself doesn't describe. And I'm happy to discuss that and why that case is not a point. Can I ask a very basic, ignorant question, which is you've got two different terms in the claim. Does the board's analysis on both of those have to prevail in order for these claims to go down, or is one of them sufficient? There we go. Are they in different claims? They're independent grounds, yes. They're independent? Yes. Okay, because they were in different claims. They're in the same claim, but either one of them, if absent from the prior art, would be grounds for vacating the board's decision. Okay. Any further questions? No, thank you. I'm dissatisfied with my answers. Thank you, Your Honor. Thank you. Okay, we'll restore three minutes of rebuttal. Your Honors, I think what we just heard actually establishes my point perfectly. Those final three elements, they are not in a dependent claim. They are in the independent claim. They are further elements of the word deploying, so they must be part of deploying. To start with, if we look at APPX 51, which is where the office just pointed us, and it talked about how we conceded that these elements were in the prior art. But if you look at what the petition says the elements mean, it says, Petitioner contends that Miller's LBM receives VM configurations from the self-service portal and uses them to create new VMs with those configurations. So in the petition, to meet these final three elements, which define deploying, the petitioner pointed to using a self-service portal to create a VM. Now, if we go to the specification, we were asked about, like, you know, what do these elements mean? Well, the specification, there is no other embodiment. There aren't alternative embodiments. There's one description of how this works with the templates, and I will tell the court I'm not going to read it all, but it starts at the bottom of column four on page APPX 784 on line 63, and it says- Hold on, column four. Column four. So it starts with, in the example of the cloud computing system of figure one, self-service portal- I'm sorry, what line in column four? I'm sorry, 63. Okay. In the example, figure one, self-service portal exposes the user interface for access who is authorized to install a VM on the user system. Let me go over to the next column. Let me move to line-I'm just doing this quickly because we don't have much time-page seven. The user specification includes for each- I'm sorry, where are you now? Column five, line seven. Five, line seven. Okay. So this is using the portal. The user specifications include each VM, specifications of the computer resources to be provided as a VM, types of different processors, quantity of random access memory, hard disk storage, input output, applications, et cetera, and it describes that. Then you move down to the next paragraph, and now it describes what happens. And this parallels the language and the specification and the claim that I just read. So your view is basically that this entire process and the entire invention here could never use already created virtual machines. The way that works is when you need more resources, somebody puts in a request or it's automated, however it's done, with what it needs, and that process will always result in the creation and deployment of a new one. Yeah, I mean, that's what the patent is all about. I mean, it says- But we're looking for the best language that tells us that. The best language, yes. So here I was, column five. Beginning line 20, it starts with the self-service portal. It describes in this paragraph, I will just point the court to it, but the self-service portal, and it describes how it passes. The deployment engine selects a VM template that matches these specifications, and it describes if it's an Intel processor, et cetera, et cetera, exactly what I said in my opening thing. Then the data center administrator at line 34, the data center administrator server then calls the hypervisor on a cloud computer, same language in the claim, to install the instance of the VM specified by the selected completed VM template. The data center administrator server records a network address assigned to the new instance of the VM, as well as a unique identifier. Each time it does this, the whole patent is about specify what I want and then clone me. A new instance, I mean, there aren't many. This might be the only reference to a new instance in this sentence, and it could just mean newly deployed instance. Well, it says, I mean, so. And this is the question mark for me all along, which is we're talking about all these user specifications and then get a design of exactly what it is that the user wants, and then all of a sudden the patent talks about, then you deploy the instance of the specified VM, and it doesn't itself explain how it got from A to B. How did it get from the receipt of the user specifications to the now deployment of the instance? And that is, there's a gap in this patent in terms of how we get from one point to the other. Did you have an expert declaration that says, here's what's going on here? I don't recall if the expert declaration describes this exactly, but I think if you go to page 51 of the final written decision and you look at how the petition meets those claim elements, in each case, I mean, when the petitioner's talking about, just like the office did here, when the petitioner is meeting those final three elements of the claim, it's always talking about creating. And they actually, the petition actually uses the word creating. Where are you on 51? In the final, it was appendix page 51, I believe, which the solicitor's office pointed us to. And it was how we, it's the discussion of how we, how the petition describes meeting these final three elements of the claim. Let's see, where is it? But that's just talking about Miller. I mean, Miller, they contend that Miller then uses them to create VMs with these configurations. Right, so I guess my point is, so if we just back up for a second, how do you meet these final three elements of the claim? Can I ask you, does you, let's say there's a hypothetical system out there that doesn't allow the user to ask for specific customized VMs, that you're given a menu and the menu is option A gives you this kind of resources, option B gives you this, option C gives you that, the user does that and picks one of them, and then the hypervisor or whatever already pre-stored a bunch of different versions of A, B, and C, and all they do is deploy all the already pre-stored. Does your patent, would your patent cover that? Or if that's the system, it doesn't infringe the patent? I don't think so, because this deploying... You don't think what? I don't think it would cover that, because this deploying step says, I mean this is cloud resources. The cloud provider, this is in Kigi, this is in the expert declarations, I don't think this is disputed. The cloud provider wants the resources to be available and paid for by as many people as possible. So we don't just set aside Intel processors for people to come along and use with Word installed on them. It's just not the way the systems work in the first place. But second, the actual deploying step says not just go grab something, but install the things that I want. And when you read the... Right, right. But what I was trying to get at is if the install the things I want are limited to choices that have already been pre-selected, then it could be the fact that they're just deploying instances of already created VM terminals. Do you think that your patent can't be read so broadly as to include that? I don't think it is to include that. And I think the whole purpose of the patent was before a system administrator had to sit there and specify, I want this processor, I want this stuff, and then the system would go allocate the resources. This is all about autonomical scaling. So it's saying, I have a workload, I need all these documents processed, I want it done on this computer. If this one gets too busy, clone it. If it gets too busy again, clone it. If it's not busy anymore, terminate it. I don't want to pay for it anymore. Terminate it. What did the board say about these final clauses of Claim 1? The board didn't really address it. Well, did you argue it to the board? It was in the patent owner response. It absolutely was in the patent owner response. I won't give you that page right now. It's appendix 455, page 9 of the patent owner response. I mean, I do believe the director was being forthright in conceding. I mean, these final steps are about creating a VM. The same thing goes with regards to flagging for terminating, which we haven't really got to, I would say, very briefly. But no, you're not going to get to flagging at all. But why do we have to read the word install at the end of the claim as necessarily meaning creating? I guess I'm still a little bit stuck on that. Creating is not a natural synonym of deploying, and I don't know why creating is necessarily the same thing as installing. Well, so I think deploying here is bigger. So I think this is one thing that the director was saying. Well, just assume for now that the most natural understanding of deploy doesn't include creating. Now I've got to ask myself, what does the word install mean? Does that somehow trigger an understanding that we're doing some creating here? So I think I went very quickly through the end of column four and column five, but it really does talk about the entire process that I described in my opening argument, which is you have a bunch of processors out there and you have a bunch of memory. This whole patent is about pick the specific template that you want, run a workload on it. If it gets too busy, clone it. If it gets too light, delete it. That's what the whole patent is about. The claim doesn't say clone anything, but I understand the point. I'm simplifying for sure. But if you read column five, that's exactly it. So when it's creating, it actually is allocating a particular processor. I mean, it's described here. It allocates a processor. It allocates an I.O. bus. And these are physical resources that it has to pick. Well, I'll make sure we read column five. Yeah. And then I guess my very final comment is deploying in the claim is these final three elements plus putting the workload on it. So in that context, it is deploying. It creates it and deploys it and puts it to work. So thank you. So thank you.