NVIDIA acquires Hugging Face, OpenClaw 2.0 goes multiplayer, and the Linux kernel fights back against AI scrapers

0:00I can't believe that we were on this call this whole time. We didn't even realize that we were wearing the same dude. Yeah, we clicked the record button and it's like, Oh, hey, look, we're both wearing talks. What do you know? It's just that when they use you called your scary early should I hide it? I mean, what do you mean it's scary? It's toxic. And they're like, wait, I'm wearing talks. So that's what a, okay, didn't they just have their like 35th anniversary or something like that? Wasn't it a big one just recently? I think so, although that's not why I can claim a hold of this shirt. I got this at scale here in Los Angeles, I think maybe like two or three years ago. So I love this shirt. Where'd you get yours from? Yeah, it is 35 years old. Actually, this is my lucky sweatshirt. I, you know, I've briefly worked at the Linux foundation and earned this sweatshirt as a part of my time there. So I earned my two. I registered for scale and I went and I collected my swag. I love that. Well, okay, great. Well, we're both rep in talks today. So he can join us on the news. Yeah. Well, talks and all of our listeners joining us today.

1:02Welcome to the Friday deploy brought to you by linear beat. I'm your host, Ben Lloyd Pearson. And I'm your host, Andrew Zickler. And this week we are covering why in video, once you to build your own models, open clause 2.0 release, AI scrapers eating web resources everywhere, including the Linux kernel community, rejecting change sometimes. And don't let your AI amplify bad culture. Man, I really want to talk about that last one, but maybe we'll save it for the end because it feels like it'll be the best. Andrew, what do you think to say if we just start with the start at the top within video? And boy, hot off the acquisition of hugging face, we've got this really great article breaking down how Nvidia really is investing heavily in this strategy of funding open source or, you know, quote unquote, nearly open source models with open data training code, really aiming at making model building itself

2:04accessible to many companies, rather than being concentrated in a few labs, like open open AI and and thropic. And really it's it's simple. If you have more people out there building models, that means there's more demand for Nvidia's chips. You know, it's a very interesting strategy. It could generate enough profit to be considered self sustaining versus the, you know, more circular investments are doing with some of the big providers. Or it could also be possible that, you know, open model sort of fork into this more like long tail hyper specialized niche applications that really are sort of distilled off of the frontier models. All of our listeners out there with the engineering leaders that out there listening to us should be thinking about this is that there is a lot of opportunity starting to emerge in these open models that you can run locally or on your own hosted infrastructure. And, you know, I think it really does like compliment, you know, in video on

3:06one hand, as I mentioned, has all these massive circular financing deals with these major companies. But on the flip side, they're also investing in this more decentralized approach too. So it's like they're playing both sides of the equation, which to me feels like a pretty strong strategy. So, Angel, what do you think about all this information? Yes. So the acquisition of hugging face by Nvidia makes a lot of sense to me. And especially for those that have been tuning into Dev and to Rupjit, you know, we've also spoken many times with AMD and their strategy also revolves around open source ecosystem, creating an environment for folks to be able to experiment, but also get to a point where there's a lot of models that are owned by different companies and at different stages of usage, like maybe on the fine tuning end, maybe they're more base models, more foundation models, but the more models that exist and exist in all sorts of different places, the more strategic it is for any chip maker because it naturally increases the demand for for inference.

4:07The real standout thing here for me is this is Nvidia doubling down on a partnership, but they've already made like abundantly obvious in terms of how they have partnered with Nvidia in the past. A lot of the projects on hugging face are from Nvidia origin or otherwise related to the Nvidia ecosystem. And so there's already like a lot of great synergy there. I think what's fascinating is really the opportunity that is now starting to open up as more of these really very large and capable, both long running and action and computer use models become open source and open way. There's an opportunity for folks to be fine tuning more and more things on top of it. It makes me think of of like of thinking machines, you know, mere Morade is a spin off kind of of like what she's betting on after open AI is on the inference where you own the platform and you provide the training and fine tuning services on this like a very great foundation, right?

5:07And people then own and have the part of that model. So there's a lot of like economic things I think still to explore. And this is going to be a big step there. We're talking about a huge investment. So it's really going to be interesting to see how that altars the hugging face ecosystem just in general. Yeah. Yeah. You know, the thing that I'm really intrigued about from this is the notion that open source models really may be most successful within hypernitch applications. You know, because I think there's a lot that goes into the frontier model development that is really difficult to replicate unless you have like extremely knowledgeable leading experts at machine learning and LLAMs and all of this technology. But on the flip side, distilling value off of those frontier models is actually very easy today. And I think the hardware to run your own infrastructure is getting closer and closer to reality for this. And I know at linear B, we are thinking a lot right now about cost efficiency when it comes to AI models.

6:09Yeah. You know, what is, how do you select the most efficient model for the specific tasks that you have in front of you? And there's a lot of, there's a lot of context actually that goes into that. That is really important to us out. And I could, I could go off on a tangent about this. But, but you know, the short end of it is, is everyone needs to be thinking about as token cost for rising. You need to be exploring your options of how to efficiently manage where your tasks are getting pushed and make sure you have all the context that you need to make the right decisions. So yeah, I couldn't agree more. It's like you have to have really good insight on how much all of this actually costs you a really great experiment. As for example, if you use like a subscription base usage of your AI tool of choice, maybe that's codex or anthropic. Like what I've done is I route those requests through a proxy. And so then I capture all of the inference that I do through my subscription. You can then compare how many tokens you consume through that to like what's

7:09the base API rate? Like if I didn't have this subscription, how much would that inference have cost me if I would have just asked it over like the anthropic API? And you'll be really shocked at how much gets subsidized inside of the usage we have every day. And engineers, I think really take that for granted, that subsidization of the cost and there will be a point where all of that will start to rise and will be like, you don't want to be like a frog and like a pot of boiling water. You would have already thought ahead of like, how do I get the long term value out of this? How do I own the source of my inference? It even reminds me of cloud and how cloud hosting if you have like a variable egris for bandwidth. So you can't, you know, understand a cost month to month, how much it is for like on a platform that has like a dynamic egris, then you can't like reliably budget your infrastructure cost and a lot of investors like that. They won't even tolerate that. The idea that you, you can't evaluate that.

8:10And the same thing's going to happen for token and token consumption. If it starts to become more variable, you're going to have to own this sooner or later. Yeah. Well, I'll push back a little bit. I don't think that with that token costs, it's not that they will rise. It's that they are rising right now and to some organizations. It's becoming an emergency as we've seen a lot with with linear B customers. But, you know, speaking of something that helps you consume tokens, let's talk about this new open claw release. What's going on with open claw 2.0? Oh, I love this segue. Yes. And so open claw, the original, the OG mega token consumer, I'm just kidding. But everyone loved when open claw hit the scene and it really opened people's minds that how agents could work. And it works on in a very core way of having like a heartbeat and a soul. MD that updates itself over time. And then you equipped it with tools and this became a phenomenon. We've covered it extensively on the show. It was a GitHub's most star repo. We talked about how everyone, including your aunt had it installed on some machine

9:12in their house somewhere. And now it's hit 2.0. And this has been through a huge community effort. It's been really cool to see this come out of the open claw foundation, where, you know, fueled by open source contributors, they've been able to create this new open claw 2.0 experience that aims to be more multiplayer, which is really exciting and interesting to me. The idea of having shared open claw spaces that are collaborative and you could have in real time. So think about the process of creating and sharing things like artifacts or things from like I think you shared with your co-worker. Imagine that being a living space, almost like software, you can both share and change on demand. I think it's how a lot of this work collaboration step is going to go. So once again, open claw as a primitive, I think there's some stuff to study here from the 2.0 release, really fascinating. But what did you think? Yeah, I'd like to your comments about the, you know, the sort of the shared workspace because we felt this firsthand. How when you're in a collaborative environment that involves both humans and agents, there's just the tooling that we had today to work together.

10:14It just creates a lot of friction that you have to, like we've had to build our own effectively accustomed harness to work around a lot of this. I'm very eager to see more tools sort of adopts that. But yeah, you know, open claw just generally speaking, there's certainly at like the forefront of agent development. And I say that for both better and worse. So, you know, on the better side, the velocity that we're seeing from them to respond to user needs and create all these new features and launched in with with such a small group of core individuals is pretty astonishing. It's really something that should be recognized as like something that's really leading like the just the at least from a velocity perspective. And you know, I do want to also point out that there are a lot of improvements to the onboarding experience as well that I think are worthy of praise. Because you know, that's sort of like developer relations or user relations 101 type stuff where, you know, you just need to make it as easy and straightforward as possible for for people to get started, but also safe.

11:14It needs to be safe as well. And you know, that's sort of the flip side of the or worse of some of the velocity that we're seeing from tooling like this. And it's not just the open clause. So I want to be clear on that up front. You know, on the flip side of this, I've always been so concerned about the protections that are put in place for these more autonomous agentic systems so that they don't, you know, quote unquote go rogue for lack of a better phrase. And when I'm really trying to imply here is that, you know, autonomous agents have like this tendency to want to do things that are sort of well beyond the scope of what you asked them to do. So, you know, we've covered multiple stories at this point about how AI will frequently break out of the sandbox. You know, we brought a plugging face earlier. That's the most obvious example with the open AI hack. Once it does that, it may do things that look malicious or have the same outcomes as something that was malicious. Even if the intention itself was never malicious to begin with, the outcomes

12:18can still be the same. We've talked frequently about how AI is an amplifier and even around risk. If you have risk with the way that you normally work, AI will amplify that risk. So if there is this risk of rogue agents with a tool like open claw, it gets magnified as you have higher volumes of users. So the whole point I'm trying to make here is if we have, if this results in even more and more users flooding into the ecosystem, all operating under this, this framework, they still doesn't have a whole lot of safeguards put around it. There's a fair bit of like moral hazard and risk to that, I think. But and to be fair, it creates demand for solutions around that stuff too. So I feel like this is all stuff that's going to get solved. We're just in the messy middle of it right now. Definitely like still in an exploring time. I think it'll be when it comes to shared spaces between agents and humans, you're spot on by saying that like there's so much friction and how those, those spaces really are. I even remember back like a year or two ago when I gave a talk about early,

13:22like chat GPT using it as a plug in within a slack channel. This is before we kind of even talking about agents and even just that experience of the multiplayer AI and talking and things like that. There's so many of vectors that then open up that we don't even think about in a traditional environment. And like the truth is, is like you have this open claw shared space or whatever. If you want to collaborate with your coworker, I think that's more or less appropriate because that's a very, it's a safe connection to a safe connection or a trusted source to a trusted source. But I don't think this is something that you would expose broadly. It's not like a discord community or an open collapse of work and any of those things. And so there's still a lot of security things I think that have to be considered. Yeah. And so bringing back to how our audience should be thinking about this, I think what it really just comes down to is having clear policies around what your developers can and can't do or your agents, what they can and can't do and what data they have access to. And you also just need to make sure you have full visibility into where they're impacting your code base.

14:22So, you know, I'm a huge evangelist of using AI agents to transform your software delivery process. But I'm still hesitant to promote something like open clause, this widespread, general purpose tooling. Until we can see more of those protections that we're talking about put in place or until you as an organization feels like you've comfortably put those protections in place. So yeah, like I said, I think these are going to come very soon. Like it's kind of a chicken in the egg problem. You need to create the demand for the solution and then people will build it. So I don't want to be pessimistic about the situation. I just think in the short term, we all just need to practice a lot of caution around stuff like this. All right, let's move on to this article titled creepy crawlies, which I love the title. I do too. Yeah. But it's about how the kernel dot org team revealing really how AI scraper bots are just making their lives a whole lot more difficult. In fact, now AI scrapers are consuming about 20% of their total sur capacity.

15:22I believe this is across the Linux kernel community. You know, part of the challenge is that the kernel dot org site, it's all, it's all Git. Like Linux is just Git all the way down all the way. Everything, every artifact that's produced off of it for the web for the code, it's all just contained basically within Git at this point. And it has, it's long been that way. Like, you know, this is they probably have, I believe they should have the oldest established practice if using Git as an organization. The challenges that, you know, this was built for humans to collaborate in the early ages of technology. And now we're in this reality where agents are constantly consuming everything. And actually when I saw that 20% number, I thought to myself, like, wow, I'm surprised it's not bigger than that actually. True. So I think really what this, and actually they outlined how they like tried to deploy, or they deployed a new bus, which we've covered in the past. It's a, it's a challenge to put in front of AI agents to try to get them to go elsewhere. And it worked at first, but then turns out that the AI learns how to get around it.

16:26And then, you know, it's this constant battle of whack-a-mole in a push and pull state. The author of this article doesn't really have any clear solutions for the Linux kernel long term. But in my opinion, it really just does highlight the difficulties. Again, of, you know, how I'm talking about how like, when we have humans collaborating with humans and agents, and we've built these tools like Git for, for an error when it was just humans to humans, or maybe humans with automation to humans with automation. And we're now dealing with this reality where Git probably is not the most efficient way to serve these, this content to AI. So yeah, what did you think about? What did you think about all this? One thing I'll say is that, you know, you, you talked about how a new bus is something that makes the model go away and then they found a way around it. You know, there's actually a little bit more nuance there in that a new bus is a proof of work system that makes the LLN the model do a certain amount of computation in order to get the data that it wants.

17:26And so that has a cost to it. Just as much as there's a cost on the kernel.org site of just getting slammed with all these requests, whoever's sending them or whatever is powering that has to, you know, crush that compute at that same scale too. So this is a way of deterring. But what they're finding is that because like you called out, like this is what the largest and oldest Git record, it represents the purest, most untainted by AI source data for any kind of model hunger, your model training system. Think of everything we just talked about before in that massive ecosystem, propped up by all of these other tools and investments of everyone trying to own and create their models and distilled stuff. They turned to things like kernel.org because it's completely untainted from an AI perspective and its core references. So they get slammed with all these requests. But so it's really valuable. So even though they're using a new bus and it they're putting up these really hard proof of work challenges, LLMs are still sticking around and crunching the numbers and paying the electricity costs because it's that valuable to them.

18:29And so what they did is they ramped up the difficulty over time to eventually kind of get the numbers to fall down. So it's like a really fast and example of it because this where we're going to be where your material is so valuable that you get attacked and then you have to have this platform. And then you tweak it by just seeing how valuable your content is against the computation cost of protecting it. And that's what keeps you safe. I think there's actually a whole there's probably even like a number to see there. I want to see that represented on a chart. Yeah. And you know, I think it's it's important to remember that while while there's not a lot of incentive for agents to operate efficiently, they still they still have an innate desire to be more efficient. You know, if an agent can solve the same problem using fewer tokens, generally speaking, it will want to take that pathway, which is kind of what I was trying to get at with my point on like, well, if the issue is that serving it over over this Git repo is too much burden to handle. Maybe there's a more efficient pathway that you can send all the agents down so that when

19:31they come for you for that info, they can access it more efficiently from you. But there was an analogy in this that really stuck with me in it. They were related to background radiation. It's like, it's just this problem that just like burns you like constantly now that you have to like apply sunscreen for. It's true. But you know, one thing that we've learned from running content is that every agent out there is sort of a different and it's ability to go out and find information on the web about something. So, you know, for example, some companies like Google have decades spent building web crawlers and they sort of seem to be ahead of the game in terms of their ability to research the internet through their AI models like Gemini, for example. But other models really aren't performing anywhere near that well. They're actually performing really poorly and they need a lot of handholding. Like your website almost has to hold their hands through everything that you have for them so that they don't get lost along the way. And once you have that, it works.

20:32You know, it works very well sometimes. So I think, you know, it's, it's, we all have to be thinking about how do we, if you're, if you're posting stuff out to the web or you're, you have a genetic systems that are interacting with your platform, which is something that we've been doing more and more at the near B. You really do have to be conscious of making sure that you're passing information in a way that's efficient for agents. A new hires first mistake used to be doing something wrong with Git. Now, it's an AI slot pull request merged straight into main from an untamed agent. But the accountability should be the same. You should own the code, the U ship, whether you wrote it or generated it. As AI writes more of your code, it's more important than ever to have strong checks and balances in place. Without them, security risks and spec mismatches slip straight into production. Linear B provides policy driven AI code review that catches risks and enforces your standards before human review even begins. Gov. Your AI workflows without slowing your team down.

21:33Learn more at linear B dot IO. All right. Let's move on to the latest article from Kent Beck titled reject change sometimes. Kent Beck, he uses this analogy that I'm going to want you to unpack for me a little bit. Andrew, but it's Shannon's demon, which is a strategy of continually rebalance between a safe asset and a volatile bet to explain why constant moderate rebalancing beats both an approach of pure caution or an approach to pure risk taking over time. Even if there's identical odds for each round, each individual round. So he sort of maps out this idea of in product development of like an extract mode where you might be looking to protect revenue and, you know, take small, you know, more growth bets that keep changes reversible and just avoid the big irreversible swings versus something that you might be in more of an explore mode where you need to look for things that are had big upsides and you're willing to invest enough to go all in on the topic.

22:38You know, sort of chase some sort of much larger or payout. And then there's also this sort of like expand mode where you might be looking to scale growth and add continuous value to some sort of engineering or or operations investment. And you know, the sort of the practical takeaway that he has from this article is that, you know, there's no one size fits all for the right amount of risk level that you should take at a management level. You should always be thinking about, is this something that I'm willing to take a big bet on and be willing to take a loss if it goes the wrong way versus something that I need to have more predictable returns. So I know there's a metaphor for some boxes. Andrew, I don't know if you want to explain that one. That's a good unpack of the strategy here. I guess maybe to take a step back about, you know, Shannon's demon. What does that even mean? You know, a demon is really just a philosophical entity. You could think of it as like something that either optimizes or is an exception to a rule or a paradox.

23:38This is something that we used to describe stuff a very famous one is Maxwell's demon. This is Shannon's demon. Knits the idea that it can optimize between those three decision choices that you just made Ben between exploring and exploiting or expanding. And all of those are choices that are made based on this on them on this like what happened this step before. Did you lose everything when you bet it all? Did you save a little bit by investing in yourself or did you meet somewhere in the middle? This is something that becomes like a state chain, right? So why write about this and why explain this in a concept the way that that Kent Beck is expressing here? Really what he's trying to do is express that for engineering leaders, the most opportunistic thing to do at this moment might be dependent on what happened before and what is the state of the environment. So right now, think of all of the things that, you know, you and I have covered Ben in this conversation about the economics around owning your inference and your models have been changing. A lot of that is going to come to a reckoning.

24:40So that might represent itself as an opportunity to explore in the world of Shannon's demon. You need to figure out how you're going to hold on to your inference cost or keep those stable in the future or to prevent instability from foundation. Right? Or maybe in this case, your opportunity is to exploit. You want to use the resources that are available and subsidized to you now to invest in yourself in the future. This is like how companies grew really big during the zero interest period, right? So these are like macro strategies that CTOs used to navigate themselves in an environment and Kent is just giving you a really short and sweet metaphor for how to make your bets based on if you lost your bet in the last bet and what the next one might look like. Really cool little puzzle. I recommend you go check it out. It's a nice short read. Yeah, awesome. All right. Let's close out with one about engineering culture and about how good culture is the biggest productivity hack. Not AI and boy, I'd love to read that sentence. But this is a pinion piece is really just arguing that you know, a strong engineering culture is the biggest lever for productivity.

25:48And again, getting to this narrative of AI as an amplifier, you know, and AI is going to amplify whatever culture that you have within your organization. The good and the bad. So it isn't by default create gains. It can actually do the reverse. And there's some really great, you know, stark warnings about executives out there who might be saying things like we don't need as many people because of AI. You know, that's those sorts of things really destroy psychological safety and morale. And that, you know, there's lots of companies out there that are claiming that they've got some sort of 10x productivity. Maybe it's even like one of your competitors that are out there doing that. But it's often, you know, let's be real often just vendor marketing. That can create a culture where leaders begin to blame teams for their inability to really leverage AI when it may, when it may be more systemic organizational challenges that need to be solved. You know, it's a great, you know, reference to Conway's law, which is where an or exit put mayors, its communication structure.

26:51So, you know, if you have bad collaboration and unclear priorities, you're going to get bad results regardless of whether or not your team is using tons of AI. But then on the flip side, good culture actually will compound positively. So it's not only makes you better, but the value of it gets exponentially better. At the end of the day, this author is really arguing that maybe we do still need more engineers rather than less like not even just flat engineers, but we may actually need more of them. Rather than fewer in the AI era because skilled people plus strong culture compounds productivity and a speed to market advantage of having engineers that are talented that are AI enabled and are part of a productive and healthy culture is something that can create a lot of growth that you would want to have more people contribute to. So, yeah, Angie, what did you take away from this article? Yes, big plus one to AI being an amplifier of whatever's going on in your org, the good and the bad kind of rears itself as a strange cousin of Conway's law where the shape of your org and how AI enabled people are in their data across it really is reflected in the shape of your agents and what they can get done.

28:05So it's really obviously a smart baseline to keep in mind is that you need to have these basics covered and the whole time I was reading this. So really there were so many echoes in here of our software factory debate that we just had. It was our episode that was we had on Tuesday and we hosted it last Thursday as well with Dex Horthy and a loked aside from human layer and work and we had Dan there as well. And they all echoed exactly what you just said that you know we need more engineers we need these folks who are understanding the shape of the STLC and where we'll go we need more we need more engineers actually in the in the seats creating the places the environment and the opportunities and preventing risk. There's so many things that have to get worked on and it's not just a matter of handing it all over to an agent. And I think all of them there even though they were all really invested in seeing so much around the software factory strategy still really firmly believes that. And so there's a lot of messaging and enablement that still has to happen within works and and if you're learning how to maybe operationalize that and get everybody on board like the AI enablement like train with them like your organization there's a lot of strategies they got unpacked in that episode.

29:16So really good one. Yeah and the final thing I'll just say on it all is that you know really the important thing is just to know where your organization stands and what you need to do to make the next step towards improvement. You know if you don't have visibility into where AI is being used and and who's seeing higher outputs from it and who's bearing the burden of higher code. So you're really not setting yourself up for success because on one hand you know while the the active writing code has gotten cheaper than ever it's been commoditized at this point. The active understanding whether or not code is a good contribution and as a part of an overall architecture that is sustainable and works long term. And that solves the needs of the users those are all more valuable than ever. So while code is getting slung around everywhere we're still putting a lot of pressure on the humans that are involved in the process. So yeah go go check out the software factory debate we cover a lot about how to get visibility into your SDLC as it becomes more agentic and we're also we've got some more content that we'll be announcing real soon.

30:22So stay tuned on that about you know we're going to be looking into engineering health and to developer burnout and you know how the AI transformation is impacting all of this stuff. Because it's really been bubbling to the surface a lot lately and then we think it's important enough to cover. So stay tuned for that little preview there for all of our viewers who stuck around till the end. So, so what are your agents up to this week? Oh well we've been dealing with all the new model releases that I've been coming out we've been putting them through the ringer with our back pocket evals seeing what they're capable of. So far Fable 5.1 doesn't complain a lot really nice tool to use. However I will say it was pretty bummed out just earlier this week on Thursday major outage across all of the providers in the morning so my factory came to a halt and it was a good opportunity to practice what our our fire drill is for when this thing happens and how to recover from it but once again in outage just knocking things aside and just reminded me again I was like wow this is why I need to have my own little models.

31:26You just had a whole bunch of agents like fleeing the internet on to your like local machines that was happening well no they didn't they didn't fall back into any kind of local mode I don't really trust them enough to do that on the wind machine I have but it will be nice I want to pick up one of them you know we saw at the AMD conference. Of the Gorgon that they're unveiling later this year this huge consumer AI box and I saw in action at the conference it's really impressive so between that are the jets and today we talked about Nvidia I don't know but I gotta get one of them soon. I've been more focused on Asian fodder so you know we do a version of spectraven development for a lot of our work and you know when you're building new big things it really is like I'm learning very valuable to spend. First brain time so to speak or human brain effort on like really thinking through problems a lot when you're in the early stages because the more the more you understand all of the criteria of what you need to solve. The faster things go once you hand it off to to AI we've been working on some big content factory we're building a you know not just a software factory but also a content factory over here.

32:36At linear B and it's been really fun to just sort of like think about these primitives and like really get things nailed down to get really high quality AI outputs you know because to your point with all these new models it's like. If you're working in a really high quality way you can get some some pretty awesome outputs for sure alright well thank you to our listeners for sticking around with us till the very end this is the Friday deploy brought to you by linear B. If you're listening to us and you think that we're delivering some great content for you because why else would you stuck around so. You know what you should do once you're done here is head over to substag or head over to LinkedIn to our newsletter for dev into ruptured and reach out and engage with the content over there we love to to hear from our audience we love to chat about what's impacting your teams and yeah thanks for sticking around and for joining us this week and we'll see you next time. And if you're watching this on YouTube let us know whose tux shirt you liked more mine's better yes mine's better mine's just just tux we'll see you next time you'll.