Have you ever tried to debug an issue in production while switching between your error tracker, log viewer, and APM tool? Maybe you know how painful it is when you can see the error but can't find the relevant logs, or when you see slow performance but it can't connect to the root cause. Scout monitoring is all in one platform, built to make your life easier. With log management, performance tracking, and error monitoring coming soon, all of your data will be together at last. You'll go from bug to fix in one place. The result? Less context switching, faster debugging, and monitoring that actually helps
you ship code instead of distracting from it. No DevOps team? No problem. Scout works with your existing stack and gives you insights in five minutes, not hours. From bug to fix in one place at ScoutEPM.com.
Hello, and welcome to another episode of the Bike Shed, the weekly podcast from your friends at Thoughtbot about developing great software. I'm Ajis later. And I'm Sally Hall, and together we're here to share a bit of what we've learned along the way. So, Sally, what's new in your world? Tis the season of end of semester holiday performances at my children's school, which is better than it was, you know, 10 years ago, because there were seniors in high school and we went to the holiday concert the other day. And I'm just always amazed to watch people be really, really good at something
I can't do. It was really impressive to see all these, you know, their children, but they're really talented and worked really hard. And the chorus concert was amazing. I was like, I don't know where I can get this talent. It wasn't me. It's just so nice to just sit back and watch somebody do
something that feels like magic, you know? Yeah, yeah. I once had someone ask me now that, you know, how technology works, is there anything that feels like magic to you anymore? And I'm like, yeah, people that can sing, people that can, there's still so many amazing magic things out there in the world, just because I know how computers work. Yeah, it doesn't take away the magic in the world. Most art feels like magic. Yeah. So it's no one in your world.
Oh, nothing so grand and hopeful. I have been getting to work a little bit with administration lately. That's the thought about gem for making admin pages that I had used kind of cursorily over the years, you know, to make admin pages, but never had to do anything really special or fancy with it. And I've been kind of digging into not only the new features now that it's gone 1.0, but also just the bells and whistles and the customization that you can do just out of the box has been really great. There are so many little off ramps to doing something completely custom that feel really easy to do. You just sort of generate, you get a couple of stock field
partials, and then you can make those field partials do whatever you need them to do. And you can customize it from the edit view, the show view, the index view, all of these different things that can add a lot of power to workflows for the people who are on the other side of that admin wall of a product, right? They don't often get the attention in the work timeline and the compound board that they really deserve, right? Because they're a vital functioning part of the product as well. So yeah, I just love the ease that Mr. brings to all that.
Yeah, I did a little bit that during Ruffle Palooza too. And it was, it's always nice that moment when you're like, oh man, I had to do something custom. And then you realize that it's not actually the type that you expected it to be. Yeah, yeah, especially in the Rails world, that's often you need to do something counter to the gem or rails. It's going to be hard. You're not going to
have a good time, but you can do it, not so much in this case. Yeah, it's one of the gems that I think was built with the intention of being customizable and accessible, which is really nice. I feel like we do both ends of that spectrum. Some things we make where we have opinions and those opinions are built into this tool. And if you want to do something different, then these opinions, this isn't the tool for you, which is also useful. I mean, not what I've already said, standard RBE, I find that so
helpful because it's like generally a big group of those opinions. And then I don't have to customize it. Yeah. But it's also delightful when it's like, here's our opinions. And also, you can do whatever you want to do with this. So test double-adjusting circles, the ones behind standard RBE, did a session
at a RailsConf a few years ago where they held a live vote for the way that the rules for standard Rails should go. So they started with a blank slate of looking at all of the different Rubocup rules for like Rubocup Rails or whatever that gem is, and gave everyone in the room like paddles that were like red side and green side. And we all voted. And every time we voted in a way that Justin didn't like, he played an ad for test double from his computer. So that was a lot of fun.
Yeah. That does have fun. I like that live, IRL crowdsourcing. It was basically the eventual GitHub issues and pull requests and all the comments and everything, but just in one room for half an hour. Yeah. Very efficient. Sometimes the fastest way
to do something is to be humans together. Oh, that does sort of segue into the topic that we are going to talk about today because some people might call yourself and myself AI skeptics. Is that a label that you use? Yeah, I think some people maybe would also call us haters. But I feel more comfortable with skeptics. I think I can see skeptic clearly in my rear
view mirror. I think I'm all the way all the way over. And yeah, it makes things difficult to navigate sometimes in this world that we're in, doesn't it? Yeah. Yeah. And it's hard to not just be
knee-dirk angry in response to pro AI initiatives or conversations. And I tried really hard to go, okay, these are people I respect with intelligent, researched opinions. And so I've been working hard on trying to be like, okay, what can I say is good about AI? So I won't start there. Okay. It'll be quick.
I think maybe we might need to distinguish what exactly we mean when we're talking about AI. I think we're both thinking of that in the vein of LLMs in this kind of large language model world that we live in at the moment, not necessarily other like machine learning techniques. Right. Yeah. Yeah. I think in general machine learning is morally neutral, I'd say. In my opinion,
but I feel less that way about AI. But right. I was going to start with things I like about good things. We're going to. Yeah. So I have used some AI things a few times in my work.
I find it most helpful for a subtle nuance of a language I'm not super familiar with is throwing me off and I cannot figure it out. And I haven't actually used like Claude or any of the other like tools you can have, you know, in your IDE with access to your whole code base. But I've used chat GPT and I've used honestly just the Google search AI summary. And sometimes those are really helpful. So I can see the value there. I think, you know, there are things that I probably would
have gotten to on my own eventually, but they did save me time. Yeah. After some arguing. I've had a pretty similar experience about I don't know if it's actually 50 50, but it feels like that. Then we go in circles and waste an hour or they get me what I need right away and I can just get back to making things work. I will say when I try to use it for personal things, it has been a lot
less successful. Like I tried to get it to help me figure out how to cut up and sew a piece of fabric back together to make it a specific size. I think I've mentioned this to you. And it did a terrible job. It just like ignored the laws of physics and then got kind of saucy with me when I was like, well, but the fabric is only 44 inches wide. As I mentioned, it's like, well, I didn't know we had
the constraint. For my last email, I just kick out, use a ruler and paper and figure this out. Yeah. I think that might tie into one of my points later on. So we'll keep that in mind.
So those are, those are the main things that I can say good about AI. What you got? Anything there? Yeah. Yeah. I mean, I think like you were saying, machine learning, LLMs, any of these different
things, robotics, they're all neutral statements of fact, right? It's the execution of them that might be problems with. But I have found that they're very good for go figure language things. They are able to parse and manipulate and churn through linguistic tasks, things that don't require facts or mental models of the world, anything like that. Just sort of like, here, I'm going to give you a giant amount of text. And I would like you to use that as the basis for answering some questions.
So far, I found that they're pretty good at doing that. And that kind of makes sense, right? That's how they're trained. That's the axis upon which the model is going to kind of view its job. If we want to add the poo-moroth-vise it already. And yeah, so things like that, that's like
rag retrieval, augmented generation, and different things like that, summarizing and things that are manipulating one corpus of text into some other form. I find that they're really fantastic for that, helping with voice, that kind of thing. What that isn't is coding. I know that there are a lot of people that have had a lot of success with that and don't want to act as though I don't understand that that that is possible, just because I maybe haven't had those experiences or haven't like sought them out or anything like that. But that's not generally the area that I will reach for these tools, if I do.
Yeah. I feel like the buzz though is that that is where these tools are really helpful. You know, that's the industry, the internet, the universe wants us to believe that AI tools are really, really great for writing code and for making developers more productive. Maybe a little bit of foreshadowing. I think it's six billionaires who sit at the top of this industry that want us to think that. But yeah, yeah, yeah, there we go. Yeah. The main thing I found,
and I, again, I don't use these tools to generate code myself, but I've worked on teams where other, many, many other people rely on them heavily. And I think Claude is pretty good at being wrong really fast. Yeah, it's a computer, right? Yeah. Yeah. And I think also, like,
I try to develop the best software I can. I try to write code that I hope is always or almost always above average, has some thought in it is really the best solution I can come up for, for the situation given the constraints. From what I understand about how AI and large language models work is that they are going to, it's a lot of using probability to predict names. Yeah. It feels like it would be nearly impossible to get something above average generated by AI because that's sort of the goal is like, what's the average way to answer this? What's the average way to approach this? Yeah.
And I, I don't like that. Yeah, I think that's something that even, I can't remember if it was anthropic or open AI, published a couple of blog posts that, because they understand this technology, and they maybe despite their best practice for marketing, they wrote about like, yes, technology will always have hallucinations because that is just how this works. Right. And kind of saying things along that line of, this isn't AGI. And so like the people building these things, maybe not the ones talking about it publicly know that that is a limitation just because of how they work, right? Like you're saying, it's probably a little bit. But in person, you said AGI,
what does that say for AGI is artificial general intelligence. Okay. So it is a truly thinking, what we would call a conscious artificial mind. Right. Yeah. Which the tools that we have right
now are generally not that. Yeah. And that's kind of where I, let's go back to what you're talking. I don't want to get ahead of myself here, because yeah, I want to kind of add on to what you were saying that as we're trying to make the best code that we can for the moment for the need, I have recently had to think a lot about like, what is the best code? What do I think is going to be better code? And I always come back to
like readability and extensibility. Does it do the thing that it needs to do? And is it not going to add weeks to someone's time to change it later on? And I feel like every time that I've encountered or I have generated some sections of code, it is not incomprehensible. Or it is, but it takes so much time to understand it like a human, which I am. I've seen some teams that will generate a
section of code, not be able to understand it themselves, the person who was orchestrating the generation. And so the solution to extending it is to just throw it back into the LLM. And that's terrifying, you know, in a little T work terrifying kind of way. If the system is getting like farther and farther out of your understanding and control, then that seems like it opens up a lot of risk, a lot of danger. Yeah, it's like a faster way to create those bits in the app that nobody wants to touch because only the person who wrote it understands it. But the person who wrote it isn't a person,
it's a model. Right. And I, as much as possible, try to avoid those avoid creating those. And when I encounter them, if possible, I like to try to make them less terrifying, less unapproachable. And so it feels like we're moving in the opposite direction of that by committing code into projects that we don't fully understand. Or maybe we understand it
enough to go, okay, this is what I wanted it to do, but not enough to go, how do I change it? How do I add to it? How do I modify it? Yeah, because it's trained on all of the code that is publicly available. And just by definition, half of that code is below average. Or even more so, like
stack overflow is full of people asking questions about problems they have. Right. It is not primarily a place for people to just post snippets of excellent code that they like for no reason. That's true. Right. And I think it's a super valuable resource and there's nothing wrong with that.
But by like inherently, that source is going to have a lot of flog code in it. Right. Right. Because the question is, hey, this doesn't work. Here's a bunch of code.
Yeah. And you know, I imagine that these models are smart enough to differentiate the question from the answer, right? At least in that context. But it's not as simple as that. I think one of the sort of comments that always comes up in this train of thought is, well, there are always developers that are committing this kind of code. Like, how is this?
How is this any different than just someone on the team that isn't trying very hard and is just committing a whole bunch of code that maybe the rest of the team doesn't appreciate? Yeah. And I suppose the counter is to like, is that what you want to aim for? Is that your goal? Do you want to intentionally hire a bunch of those people into your team?
Right. Who by definition aren't going to learn like a early career dev that joins your team is going to get better as they get more experience and they progress and they learn the practices that your team uses, they learn more idiomatic code as they're involved in the community, different things like that. But the language model is not going to do that. There might be some change as your code base changes and it scans at that sort of thing, but it's not going to level up in the way that a junior is. I think there's also an amount of like automatic trust that people give to AI tools.
And that if you have a coworker that someone on your team who doesn't really care and has a history of writing crappy code, you're going to approach pull requests and code reviews with a slightly more critical eye or with just that context, right? I know that there's a likelihood that this code has some mistakes in it or has repeated some patterns that I'm familiar with. I think in general, people don't have that approach with AI generated code. I think they have the opposite where their first assumption is that the code is probably pretty good or does what it's supposed to do. So I think that's a big difference is that people are not blindly merging in code written by a coworker who's not trying very hard, even when they don't understand it. But we are doing that
with AI coding in many cases. That's true. And the usual sort of thought bot people first approach to having someone who's not trying very hard on the team of being generous with our time and expertise and helping them to level up and usually finding out that that is the easier thing to do in the first place. And all of these different things is not going to change the AI generated code. Yeah. And assuming that the person who wrote that not great code was doing their best and has
good intentions and wants to write better code, and those are not necessarily some turns I could make about an LLM. Because it doesn't want anything. It doesn't think either. I think that was a... We're going to do it just as humans anyway. That is the way that our brain has evolved
to the promorpies and put things in those terms. But there have been some coinages of the AI industry, if you will, that have given it more of that sheen. There are reasoning models. They type out their responses to you in the UI. There are all of these different things that make it seem as though it is thinking and it has feelings and different stuff like that when it really is just incredibly sophisticated, but fancy predictive text. Yeah, I think they're like natural language I'll put
in the conversational tones and the fact that we give them names makes it a lot easier to forget that they're not people and they're not thinking. They're not participating in the back and forth in the same way that we are. Because there have been plenty of times where I've walked myself and model in a circle where the model has commented on the person that first wrote this code as having made some mistakes. And I'm like, well, that was you. I mean, I'm guilty of saying that too, but I usually have a cheeky self-dever getting away. Sure. Sure. And not within 20 minutes too.
You can remember that you did that. Most of the time I can. Yeah. That's it. That is a good point.
Most of the time I can as well. Yeah. So I feel like all of this that we've talked about so far is the limitations on quality. What can it do? What does it do? Well, what does it not do well?
And with how we use it, which I think one argument response of that could be like, well, it's new. It will get better and you will get better at using it. So just keep trying. And that, I think doesn't feel great to me for a lot of reasons. I think we do a lot of just completely ignoring or sweeping to the side, the downsides that use in these tools at large scale.
And so I'm hesitant to put more time and effort into getting a tool to do what I want it to do when I know that it is also causing harm at the same time. That's the part that I usually check out of a conversation is when you bring up like, yes, we can talk about how well it does on coding, but like, what is the cost of this as a society? Right. And that kind of, I think sort of foundational consideration isn't usually taken into place. We jump right ahead to assuming that these things are inevitable and that they are good. Yeah. And that the side effects of them are inevitable too.
Right. Like towns running out of water might have multiple contributing factors, but the giant AI data center that uses a significant chunk of the town's entire water supply needs to be taken into consideration. That's a real thing that happens. Yeah. And the small towns of usually historically marginalized folks, right? These are these communities that have
in the states been redlined and all of these different things. They're the ones that are getting these data centers next door. Their electricity prices are going up. So they are in a way subsidizing this data center or they are feeling the cost and the burden of it greater than other communities. And like at best, that was an accident and people are getting hurt. And at worst, that is the reason
they are doing them where they are doing them. Yeah. I think those areas, the land is cheap. The electricity is cheap. It's, you know, they just need more or less a throwaway location on earth where they can put stuff. Right. And that's going to be in a place where people with fewer resources live.
And then they go there and they take away even more of the resources that are already limited. Yeah. And why are they areas that generally have fewer resources? The same engine of capitalism has created them in the first place and continues on. I feel like Sandy Metz's talk at, was it really calm? A few years ago about redlining in Durham, which is also my hometown, I think gives a lot of
information on this if any of you are wondering sort of how what we mean by redlining and historically under-resourced people and how systems can perpetuate cycles of poverty. Go check out that talk because it is really, really informative and moving. That talk was, I was lucky enough to be in the room when she gave that talk. And the title of the talk, I think, is something like, how lucky are we? And the blurb and everything is like, how lucky are we to work in a programming language that we love? We just love Ruby and it's so great. And so everyone walked into the room, not ready to get
blindsided by the like, how lucky am I? All of this privilege, all of these things that the system has given to me and kind of going through that story of like created, intentionally created inequality and just looking around the room and like seeing the people that are impacted by that, right, who are maybe seeing these sort of things for the first time. And then other folks that are upset because they wanted to hear about programming. It's just a lot of different feelings going on in the room. It was an incredible experience. That is an amazing talk. You should definitely watch it.
Yeah. Yeah. And I think it's very relevant to the way these handful of billion or corporations are deciding where to plop down their data centers. They're not putting them where they live. They're not putting them where people in the tech industry live and work. We are lucky. I don't
imagine anybody will ever plop a data center down in my town because it would be inefficient, not cheap enough. Right. Right. I pay incredibly high rents comparatively living here in Edinburgh. And so I know that by the square foot, this is not where you put a data center. And yes, I am very privileged
and very grateful and very lucky to be working in an industry that allows me to afford that. Yeah. That's another thing that's a little lopsided about this is that all of the people involved in creating and most of the people involved in regularly using these tools are not at all the people who are affected by the placement of these data centers. Like it's an entirely separate group of people and makes it very easy for all of us to just not think about it and pretend it doesn't matter. I have the wild, wild opinion that people actually do matter. But
that reminds me of the one, it's one of those like reply images that you see on the internet from time to time. And it's just white with black text on it says, I don't know how to tell you that you should care about other people. Yeah. Yeah. Yeah. I mean, though, like these billionaires,
they need more money though. That's the other thing that frustrates me is that this is not a need that we've been trying to fill for years. We have not spent long times going, gosh, it is so hard to write code quickly and efficiently. And this is just not a great way to do this. And I really wish I had a better way to create software because this isn't making people rich faster. I was thinking, it's like, you know, I think we have been thinking that. And then that
last sentence is like, nope, no, that's why that's why it's different. I can't think of very many actual human problems that could be solved by people writing code faster. That's not going to make my knee hurt less. It's not going to make my in-laws have more access to their community. It's not, you know, that's not the solution. But what it is doing is it's making, you know,
there's handful of billionaires, Richard. And it's very frustrating that people have just stopped saying just because you have a hammer doesn't mean everything's a nail. And AI feels like this shiny new hammer that we're all desperately looking for nails to hit with. I saw an article. I don't want to say what publications front because I can't remember. I will put it in
the show notes. The headline was, humanity's running out of ideas or like humanity's been running out of ideas. Here's how AI is going to save us. And I did the look, look up and look around and be like, was anybody running out of ideas? This is the first I've heard of this. And it's exactly that. They're
the narrative is starting from this assumption or presumption that there is a problem that this solves because I haven't seen it yet. Yeah. And I feel like the idea that AI can have good ideas, that people can't think of new ideas anymore, but AI can, when everything AI comes up with is a amalgamation of things that people have already done. I don't see AI taking cognitive leaps and making innovative ideas happen when it's really just complex probability on data that it already has. And I think that's a little bit speaks to a lot of how I feel about this. But when you were saying
before that people sitting around or working or whatever, saying, you know, this is a really inefficient way to do this. And this is not a good way to code and stuff. And it's like, we have been thinking that and that has driven us to make new tools and make new things, right? And that kind of innovation and growth, I think is at danger or at risk of taking a hit. Like, we're not going to be innovating actually as fast because we will not be trying to solve problems of like the grunt work. And the alarm can just spit out code. And I think we'll probably
stagnate as an industry if we lean too hard into these things. And it's never going to be adopted 100%. It's never going to, et cetera, et cetera, et cetera. But it's going to have an effect. Yeah. That makes me think of sort of the joy of crafting things versus the having a thing that
fulfills a purpose. I know that I'm slightly unusual in this. Like, I am wearing socks that I did myself and have meant it several times. That's not a normal thing to do in 2025, but it's a thing that makes me happy. And I think as we get better at making things faster with less human input, they inevitably get worse. The handcrafted furniture that your great-great-grandfather
made out of a tree, he cut down himself is probably still around. And somebody can still sit on it, and it works really great. The Ikea chair that I bought last year is dead. That's true. And even the higher end, well, you know, mass-manning back-grid stuff doesn't last that long. And I don't think we need to move towards more of that. I don't know. This is now I just feel
like a cranky old woman being like, get off my lawn. But you know, things antique things that are still around are valuable for a reason. And it's because they don't make them like that anymore. And I feel like AI is leading us to a spot where software is going to be, they don't make them like that anymore. And that feels gross and bad.
Yeah. I think this is a technical term. So I think I'm okay saying this on the podcast. What you're describing is inshitification. Yes, inshitification. Yes. Yes. Further raising
the inshitification of all software. The wonderfully descriptive term coming by Cori Doctor O to describe the modern world's march towards mediocrity. Yeah. And I think the other side of that, though, is that I get, I admit my own socks, not because they're better than the ones I can buy at the store, but because I enjoy it, right? I enjoy the process. I enjoy learning and getting better at it. I enjoy the satisfaction of completing
it. I enjoy all of those things with code, too. I don't think AI is capable of enjoying it. And we talk about developer experience and one of Bob Bob's values is fulfillment. And those are not things that AI can experience. And those are not things that I have found AI to increase for me.
The satisfaction of fulfillment of solving a problem drops significantly for me if I just asked a computer to do it for me. Yeah. Journey before destination, right? Yeah. And I know that that's not how capitalism works, but I also don't care.
That is, I think, ultimately where my kind of wall comes up with these kind of agent coding tools. And I know that there are other folks like me. I know there are other folks that have the opposite experience. And their experience is absolutely valid as well. If they get fulfillment from using these tools and everything like that, I'm not saying that they're bad or wrong. I'm
speaking for myself that and you were speaking for yourself that this way of working is not fulfilling, right? Ultimately, it might meet other, there might be other trade-offs that it meets or anything like that. But at the end of the day, I will not have enjoyed the day of work if I have been solving problems by inputting the problems into a plagiarism machine and getting an answer out. And that is all right. That might mean that I, there will come a time when I will not work in this industry, because I will not be fulfilled with it. And that's okay too, I guess. I would miss it for sure.
Hey, when I see you, I believe. I think that's the thing that I'm mostly afraid of is because not because I'm not going to be able to learn them or because they're not going to be good enough or different things like that. It's just it will not be fulfilling and I will need to look for my next thing. Can I recommend knitting? I've been wanting to get into a little bit of embroidery, but that sort of besides the point, that is an art form. And I think one of the things that gets kind of touched
on a lot with generative AI, large language models or the generative imagery and video techniques that are out there. So Nick Kleg, who was a former deputy prime minister and an executive at META, said that basically having to follow the law would mean that the AI industry couldn't exist. Speaking of taking works that had not been published in a way or under a license that would have allowed or appreciated the kind of training that their models have done, we know instances where these companies have literally just torrented books because they knew asking the publishers and authors would cost too much or take too much time. And I think if your industry requires you to be above the law and above democracy, then maybe your industry shouldn't exist. Yeah, I don't have a lot to say that Yeah, it is tempting to compare it to various human rights violations and other sorts of things
that have been justified by, well, we can't afford to do this if we don't do it in this way that hurts people. Well, but then should we? Right. It's again presupposing that this is good and right and inevitable when in fact, it's being made by people who are choosing to make it. I'm still just not convinced that the benefits of using these tools and building these data centers and stealing this work outweighs the harms. And I think that that is an important question to ask ahead of time,
rather than later going, Oh, whoopsie, how do we fix that? Yeah, absolutely. I am always willing to be wrong. And I continue to educate myself in this realm. But yeah, I have yet to be convinced that this is a thing that is good. I would also love to be proven wrong. I would
love to wake up tomorrow and have AI not hurt people and be good at helping me do my job faster. But it's not doing that yet. Yeah, I think that was that was an interesting conversation that I had some of my sort of assumptions and beliefs challenged by someone in a conversation within the robots lack that I really appreciated, by the way, it was really well thought out, kind of feel self-agreement in a way that I was talking about how these models were trained kind of unethically, that they have gone around our societal agreement of how people should act in order to money more.
And the question was raised, well, is a model always going to be unethical? Like, is there a moment after which it has kind of earned absolution or something like that? Can a model that was unethically trained to be redeemed? And that question was put to me in a much better way than I've summarized it here because it gave me a lot to think about. But when you get to the core of that, it's like, no, it cannot be redeemed because it is not alive. It's a blood diamond. It is less
beautiful when you know how many people die getting it and its kinds. And a model that was trained unethically will always have been trained unethically and will always have extracted the value against someone's permission of their work. So, no, it can't be absolved. We need a different one that was done better. That makes me think of, I watch a lot of crime TV, right, long order, etc. And evidence
that has been tainted can't be untainted. Evidence that has something has happened to it to make it unusable doesn't get outweighed by the fact that like, oh, but it's really good. Right. Yeah. It just is. And it, what is it? For the poison tree, something like that, where, you know,
if it starts out bad and then you build on it bad, even if eventually you stop doing the bad to keep building it, it still has the foundation of that. And it goes back to what we were saying at the beginning that these technologies are not necessarily morally corrupt or anything. But if you keep building on the thing that was, instead of starting over and being better about it, then that's always a part of your history. It's always a part of that tool, right? A person can be absolved because they can learn, they can grow, they can change, they can understand why they have changed. But it's not even part of
its data set, really, of knowing how it was trained ethically or not. On that note, shall we wrap up? Yeah, let's run away as fast as we can. Sounds good.