The Standup with ThePrimeagen - Amazon fires middle management? Did AI kill game dev?I

Episode Date: May 8, 2025

ssh terminal.shop

So this is the stand-up episode two where we talk about all of the greatest things you see going on in tech. And to get us started, we are going to look at everybody's favorite person, I believe, hold on, let me pull this up, Andy Jassy, the CEO of Amazon, who has recently been saying a lot of things on the internet. And this actually does include the following. Before the pandemic, not everybody was in the office five days a week. every week. If you or your child were sick, if you had some sort of house emergency, if you were on the road seeing customers or partners, if you needed a day or two to finish coding in the more

isolated environment, people worked remotely. This was understood and will be moving forward as well. But before the pandemic, it was not given that folks could work remotely two days a week. And that will also be true moving forward. Our expectation is that people will be in the office outside of extenuating circumstances like those mentioned above. And if you already have remote work exception approved through your S-Team Leader, so one of those two things is what he's saying. So he's kind of building up this case that remote work is coming to an end. And he even said that we want to operate like the world's largest startup.

And then just a couple days ago, on Bloomberg, he said the following. If you are a culture that invents a lot and collaborates a lot, some incoherent rambling where he stumbled on some words, if you don't have people in the office together doing that intent invention, it's just meaningfully worse. You don't invent. You don't collaborate the same way. You don't connect with each other the same way. You don't learn the culture the same way.

And so, having people back in the office together more frequently, we felt very strongly we felt it would be better for customers and the business. End quote. Now, Casey, you've never actually had a real job before. And so since that is the case, how do you? you feel about this whole remote work and devs? Is it a requirement? Is it actually something that makes meaningful differences? Or is this just a fabrication of the man trying to hold us down with his real estate prices and trying to get these properties and these commercial properties

rented, as I've been told on Reddit? Well, you know, I will say I did have a brush with the real job because obviously when I worked at Rad, one of the things that we did do was we did a, there was a very large contract for Intel that was being done. And I got to kind of experience secondhand that amazing force of multi-level management. And it was pretty special. So I kind of have a vague idea of what goes on. But thankfully, it was sort of like, you know, what, you know, like sort of like on a whale watch. We kind of see it off the distance, you know, doing its little breach thing.

I don't have a lot to add to this. I feel like I wanted to mention one thing about that memo. and I feel like then you guys should probably have the majority of the reaction to this. The thing that struck me in that memo is it read almost like someone who was, you know, getting up to give like a speech at a funeral who had been like crying beforehand and was like just barely holding it together and just almost got the thing out and then just broke down anyway. And the reason I say that is because when I read it, there's this big long, like two paragraph sort of thing in there that's like, look, we're, you know,

We used to be really lean and, you know, we want to run like a big startup and we're going to invent and all these managers are everywhere and they're ruining everything. And we're going to get rid of that and we're to come back to the office. It's going to be awesome, right? And it sounded like he really believed it. Like I was wicked into it. I'm like, okay, man, like, you know what? You've got a vision. You're going for it.

That's cool. But then in like the very next paragraph, it was like, now there are some things like you're going to need to coordinate with your. team. I have no idea what that is. I assume it means a team that's way better than the A team. And obviously there are it's a tier list. They rank all their teams into a tier list. And they're all

X tier size. All your team. You got to coordinate with your S team. Obviously if you have a remote work exception, which was all capitalized, like that's the thing. So he was able to hold the like lean startup. We're not going to have managers together for like only a few

sentences. And then it went right back. to like all of these Amazon jargon structural like things and I'm just like okay you're you're still just going to have like a bunch of middle miniature it's like you're not really fooling me that was what I got out of the memo maybe maybe I'm over reading it but it was hilarious how quickly it went back into weird jargon stuff where it was just like okay but can a casey can a company even escaped at the size of Amazon right if you have 10,000 engineers like how can you not just have tons of mental management right because you're going to have at least How many, like, if you had for every 10 engineers, you had one manager, that'd be 1,000 managers.

So that means for every 10 managers, you might have a director, which would be 100 directors. For every 10 directors, you have a VP. You have 10 VPs. Does that mean you have to have, like, one senior VP for every 10 VPs for every, you know, like you could just see, like, that structure is almost inevitable to kind of form. Is it possible to avoid that? Well, I mean, I guess the question is, is it possible to avoid it and do what, right? I mean, obviously, we all agree that it's possible to avoid it in the sense that you could just not have it.

The question is, how effective is it? And I guess the one thing I would point out is we do know that you can be, let's say, 500 employees and not have any management. Because obviously, Valve is, like, massively dominant and makes infinity cash per year. They almost certainly have a massively higher per employee income for the company than does, say, an Amazon or something like that. And they don't have that structure, right? So we do know you could do it with 500. The question is, can you do it with 5,000? Maybe not.

Can you do it with 50,000? Maybe not, right? And so there may be a natural size at which if you don't go ultra-hiercy, there's really nothing you can do. Trash, you're the only one left of us that actually work at a company that fulfills this kind of size and requirement. Now, before I left Netflix, Netflix used to be known as this bad. away from all the Silicon Valley, which was you come there, you're a senior engineer,

you get ownership, you just go in and you get after it. And then 2020 happened, COVID happened. We must hire younger, more green people happen. We actually brought in a whole group of Amazon management specifically from Amazon. Then levels started happening. Process started happening. Is this just inevitable trash? Are you feeling the burn of middle management over there? I thought that was how trash got hired. How dare you? First of all, I want to start this with all opinions on my own. And if my, if anyone, when my company sees this, it does not reflect any of their views. I'm just going to protect myself here.

Okay? Smart, very smart. I was like, oh, no. They have at least a two-password policy. Like, my stomach got, like, very tight when you brought up when you started introducing me. I was like, uh, I mean, I'll say this. I've been around the block, and I've been a part of, like, some big company. like banking and stuff like that that have like way bigger engineering structures.

And in my experience, to be honest, like so I've been from big to startups. Netflix is kind of like in the middle. I still think they run fairly lean compared to like an Amazon, obviously like Facebook and all those other big thing companies. So like in that sense, like for me already seeing like the extreme spectrum of like huge middle management like in banking, you basically end up in meetings where no one even has the power to make a decision. and then you just basically keep passing that on.

And then you ultimately get to the end where whoever can make the decision will potentially make the decision in the complete opposite direction of all the other people having the meeting. And I've seen this time and time again about my previous companies. Is it inevitable?

I don't know. What I think is going to happen is Amazon is probably going to fly in it. I do like flat. The advantage of having a flatter structure is faster decisions. You don't have to go through all this bureaucracy bullshit. it, like, you as an IC actually have, like, you feel like you have control of what you're doing.

So I think this is what's going to happen at Amazon, but I absolutely think that they're just going to backtrack in, like, three months when everyone in every team just starts building things that no one even cares about, right? If you put, like, sometimes when you put engineers in charge, they usually build what engineers think are important and not necessarily what the customer wants. And I've seen this basically in every company, not even excluding Netflix, right, where I think this is important. I have a use case for this

and then you go three months and you're like, oh shit, we just put the wrong thing. So I think that's actually what's going to happen at Amazon. So I guess we'll see. I have a lot of other mean thoughts

about Amazon. Like what is the culture that they're referring to? I feel like my simple size is pretty big. I feel like everybody I know that either works at Amazon or has left Amazon

has completely hated it. And I actually try to like Google the culture doc before this and like it didn't even show up in like in the first results i don't even know what the culture is so i would leave i would love to know like what they would be i believe if i'm not mistaken the amazon culture is that of the judge dread culture of megacity one people go in meat comes out like i'm pretty sure that has always been the general goal of can you correct me though if i'm wrong here isn't amazon the company where you have like everyone's dog is at work

no that'd be more like that was like the old google days right that was the old, that was really just like old Silicon Valley days. I think those days have largely passed where everything is just like, hey, bro, we got our dogs at work and it's super cool. So they're not allowed to do that anymore? I don't know if they're allowed or not allowed to bring their dog to work at this point. I know there's a couple dogs at Netflix. There's a few there, but there wasn't very many. It was not really considered like a thing people did, generally speaking. Because in the early days of Amazon anyway, the dogs were everywhere. It was just like a dog fest there. Really? Is this like a well-known fact?

Oh, absolutely. I knew. I knew. tons of people at early Amazon. Trash, why do you hate dogs? This is our culture. It's Amazon's culture. We like dogs. We're not cat people. No, like dog poop in the elevator as a common thing is like early Amazon. No, like, I mean, I shouldn't say early early Amazon because there wasn't an elevator at very early. But I mean, you know, at the tier where they, you know, like early 2000 N or something like that, then. And so like I was off the return to office thing, I was like, when I saw that part of the memo, I'm like, gosh, I hope there isn't dog poop everywhere because that would be a good argument for staying home as well, because that just sounded awful, especially as a cat person. I'm like, I don't want dogs everywhere. Anyway, sorry, I didn't mean to derail that. I just, I think, Amazon, I think dog poop. I was unaware of the lore of Amazon. Now, TJ, you actually have been remote working for quite some time.

Now, I've been led to believe that remote work is actually not possible and produces zero engaged workers. Is this actually how it is? Well, I started streaming while I was working remote and then I quit my job. So, I mean, you're asking the wrong guy, dude. Don't do not put me as the poster child of remote workers here. Let's read, okay, it's good for your personal development. Let's say there's a way to phrase it. Career advancement was high.

I went from employed with a job my parents kind of understood to unemployed with something my parents definitely don't understand. Oh, my. I have a question, though, for trash. Trash, do you think currently the managers at Amazon are solving problems that customers, like, want? Because I liked your point that I think engineers do to, tend to like to solve the problems that interest them. But I guess, like, it's kind of compared to what? I'd be interested if you think, like, Amazon's current scenario is such that they're

currently solving the customer problems. I feel like in middle management, that objective kind of disappears and everyone kind of just cares about their name being visible because there's just so many layers to the cake that ultimately they just want to make sure that, you know, whoever's above them knows that they're doing their job. And I'm not saying this is what's happening at Amazon. I've just seen this happen personally. You're just saying it happens at Netflix is what you're trying to say.

Flip, you got to edit that out. You got to fully. What if I actually get fired the next week? I'm like, all right, I guess I'm just a full-time streamer. It would be because of the password. That's why.

It's not this episode. No, but seriously, like when there's just so many, like, layers of power, people just need to impress and just have their name there. I mean, that's also why you have project managers too.

It's not necessarily the actual management. Like, managers exist for these specific reasons. If you cut them out and it's just a manager and they kind of have both the roles, I don't know. I can't say for Amazon, but I would give it like a 90% chance that they're not really actually thinking about customer problems. Everyone, there's like selfishness to a degree, right?

So we'll see. I guess we'll find out when they fire all the managers and they make nobody happy. Well, the funny thing is they just said all they wanted to do is increase the ratio by 15% or something like that. It's not even like they're saying we're done with this as a role. We no longer hire middle managers. They're just like, we need less. And ever, you know, all of a sudden

Reddit loves middle management. I thought that was that was the strange favorite class. Like maybe they like billionaires less. But there's like so few of them, right? So it's like, I thought Reddit's most hated group was middle managers. They just suck the corporate soul. They're just useless. And then all the sudden, think of the core. Think of the middle

managers. Oh my goodness. They're just pulling up the ladder. Skill ladders being pulled up, filling it with just useless MBA. So it turns out it actually goes billionaires, then MBAs, then middle management. Right. It was the ordering of Reddit hate. Yeah. How do you even get to the 15%? They're like, Jimmy, give me a number. 15%. Let's go for it. Yeah. I'm sure it was highly logical trash. Okay, they went in there and they just said, yeah, it's just too many. Yeah. Mm-hmm. All right. One thing just like about remote work that I think is funny that like a lot of people, so I'll get canceled again for this one.

So that'll be fun. At least one episode, every time an episode I'll have to say something that Reddit will hate me for, which was that even in the announcement, they mentioned this thing called remote, you know, work exception, which I find funny because basically like they're saying,

if we care enough about you, if you're actually a good enough worker, we will bend over backwards to keep you here. So literally like getting called. back to office for a bunch of people is like just Amazon just saying we only mildly value your work. You're not good enough to get an exception. Even at Epic where they have literally just one campus where I work, there were like three people that still could, they had, they somehow got the approval, right?

And why is it? They're just so valuable to the company. So I just love sort of like imagining a bunch of Reddit people fuming that, you know, oh, Joe or Sally over there has the remote work exception. And it's not, it's like, yeah, dude, you haven't done anything for the company. Well, it might be more like a driver's license, right? It's like, okay, you have to go take a test and they put you in front of a computer and they see like how often you click on like Facebook or something. And that's how they know whether they're going to give you a remote work exception, right?

It's like they're, they know like, okay, this is a person who doesn't get anything done unless there's someone sitting very close to them who can see them screwing around. And those people, you don't get a remote work exception. They just have a series of social logins. That's all it is. It's like, now log in with Twitter. You're like, oh. Yeah. Dang, okay, I guess I log in Twitter.

They're like, sorry. You yap too much on the internet. You don't have to two-factor off to each of the social things on your work computer. That's how they know. They're like, get back in the office. Get back in. Well, Trash would ruin this test because he would know all of his passwords since they're just one password. And it would look like he logs into social.

me do all the time. I got at least three now. Oh, okay. Okay. Wow. Moving up in this world. Moving up. That's progress. Getting secure. Mm-hmm.

All right. Geez. I was not expecting that, but I am a pretty big fan of Reddit and just the about face. I think the thing that's most confusing about Reddit is this, there's this idea of this shadowy cabal that the reason why remote work is ending is because they want the offices to be filled by people who run Reef. And so there's like this shadowy cabal that's putting pressure on the commercial companies on behalf of REITs so they can have better dividend payouts. But Amazon owns its buildings.

It's too late. That's not how it works, buddy. Oh, no, no. So you don't understand. In the Reddit comments on this thread, they've explained it that the reason that they're doing this is because of corporate real estate. Okay. So we know this because they commented it on Reddit and they're a real estate expert. Yeah, you're a fool, Casey.

The thing is, is a building that Amazon owns sitting empty is more expensive and hurtful to the company than a building filled with staff and maintenance and having everybody in there and treats and all the stuff and micro kitchens and all those things. It's actually cheaper to run the building than to let it stay vacant. So it's behooving Amazon to really screw the people by getting the back in. That is some quality Reddit reasoning right there. That makes literally zero sense. Like they built the skyscrapers for their headquarters themselves, and they own them. They are not leasable buildings. They own them.

It's the re-companies. We call them Bezos's balls if you live here in Seattle. Have you never seen them? Bezos has three balls? Have you been, you didn't know about the dogs? Do you know about the doors, the desks made out of doors? You guys don't know any Amazon lore here. There's so much Amazon stuff.

Yeah, if you go to downtown Seattle, you can go to their, campus where they built, they built, not they had someone else built in lease, they built several giant 40 plus story skyscrapers. And as part of the agreement with the city, they had to create something that wasn't horrible. Like, you know, the city, obviously you do this kind of stuff. You're going to have to, you got to grease all these palms. So they built these three spherical, giant spherical like greenhouses. I don't know what, I don't know what you call them. and they put, maybe because their name is Amazon, I don't know, they put like a rainforest in there.

There's all these trees and botanists. There are full-time botanists who go around and maintain all these weird exotic flowers. You guys never heard of this? No, I don't know what you're talking about. I swear to God, you go look it up. I'm not making this up. All right. Well, Jeff, we'll do a deep dive on Jeff Bezos's balls next week.

But this week we got to stay focused. Okay, we've got to go on to a topic that's near and dear to your heart, Casey. Okay. This, right here. Okay. Do you see what I am seeing, Casey? I see something. Yes, that, my friends, is a high-end state-of-the-art helicopter for Peter Levels I-O video game that he's been making with Cursor.

And if you haven't seen it, it's pretty fantastic. There's actually a whole upgrade screen where you can purchase different types of vehicles. At the end of the day, there's a whole bunch of sponsors. And this has, in fact, landed Peter Levels to say that he is making 67,000. thousand dollars monthly recurring revenue of which he actually just updated going all the way up to 72,000 and even in that tweet more people signed on so i'm kind of seeing something here i'm seeing peter levels a novice game developer powered by ai has made more successful products than the average indie game dev casey does it hurt is game dev over are we cooked is this the end of the

expert? Like, what's happening here, Casey? No, this is total amateur hour, and let me explain why. So if you want to take a look at how the professionals do this, you can see what it really means to have some monthly income. So,

if, for example, you rewind the clock and go to Cloud Imperium Games, you will notice that right out of the gate, they got people to give them, was it $13 million, $30 million for Star Citizen to do Star Citizen.

And since then... A Star Citizen highly regarded game, by the way. Without ever actually shipping the game over the past 15 years, they have now... What? They have YouTube videos.

They have YouTube videos. And there's an alpha, much like Levels I-O has. There is like a thing... It doesn't really work, much like the Levels I-O-1. But it's there. you can go on, right? They've generated half a billion dollars in spend total. So, so no, like, basically what you're seeing is an amateur sell an, sell an incomplete buggy product, right?

And if you look at how professionals do that, they crush him. They're so far ahead of where he is. So they deliver later, they sell more bugs than they get more money is what you're trying to say. Way more money. And employ way more people. Like, it's, I mean, right? And it, so you can, you can see the difference. And have higher production values.

If you go look at a Star Citizen Alpha, the art is gorgeous, right? There's all this stuff in it. So, you know. So I think all we're seeing here is like, yep, I don't know that the AI was necessary. I mean, the thing that he showed kind of looks like you booted up Roblox or something. So I don't think you need AI for anything there. You could have just done this with anything. And so the AI part is not really the interesting part.

Although I guess I would point out the AI. did give him something, which is that probably if he hadn't have been talking about how it was made with AI, no one would have gone and looked at it, right? So the AI gives you, does give you something, and that something is the ability to attract attention because people

for some reason think it's interesting. So I'm just using Casey's logic here. So game devs, they can pull in enormous amounts of money pre-execution and be able to deliver buggy code. I mean, I'm personally just thinking trash dev, I think we found a calling for you.

Like if you get fired, it sounds like AAA gaming is right up your alley. Quadruplea. Yeah, dude, you can make a true, true quadruplea game experience. If there's anything about not delivering,

I'm your man. I got you. Yeah, if anything, levels, like, why did he even make the game? Right?

That seems kind of like an error. He could have just, he could have just had everybody pre-order and then walk away. That would have been so much smarter. What was he thinking? That's where the money's at because you don't actually have to execute, right? But, but TJ, do you think that it's actually just, so Casey kind of put a little pressure against it? Do you think it's actually, like, TJ, you can go with this one?

Do you think it's actually because he mentioned AI that caused all the ruckus? Or is it, or is there something more? I think it's part of that. I think people underestimate levels skill at telling a story. There's like a lot. And, you know, there's lots of people who have more followers than he does. I mean, he has a lot of followers. they're really engaged with him and everything.

But like part of it is just like AI, I think it gets some people in the door for sure. But like everybody posts an AI thing every day now. I don't really think, you know, you got to say like what's the competitive thing that he's doing differently. And I think people underestimate how good of a storyteller levels is. And they appreciate like his candor and like just the way that he tells the story, publishes what he's doing, where he's going with different things. And so I feel like people are really underestimating that. Like, Levels has a big audience in large part because he's a very good storyteller.

I'm not saying like as a, like, I'm not saying like he's lying. You know what I'm saying? I'm saying like he tells good stories. People enjoy listening to him and hearing what he has to say. So I think that that's a big part of the success. I think he could have built the game potentially like without AI. It probably would have blown up less. I do do agree.

Would have probably taken him long. too because there was like a bunch of stuff that he doesn't have any like domain expertise about at all right like that's not his normal thing but if he had some other like flashy feature that people thought was fun and interesting i feel like he could have woven a lot of the same story i mean the main way is making money is not people buying skins it's buying advertising right because people are seeing the game so it's not really they're buying for the game like the money is making is an ads which i'm not like there's nothing wrong with that that seems nicer than like like dark patterns, tricking 12-year-olds into buying upgrades for your game. Like, I'd way rather charge a company for ad space than tricking 12-year-olds into getting mom's credit card, right?

So that seems like... Classic Roblox. Yeah, I wasn't saying anybody in particular, you know, I'm unemployed. I can't get fired from that. I don't think, but I want to make it back on the podcast. So, yeah, so I think there are like elements of, like, yeah, it's part of his, like, large following and all this other stuff. But people are underselling.

how good he is at telling stories crafting a narrative, making it fun and enjoyable to follow along. Like, if he had just only posted the game at the end, I don't think anyone would care. It's seeing each of the little building blocks and he's excited that it's happening. That's what's getting people like, oh, cool, I'll go fly around. And then he gets a lot of traffic. That's because he's talking about it in an interesting and intriguing way. Like, I've clicked on the site. I've flown around.

I wouldn't have done that if Levels wasn't so good at telling stories. So one thing I also kind of observed with the sponsor and the monthly revenue is I don't think people are purchasing it for the chance to really get after the full audience because he even posts. He gets 20 to 50 concurrent viewers. I'm not sure if that's really like the bang going into that doesn't sound like what he's actually or what the companies are expecting to get back out. In my head, a lot of that is just the fact that he's going to be tweeting and talking about it and these pictures are going to. going to be flying around and they get to be associated with the levels I.O. brand. It is that storytelling. It is that thing that makes it so compelling. If anything, the big thing that this taught me is that people with AI that don't know how to enter into a domain that

they're just like unfamiliar with and it does have like a rather large, steep curve can lightly enter into it and make some level of splash and to be able to use their kind of maybe disproportionate social weight to be able to make like actually something that makes some money or become something that's popular that maybe probably like in all reality probably shouldn't have a plain game that you're just driving around shooting basic bullets like there's nothing to that game to begin with but it's the fact that you might just see levels i.o in the game or he might tweet a picture with you in it right like there's all these like exciting facets that are completely i guess away from the idea that it is you know just a game that people are buying i have a question yeah go for it so

this might sound pretty basic but to have MRR doesn't it don't you need like a couple months before you can like determine like like 70,000 MRI it didn't it just come out like three weeks ago like how is this recurring and this isn't like a dig on him this is more for me like understanding like how do you calculate these things except see like dax like call this out where someone's like

I have 100,000 MRR but it's only been off three weeks and those people could just drop off like the next month right so I'm just curious like how like that even comes up like with those numbers not saying like he's lying. I think levels is great. But I was just curious. Like when he first, when he first dropped that tweet, I was like, hmm, okay. Like, how do you, how do you get that title there? I don't know. Does that make sense what I'm saying? Yeah, I think this is like a pretty common, like, sorry, I made fun of Reddit and I'll make fun of indie hackers. Like, every dollar that they get goes into recurring for some reason. And like, that's not how that works. Like, a subscription is recurring or like if someone signs a contract, that's recurring. But like, if I release a course and then on the first day it sells $10,000. I don't have $300,000 of monthly recurring revenue, right? Like 10,000 times 30, right?

I just, I don't have that. I sold that much money and then maybe never another course again. That is like, I don't know, it feels like, oh, it always just goes in the recurring MRR just stands for money I made this month. You've just totally disqualified yourself for any future CFO position. Oh, shoot. You're right. Like that, like, you're going to have to have.

Return to office now. Or you're going to get fired. I'm back. I'm back. C-suite all over right now. Dreams of working at Enron are down the tubes for you, my friend. Every dollar made is booked for the full future. That is how it works.

I have one last point I want to make. I think there's anything I learned from this. It was how upset people got about this getting so much attention. It just, it was kind of like high school. Everyone's like, they're just kind of like, it almost came off as like jealousy in a sense that like, oh, he got so much more attention than what I'm doing over here. Like, I get it.

But it's just like, man, I don't know. It kind of rubbed me the wrong way. But it was also interesting to see how levels even like responded to like these claims. Like, oh, you don't know like these kind of concepts. He's like, whatever, man. I'm just like doing my best or whatever. Vers like. And then I started, I started playing it from the beginning.

and it was complete trash. It was so bad. I couldn't even, like, move. And then I tried it, like, yesterday. And I was like, oh, he's, like, actually making this better, which was interesting, which I think makes the whole thing better because he's actually trying to make it better. But anyways, I saw a lot of light-painted people in a weird way for me

after following this and, like, they're kind of takes on, like, instead of, like, encouraging. And, like, that's kind of cool. It was more like, you're dumb. You don't deserve this. And I'm just like, if I had your following, I would do the same. Yeah, I was just like, man, just, you know, just congratulate him. It's pretty cool.

I didn't know a rift off of that. But I also have a real answer in a sense, which is that, like, people for some reason forget, like, that Flappy Bird was, like, the most popular thing on iPhone for, you know, a good month. People seem to think that good and popular are, like, have something to do with each other. But they don't. Like, that's not how it works. And so I understand why people would be upset, like, seeing like, oh, some guy, like, you know, type some stuff into an AI and this thing, this poop came out that doesn't really work. And he got all this attention for it. And I'm, I'm, you know, upset about that.

But in reality, it's like, well, okay, but that's just kind of how the world works. The world, like, there is a segment of gamers who care about quality and you can address that market. And there is money there to be made. but there's also a huge market of people who are doing other things and that they're not like assessing this purchase based on that and like to the cloud imperium point for like star citizen there's also a bunch of people who are you know chasing a dream like they want this big open world and they're willing to pay a bunch of money like a like a like a you know like a Vegas crapshoot because maybe it happens right and and there's all of these other dynamics there and that's That's just how it is. It's messy. That's the game industry. It's always been that way. It's never been that you just do the best job and you get all the attention or anything like that. So you just have to kind of be okay with that. I certainly don't begrudge anybody, any of that. And yeah, like, I was joking about the Cloud Imperium thing. But yeah, it's like, I kind of said that for a reason, which is like nothing levels I owe is doing is anything worse than we do in the bigot. Right? Like, this is not an anomaly. This person is not. stealing money, okay, but by the industry standards. So just keep that in mind, is what I would say. In fact, to me, it even says something about just like recently there's been a lot of game

flops and there's a lot of things that haven't been working out. If anything, what it says to me is that perhaps the gaming industry is having a hard time really understanding their audience right now and Levels understands his audience and is able to break into an industry that he's not in and do something, even if it's just for a moment, that say the gaming industry has had like some pretty, massive flops Concord included to where it's like hundreds of millions of dollars just disappear in a single day,

which is pretty wild to think about the vast difference between these two. But we're going to keep on going with the AI because this is my favorite segment of the day, and I hope everybody has done their homework, because we are now going to look at the CEO of Anthropic, and he says

the following. Now getting to kind of the job side of this, I do have a fair amount of concern about this. on one hand, I think comparative advantage is a very powerful tool. If I look at coding, programming, which is one area where AI is making the most progress, what we are finding is we are not far from a world. I think we'll be there in three to six months where AI is writing 90% of the code.

And then in 12 months, we may be in a world where AI is writing essentially all of the code. So is AI going to be writing a single? essentially all of the code in one year from today. Oh, gosh, TJ, right off the rip. I'm ready. Yes, it will be because people will be able to generate one billion lines of code in an instant. And it will all be sloth. But it doesn't matter because by the metric of who's written almost all of the code, it will be AI.

It will be impossible for me to keep up. I cannot write one billion lines per day. It's a fact. You know, TJ, there's actually, there might be an accidental truth here. he didn't say it technically is... Accidental. What do you mean? What I mean is that I assume from his side, he's actually saying that all companies need to adjust right now. Like all of your engineers are effectively going to be not as valuable.

It's actually going to be the tool that they're selling that happens to be where all the values at. Your engineers are really just the cart driver, but it's the goods that are really and the cart itself that you kind of want to purchase. But it might just be that, T.J. That is so much overwhelming code going to be generated that people who are not using that or people that are even editing it are so far behind by accident he is correct and in 12 months it will be just so large that it'll be like 0.1% will be hand artisanal code and impossible to keep up with now casey i know you love ai and so you're probably going to side with this guy and probably a shorter time frame do you think that you will be writing your c and your game engine

and your optimizations with ai within the next 12 months oh man uh so you know you know the weird thing is you always pick these like very complicated topics and then you're like we're going to get through it in 10 minutes guys uh i have no idea what a i is going to do in the future i've said this many times the current kind of i that they've got is never going to do much in my opinion like when i've looked at the current i that they've got basically what it is is it's it's able to sort of mush stuff together in a way that the code that it's generating shouldn't probably have existed. So it's generating code that a better programmer

wouldn't have even had that layer that's there. They would have done something much better that didn't require the code that the AI actually put in there. So AI right now is not generating very usable code from the standpoint of actually doing a good job. But on the flip side, the average programmer in their job is also not generating that kind of code.

Like they are mostly doing that kind of glue layer stuff that really shouldn't have been there, but it's because you're going to use this library and this high-level database thing and whatever, and you're going to mush them together, and they're sort of sometimes going to work, and they're going to have a bunch of exploits because the quoting doesn't work

and all this other stuff, right? And the AI can easily generate that code. So when people talk about how much code is going to be written by AI, I'm never quite sure what they mean. If they mean AI will eventually get to the point where it can actually totally do that job, I vaguely believe you.

I can't say for sure that I think that will happen, but I could believe that that would happen in a year. If you mean that the AI is actually going to write any actually new code that's useful, like it's going to actually do a better job writing this PNG reader than a human would have done or whatever, I don't believe you because I haven't seen any evidence of that yet. It doesn't mean it can't happen. I just think it's going to need, like it could happen, but it's going to require them to make some breakthroughs.

And it's impossible for me to predict when breakthroughs. will happen. Maybe the breakthrough happens next week, and then we can have it in a year. Maybe the breakthrough doesn't happen for 10 years, and then we don't have it for an entire decade. And I just don't know, right? I don't work in that field, so I don't want to make predictions about their breakthroughs. So that's my honest opinion about it. I think, you know, I wouldn't want to be out there saying that in three months it's going to write 95% of the code. But, you know, I mean, like I said, I'm not even sure what that really is talking about. So it's hard for me to say. So, Casey, why can't you just embrace the vibe? That's, I mean, honestly, that's what I'm hearing right now is you're just not embracing the vibe and being the vibe code. Yeah, dude, I actually feel like a level down right now. Vibe kill right now. I don't even think I can program for the rest of the day after hearing you talk about this. If my AI's down, I can't either.

I got two points along with Casey's thing. The first one is I have yet to see like a long-running LLM kind of thing where it's not guided by humans. There's no expertise involved where it and we want to say like reduces the entropy of the system. Does it actually take something that's complex and make it simpler? Does it work towards a goal of making it simpler? Or does it simply add on things on top and make them more complicated? If it can never reduce entropy, then, well, then my earlier prediction slash joke will come true because it will have to continue generating billions of lines of code to continue to wrap all of the complicated stuff like until until it can solve.

So if it can't do that, then I don't see human expertise being useless, right? Which is kind of like what sometimes people pull out from this, which is not true. Like, it could be that the AI writes all the code and my expertise is still very useful, right? That's a separate question. And then the second one is once like Anthropic or, you know, chat GPT or whoever, when they can finally make a chat app better than T3. Shout out Theo, right? Then I'll believe that their LLMs are good enough to take my job. because why are they not making their own products good?

Why can they not make a chat app? Good. There's the related question, which is, if you actually thought that you could ship something that could generate all this code, why are you giving it to anybody? Why don't you just use it to make all of the products yourself and then you can capture all of these giant markets?

The lie that's sort of underlying it is very clear at that because you're like, oh, actually what you want to do is just make something barely good enough that an engineer might use it, and then sell it to them. So you clearly don't really have any plans to make this thing be actually good enough to take their job because they're your customer and you kind of want them to buy it. So you want as many of them as possible. So not only do you want them to be buying it and needing a human there, you don't even want to reduce the number of humans that do it.

You want to keep that number as big as possible because that's where your paycheck comes from, right? So it's a bit weird. Yeah, I've said before, like, Open AI will never, you will never find out that they have like proper AGI, whatever that means from a press announcement. You will find out when they've remade your business in an afternoon. Like, that's when you'll know, because it will just be a matter of compute. So there's like, they're not going to tell you. By the way, hey, Microsoft, we reinvented you. Hey, Amazon, we reinvented you today.

Airbnb, we reinvent. But they are just going to do that and say, congrats, we're the largest quadrillion. We're the first ever quadrillion company. We are all the companies. Like, they're not going to give you that shovel. They're just going to use it for themselves, right? Yep. I do like that idea of entropy and all of that, T.J.

where it's just a continual addition to the code base because we were talking about that, Casey, where we actually walked through some code and we saw exactly that, which is, as of right now, every single instruction that I gave it was just additional new lines and layers onto the code base where you could evidently see that. And also the T3 chat thing, it is so true. It's like how it just goes to show the difference between expertise and pretty good. Because I would argue that the people working on the UI and stuff at OpenAI at Cloud. any of these, they're probably pretty dang good engineers. Better than us.

Yeah. Economic UI. For real. We've already tried. tailwind. I'm like, ugh, like, but obviously, like, Theo's whole thing is making, like, these U-I's nice and very, very nice. And he just went and showed that he outclassed all of them by himself. And so therefore, or with one or two engineers or whatever it is, but they don't have this multi-billion dollar budget. He was, you know, off with the few engineers probably working

on very different wage scale and was able to out-compete them. So it does show that there's, like, a whole realm of human ingenuity and expertise that is, like, hyper-valuble still at this point. Now Trash, we haven't heard anything from you. I know you guys talk too much. No, I keep hearing your keys click around. Can you shut up with your keys, keys clicking? Yeah, I hear that. Mr. Clickety Clacky over there. I can tell because you're unengaged looking to the side. No, that's not, I'm listening and I'm trying to find a time to jump in. Okay, then get in. It's like double duck. You can't wait for double dots. I know. I know. Okay. Well, this is actually how I am a real stand-up. Just completely tuned out and just

ready for someone to wait for the screen to just turn black and I'm like, oh, the meeting end. And then I go do something. else. Okay. So, but actually, my whole take of this is I'm not very concerned. If anyone has been part of a company that has existed more than three years, they know that there's a legacy system that is completely undocumented. Nobody knows what does what. Tribal knowledge out to wazoo, there's absolutely no way that this LLM can go in and figure out what this, like, library did, that this engineer did 10 years ago and he quit, like last week, and he named all his functions wrong. Like, there's just no way. So for those reasons, like, I was born in Legacy Code.

Like, my specialty is to go into complete slop and just figure it out. Yeah, you know what I'm saying? So what you're trying to say is Netflix engineers produce complete slop. We got to stop saying that. We got to stop saying X Netflix engineers. Latest, long-time streamer, trashed them. Can't wait to join the podcast full-time. I'm excited.

Happy to have you, Travis. Welcome to the pod. You will be seeing this every week. But you know what I'm saying, though, right? You've all been part of legacy code that's just like pure tribal knowledge. Just like it's insane. So like I haven't seen like an actual case of that happening.

I know like Meta said they fired a bunch of people. So maybe they got to work internally. I don't know. But for like apps that you're using like open source technology and all that stuff, I think it's going to be a lot easier for AI to write those apps probably because they're well documented. Lots of code bases to reference probably. But like, you know, for a company that's not Netflix, that's also big, you know what I'm saying? Like, there's just no way in my mind.

But if it happens, you know, I'm more than happy to join the pod full time. Yeah. I do think, like, I'm not saying it's never going to happen. It's just when the prediction is like three months from now. That's like crazy. That's really fast. like they know we still do like flat file transfers for banks in the U.S. and that's how like trillions of dollars gets trained.

Like it's not like they're FDPing files at nighttime. That's how the banking system works. We're expecting that nobody at the bank is going to be writing code by hand at the end of six months. It won't even be legal to use AI in these codebases still. There was a news report recently that said that like, aerospace. Oh, sorry. So banking, aerospace, finance, medical, like almost all of them have like pretty strict do not use LLM.

Like some of them don't. But there is like a whole bunch of like you can't give away any of our competitive secrets. You cannot send any of this. So it's like there's entire industries that are just like completely resilient today. Government. This idea. That's like five out of almost. That's like 80% of coding I just named.

And they can't use it. Yep. Yeah. I'm not scared. Go. Go. Casey, what were you going to say? I was just going to say I saw a news report recently, and I don't know.

I mean, it was a news report, so who knows how accurate it was. But it said something like there are 3,600 separate programs that need to be run and communicate with each other to administer Social Security currently, much of which is written in Cobol apparently, right? And so I'm just like, okay, like that's happening now, right? And I don't know how well how good these LLMs are trained up on the various different incarnations of Cobol. But like that would be an interesting challenge.

It's like, hey, can you redo any of this? But so far we just haven't seen that. Like we haven't seen people demonstrating that these things are able to do much undirected work by themselves. And even the directed work tends to be very buggy and strange. So I don't know. It just maybe they're going to cross some barrier soon, but I don't know. You may not love this answer that I'm about to drop on you. But DARPA, it's been researching a project called Tractor.

And if you're unfamiliar with tractor, converting all that insecure, unsafe, old man yelling at Cloud C into the new kid on the block rust, making it same, fabulous and fantastic all at the exact same time. There you go. They probably saw the word cargo and they're like, that's something you understand. Cargo Road. Cargo Road, that's facts. All right. I'm going to give Trash Deb the last word trash. What do you have to say?

If you only know tailwind, you're going to get replaced by AI. If you know tailwind or if you don't? If you only know tailwind. Oh, okay. Good. Ooh. Very luck. Very good take. Why am I getting, I got the last one last time.

We trust you. You're not talking a lot and you're kind of just a little bit lame these days. You're all doing that one for you. No, you guys, you're like a clean up batter. You're the most important. That is so awesome. Wait, is the teeth. The TEEDEF is fake.

Everything else is real. Okay, be like levels. You miss 100% of the shots you don't take. Michael Scott. I'm dropping an AI game. Don't you worry, man. You just wait. You just wait.

All right. And then, of course, the last comment from the top shelf itself. Junior's not gaining experience will cause a low supply of senior plus. AI will always struggle with breaking down business requirements and translating to something useful. Do you think fun kind of thought exercise? Do you think, I know, sorry. Trash, I know I said you had the last word. I'm just kidding. I don't want you to have the last word anymore.

I'm going to let Casey have it. Casey, do you think we're going to enter into a world in which there's going to be like the zero mid-level engineer that causes some sort of really wild disproportional problem in our industry? Why are you? Trash was the perfect person. I've never worked at one of these companies. How are you asking me? Trash, what do you think? That's the end of the episode.

I stopped listening because you said you had the last word. I don't even know what the question said, if I'm being honest. Take us to black. Take us out. I was like, oh, the meeting's over. All right. All right. Well, hey, the name. Hey, by the way, this is called The Stand Up.

So I hope that you appreciate it. Like, subscribe. Also, go follow Trash on Twitter. It's in the bio. TJ. I also have your Twitter. TJ streams, by the way, in case you don't know. And Computerenhance.com to view all of Casey.

If you don't want to become irrelevant in the day of relevancy, check out computer enhance.com.