677: I Accept the Battery Cost

As I sit here tonight, the kids have a two-hour delay tomorrow, and they are finally going back to school on Thursday. They have not been in school since, what day was that, January 23rd? Wait, really? So this whole ice period, they haven't been in school since then? That's right. That's right. So we got the snow, I think, 24th, 25th weekend. Then we were, well, snow and ice. I shouldn't say the snow was really the ice that was the problem. We're off all last week, 26 through 30. 26 through 30. We're off Monday. We're off Tuesday. Well, I shouldn't say off. They did asynchronous learning, which means they did a bunch of busy work Monday, Tuesday and today, which is Wednesday. It's Thursday, the fifth. They're finally going back after after a two hour delay. And I feel terrible for Aaron in particular, because I've been up here working in the office. The kids have been downstairs in the dining room doing their busy work and asking Aaron questions incessantly all day long for the last three days. Erin loves her children more than she loves anything in the world, including, I think, me, potentially. And I think she's very ready for them to come back to school tomorrow. Yeah, I think, I don't know. There's days that I do okay that she might like me a smidge more. But don't ruin my moment here, John. But anyways, the point is, Icepocalypse is finally over, and I am very thankful. All right, well, let's do some follow-up. And I would like to say that I've been vindicated and in congratulations to John, or maybe not congratulations, but in thanks to John for putting this into the show notes, even though my conversions to metric are generally trash, they at least get us in the ballpark, and many people wrote in to say, hey, that's helpful. I mean, I'm not sure if it's vindication, but you have your supporters, let's say. There are people out there who love to hear you stumble through trying to convert in your head between units that you're not familiar with. That's right, vindicated. I feel like I'm okay with you continuing to put in metric units if you use kind of the less common intervals of them. Like, you know, like I'm currently drinking about 25 centiliters of a non-alcoholic Guinness tonight. Yeah, like the only metric unit that I think most Americans are familiar with, and some people pointed this out to us, is I think we know centimeters, because as someone pointed out, like lots of our rulers have interest on one side and centimeters on the other. So I have kind of intuitive understanding of centimeters. And then a meter, we just be like, well, it's like a yard. But yeah, kilometers and kilograms, forget it. Nope, I got nothing on that. And then some of the people, I love how some of the people supporting your unit conversions couldn't help but snark about Fahrenheit, too. They're just going to get that jab at even when they're trying to support you. Right. The thing of it is, I don't plan to relitigate this because we have a lot of stuff to talk about, but we get everything wrong. Inches are wrong. Ounces are wrong. Pounds are wrong. Yards are wrong. Everything is wrong. Everything is wrong except Fahrenheit. Fahrenheit for ambient air temperatures, we got right. It's a percentage of hot. And I will tell you it is a fact that if you have half degrees on your goddamn thermostats, you failed. That is a failure of your whole system. Celsius for ambient air temperatures is trash. You can use it for cooking. I'm good with that. I'll use it for cooking. I don't because I'm an idiot and I'm an American. But I'm full support for cooking. If you want to take your internal body temperature in Celsius, whatever, that's fine. Anywhere else other than what is the temperature outside. I don't care if water is cold, y'all. I care if I'm cold. And so I think it's hilarious that all these people that make fun of Americans because of, oh, we're stupid units and whatnot, seem to have this inability to remember the number 32. Like, all these people were so smart. Look at us with our base 10 units. Man, man, man. Look at our fancy paper. That makes sense. Unlike yours, which doesn't. Really make it hard. A lot of friends and the metric supporters right now, I guess. I'm just saying, you folks have everything right except Celsius. Celsius is trash. Yeah, so much for not going into it. Yeah, well, my bad. I can't help myself. I'm so triggered by this. This is the hill I'm dying on. I know you two know this better than almost anyone. This is your metric supporter, everyone. But that's the thing. I happily do my best to give you centimeters or meters or whatever the case may be. And kilograms, I'm terrible at kilograms, but I'll do my best. We already do like, don't we do like millimeters for like when Apple gives the units in metric, we just read them in metric. We don't convert them to, I don't know, 16th of an inch. Right. I mean, again, all our units are trash. Every single one of them with one and only one exception and only when used for ambient air temperatures. Celsius and cooking, I'm good. I'm happy to do it. That makes sense. Not when you're walking outside because I don't give a shit if water is cold. Anyway, other authentication options for John's Cloudflare apps. Tell us, John, what else could you have done? What a transition. Yeah, well, he's got some time to calm down. Well, for authentication. The last episode I was talking about how I factored out the sort of passkey-based account system for all my little personal apps. So I didn't have to, you know, copy and paste that from one app to another. Since I edited, I think I have like three of them now, three or four little apps. It's nice to have that all factored out, but a couple people wrote in for possible alternatives. One of them is another Cloudflare product called Cloudflare Access. I read their web pages a few times to try to figure out what it was. Here's the beginning of their explanation. Cloudflare Access provides visibility and control over who has access to your custom host names. You can allow or block users based on identity, device posture, or other access rules. And it says the prerequisites, you have to have a Cloudflare Zero Trust plan in your SaaS provider account. And it goes on from there to name a lot of different proper nouns that you need to set up in your Cloudflare account. I'm like, yeah, okay, but what is this? What, you know, I looked up zero trust and it's the, you know, the Wikipedia page says it's a design and implementation strategy for IT systems. The principles that users and devices are not trusted by default, even if they are connected to the privileged network, such as a corporate LAN or whatever. So then bad old days of computing in my jobby jobs. Once you got on the corporate network, you were kind of in and you were implicitly trusted to do one's and stuff. But later in my jobby job that became that went out of fashion and what became in fashion is a zero trust architecture, which is even if you're in the network, Even if you're on the Wi-Fi, even if you're plugged into Ethernet on the corporate network, still you don't have implicit access to anything because that's really bad for security. So it sounds to me like Cloudflare access is kind of like, you know, if you do any kind of remote work for a company, you probably have some kind of work VPN that you have to connect to that gets you on the network. And getting you on the network doesn't get you access to anything, but you have to be on the network before you can then authenticate using your ID provider or whatever. So I don't think this is appropriate for my mini apps to have a zero trust type thing where I have to install an app to sort of I'm not sure if it's a VPN, but either way, like it doesn't sound like the right system for me to be able to log into my little personal apps. But be aware that it is available on Cloudflare if you're interested in it. By the way, to quickly interrupt the whole shtick behind Tailscale, former sponsor, potentially future sponsor, they are very big into zero trust architecture. Now, with that said, by default, you know, they kind of allow everything everywhere just to make things easier. But their actual like recommendations and the way it's really built is zero trust. And it can be extremely powerful and extremely cool. And the next suggestion was OpenID Connect or OIDC. This one I also used at work. OpenID Connect is a sort of a way it's an open standard for authentication where you have like an identity providing server. and if you try to log into a service, it will bounce you to or talk to through an API, the central identity provider, and that way you can basically essentially make one account and be able to sign into all the corporate things. Again, it makes a lot of sense for an enterprise. You know, the dream of single sign-on, as they used to call it back in the day, is this whole very large companies that are built up around providing this, and OpenID is an open standard. A couple people suggested Pocket ID, which is a tiny little OpenID Connect identity provider that you can run yourself in a little host or a little Docker container. So you don't have to get one of those big commercial services. You can run this one thing over here, and that does your identity for all your other apps, and you're good to go. Maybe it's a little PTSD for my job and working with Okta. I think Okta is actually pretty good. Okta was pretty good, all things. Yeah, it was one of the better vendors, but still like a lot of the stuff I did at work, I have like bad associations with. And also it's so heavyweight. And I don't actually like what I have is kind of what I want. Like I don't actually want single sign on. I want individual accounts on individual services. Not that it matters. These are my tiny toy things that are going to have just my accounts. Or maybe I'll put like accounts on for like family members or other people who want to like take a peek at it. But like, it's just simpler for me to be like, these things are not connected in any way. There's no central ID provider. I don't have to, you know, do authentication over there and then authorization differently. depending on, like, it's just, it's too heavy to wait for me. But it is cool that there is this Pocket ID thing where if you do want to run this, you don't have to run one of these big, you know, giant commercial things or one of the free open source ones. You can run this little tiny thing, and it's just a little OpenID Connect provider just for you and your little services. Yeah, and building off of that, there's also TSIDP, which I think is written by Tailscale themselves. And this is, like, very similar in spirit to Pocket from what I gather, but it's a little baby OpenID Connect or IDC provider that you would run on your Tailnet if you so choose. I haven't really looked into this, but it does seem like the sort of thing that I would do because I'm a weirdo like that. But anyways, lots of options for you, certainly for the public internet and also if you just wanted to do stuff on your tail net too. And these work with pass keys and everything too. Like there's no, they're agnostic to the actual method you use to authenticate, whether it's a hardware key or a pass key or a password. They, you know, all these things support, I haven't looked into Pocket ID, but the standard supports all these things. All right, let's talk about TVs for a while. Let's see. David Schaub writes with regard to per-pixel lighting control, per-pixel lighting is sufficient, but it isn't necessary. My 14-inch MacBook Pro has almost five dimming zones per inch, which David calls DZPI? And I find it quite good because my eyes cause blooming and my glasses cause blooming. Tech only needs to meet human perception. There's no reason to assume that per-pixel lighting is needed if laptops get to about 5,000 zones, desktops get to about 15,000 zones, and TVs get to about 20,000 zones. I'm sure my eyes would be perfectly happy. Yeah, I mean, there's the point of diminishing returns, but unfortunately, due to the nature of the LCDs that are in front of those dimming zones, there are still sort of worst-case scenarios that come up surprisingly often. Maybe not the 20,000 zones, but to be clear, these numbers are pushing the high end of what is available commercially. I'm not sure if there are many TVs that are 20,000 zones or higher. Obviously, the MacBooks are half of that 5,000, but they're not the highest-density screens. But star fields, if you watch science fiction, it's a black background with a bunch of pinpricks of white light. And it doesn't take that dense of a star field to essentially require the TV to have every single backlight region turned on because there's at least one star in every region. And then you've just gone back to the worst case Apple Studio display. The entire backlight is on and your black levels are raised because the LCDs can't block all the light. Right. You know, there are lots of other sources of blooming, but that's an argument against, you know, dynamic backlight regions, because, yeah, there are lots of other sources of blooming. And we're talking about that in the next segment as well. You don't want to add to it. It's like, OK, and also the backlight is blooming. And to be clear, backlight blooming is so much better now that it used to be. It used to be incredibly awful and the zones were big and now the zones are tiny and it's much better. So it is definitely getting way, way, way better. But per pixel lighting control, like, why fight this? Each individual pixel can light up. We have a technology that does it. There are tradeoffs, but, you know, the gap is narrowing with tandem OLED and QD OLED and everything. So I get where David Schaub's coming from. And in practice, most people are buying the LCD TVs as well just because they're cheaper. But if and when they sort of meet on price, the answer is clear. Per pixel lighting control, don't worry about dynamic backlights. I mean, even if it's just like because correctly using dynamic backlights, deciding which backlight should be on and how bright they should be. It's computationally expensive and complicated. And there can be situations where the backlight lags the action on the screen. And there can be situations where doing the backlight calculations adds a couple frames of lag when playing video games. It's just a complication that is not ideally necessary. But, yes, they are getting a lot better. All right. So speaking of TVs and backlights and things, Marco had made kind of an offhanded question, comment, whatever. It's less of a question, more of a comment. Micro LED TVs last episode when we were talking about Sony and TCL. And, John, you had thoughts. Yeah, I tried to give Marco a lay of land, which is basically it's always five years in the future and they're not affordable yet. And just entirely coincidentally, I didn't look this up, but my YouTube recommendation engine threw this at me shortly after we recorded the program. And it was a video from Robert Tate of the Hookup YouTube channel, which I had never heard of before. But he bought and tested a 157-inch AWOL, all caps, micro LED TV, and TV in scare quotes. And he made a video about it. We'll put a link in the show notes. This doesn't really change anything that I told Marco, but it is a much more detailed view of, like, what's the state of things now? So, first of all, this is a $55,000 TV. So, okay. Well, you know. Right. And it it's 157 inches and it takes up the entire wall. Like all large micro LED TVs, it's made up of individual panels because the whole thing is like they have to you have to put these red, green, blue LEDs on these little boards. I think they have machines doing it still like and they're so tiny and there's so many of them. And it's kind of like silicon chips chips where the bigger you make it, if you screw it up, you got to throw the whole thing out. So they always make these things out of modular panels, even the ones that are like in stadiums and for rock tours. they're all modular panels that's the deal because we can make a small modular panel they're all interchangeable if one of them breaks you just replace the panel and it's easier to make a small thing perfect than to make a giant thing perfect um these modules are 27 inches they're 640 by 360 pixels at a 0.9 millimeter pixel pitch which is how far the pixels are apart and the general rule is if you multiply the pixel pitch by 10 you get the distance where you can't see the pixels anymore so 0.9 millimeters means about nine feet away and you won't be able to see the pixels anymore um it also the the little modules the 27 inch modules have magnetically attachable and adjustable screen things so there's like the back part of it that has like the power supply and the signal and everything and then the actual micro leds attach in a little panel with magnets and they're adjustable because you're going to set up these panels and robert's setup was six panels wide by five panels tall and you you know you put the panel you put the things in the wall we'll talk about the installation in a second but you have to line up all the panels so they're you know the gaps between them are as small as possible and they're all facing the same direction so one of them can't be like tilted you know they have to all be smooth and everything and so they attach with magnets and there's these tiny little adjustment screws to make to like level them in three dimensions so they all line up it's very fidgety um anyway that that six panels by five panels that's uh 3840 by 1800 so it's not even full 4k and if you look at his setup that's because there's not any more room in his room like he couldn't like it fills the height you know there's he couldn't put another row of panels in it's just they're too big and the whole thing with these is like making them smaller is the hard part you can put one in a stadium real easy you know for a few hundred grand or whatever but it's making the little micro led small so that's why when you see those the 55 inch micro led was causing huge waves when it appeared at ces a couple years ago it's like how do they make one that's 55 inch because it's so small um god knows how much for because that's no one wants you can't have in your house 157 inch tv it's just too big so anyway this doesn't do full 4k but it's enough to see all the pixels in like a widescreen movie because widescreen movies already have the letterbox things above and below and this just has smaller letterbox things above and below um the installation was fun so to get this installed he needed three power drops to that wall and a dedicated 15 amp circuit so it's not even like you just plug it in Remember, they're 27-inch panels, and they all need to be powered. He needed 18 Cat6 Ethernet wires to run to all the different panels. Oh, my God. And that was all run to a controller in a separate room. And then he just put three-quarters plywood over the entire wall mounted to the studs that he's going to now mount all of the panels onto. You should watch the installation. It's quite a thing. Bananas. This video was relatively long. I forget exactly how long it was, like 20, 30 minutes. but it was bananas watching this installation happen because he he didn't literally bring that wall down to the studs for the most part but he kind of sort of did like there was a lot of work yeah because he had to get all the ethernet cables in there and all the power and like the wall wasn't set up in this way before and so it just really needed to be ripped up so the performance if you do that you pay your fifty five grand what do you get so the brightness is only 800 nits but it's 800 nits on any size window as in you can light up one pixel at 800 nits or you can light up all the pixels at 800 nits because they're individual panels and in fact when he did a full screen test it was actually 840 nits for the whole screen so that's the beauty of micro led is doesn't have these sort of brightness problems of oled where they can't drive it that hard because of heat things will cause burn in and degrade the organic components and everything so oleds are like super bright in a tiny region but when you light up the whole screen with white it's way dimmer no that's not a problem here um much less blooming because there's just simply less glass and stuff over the pixels. I think there's still something over the pixels, but looking at it, it was hard to see it. And blooming is caused by, you know, our past thing that we just read from David Schaub was like, your own glasses, the glass on the screen, the glass that's on the front of your television, in front of the pixels, that causes blooming. And of course, within your eyes, I'm sure we'll get some vision doctor to tell us the details of that in a future episode. But this has much less blooming because there is much less glass over the individual LEDs. I'm not sure if there's any. It just seems like if there is, it's very thin. Color performance, 83% BT2020, so it's not 100% BT2020 like that RGB backlight thing we saw. 99% P3, which is still pretty good. Almost perfect accuracy out of the chart, which is nice. Didn't need to calibrate it or anything. Input lag is not great. It's like 34 milliseconds with no scaling and no processing and some weird bugs, but realistically about 78 milliseconds with all the features you'd actually want to use to play video games, which is not a good number for a TV, but they all fine for casual play. This is not a TV. This is a display. As I said before, it's a TV in quotes. You buy this for 55 grand and what you get is an amazing screen, but you can't watch TV on it until you add something. It comes with this Novastar MX40 Pro display controller, but that doesn't make it a TV. That's just a big box in the rack that the 18 Ethernet cables run to that makes the display work. And, you know, you calibrate it and tell it where all the panels are and so it makes a full image. but if you want it to be like a tv you have to buy something else so he does he added an hd fury room hdmi signal processor for 1 000 extra dollars because why not um and this thing lets it have hdmi cc arc and e-arc dolby vision hdr 10 dolby atmos dtsx a power button a remote and also a cool feature where you can do multi-view where you can put multiple screens you know divide up multiple signals and put it on the giant screen. And then finally, you plug this thing in with all its power drops and everything and you run it. What is the power usage like? Well, it's 71 watts for the Novastar controller thing. With the screen at 100% brightness, the display takes 1,200 watts. It's like a space heater, really. No, it's funny you say that. It's funny you say it. Let me interrupt you real quick. So when we were pre-Ice slash Snowpocalypse, I was trying to compute, you know, what can I run off of the tailgate battery if we just need it in a pinch. What can I run off our gas generator, which we might talk about later potentially? And I have like a little wattmeter thing where you plug that into the wall. You plug something into the wattmeter. It shows you how much power that thing uses. And we have a little baby space heater. It's a little tiny space heater. And that was something to the order of 1,300 watts. What is the maximum? Is it 1,800 or something? Like the maximum reasonable, like 80% capacity? Something like that, yeah. Yeah, but generally, like most space heaters are around 1,300 to 1,500. and it's like, yeah, you put one on a circuit, you don't put much else on that circuit, and that's it. So at 100% brightness, it's 1,200 watts. At 30% brightness, it's still 950 watts. So it's not even the brightness it's killing. It is essentially driving all those little computers behind all the little 27 panels. And this is the best one. When the screen is off, 600 watts. It's 600. It's like a gaming PC at full tilt when it's off. What is it doing? Well, the displays are off, but the tiny computers that are behind every single one of those little 27-inch panels are still running. They don't sleep them? Yes, seriously. This is what he measured. 600 watts when the screen is off. Oh, my God. I mean, maybe they go to sleep eventually. But anyway, like he said, if you have 55 grand for a TV, maybe you don't care about the 20 cents per hour it costs to leave it in its off state. You should watch the video, though. The picture is amazing. It is sad that he can't get full 4K on a TV screen that covers his entire wall and is 157-inch diagonal. The colors are amazing. The brightness is amazing. It's really cool, but it's $55,000. And, like, this is not, like, the highest end or whatever. Like, I think the company's claim to fame is, like, they make affordable ones. They found a good way to make these small components. But this is a really complicated, fancy DIY thing. and I feel like these type of screens are really aimed towards, so you're a multi-bazillionaire and you have a giant mansion and you have a theater room and you want to put something the size of a movie screen in it. These look better than a projector. He spent a long time, I guess he does projectors on his channel, he spent a long time saying, look, here's how this compares to a projector. Projectors suck compared to this. There's no planet on which a projector can have the contrast and bright room viewability because projectors need to be in the dark. Like as he pointed out, a projector will never be darker than the screen you're projecting on. Like how can it be? You're adding light to light that's already bouncing off the screen. So if all the lights are on in the room and you have a screen in front of you, whatever brightness that screen is, you're only going to add to the brightness by projecting light onto it. You'll never get the good, which is why you have to make it dark for projectors. And this doesn't have that problem at all. It's got perfect blacks, doesn't suffer from burn-in, has 800 nits of brightness, which doesn't sound like a lot, but when it's 800 nits full field, that's still pretty decent. Yeah, this definitely looks cool. I would love to have one of these if I had a giant mansion and a giant wall where I could fit it on there. But there's no way I would pay $55,000 for something that's not full 4K. So for what it's worth, I have three 5K panels that I'm looking at right now. I have a six-space Synology. I have some Ubiquiti Network Media all hanging off of my whatever it is, tech power, the UPS that Marco likes. Cyber power? Cyber power, yeah. I couldn't remember the name of it. Thank you. And from what I can tell on the little screen on the front of it, I'm using about 300 watts. And that's with all three of them lit up, the Synology chugging away. And that's half of what this thing uses when it's turned on. It's turned off sitting on there. Yep, yep. It's incredible. And don't forget the other room you need to have with, like, the server rack where you run the controller and the thing that turns it into a TV and you run all the cables to it and everything. We are sponsored by Gusto. Small business life means hustling and figuring it all out. even a lot of times yourself. But you don't have to spend your evenings guessing at tax forms or tracking on onboarding docs. I've done these things. Gusto handles all of that so you can spend your time on the parts of your business you actually love. Gusto is online payroll and benefits software built for small businesses. It's all-in-one, remote-friendly, and incredibly easy to use. You can pay, hire, onboard, and support your team from anywhere. You can do unlimited payroll runs for just one monthly price with no hidden fees, no surprises. Gusto supports all sorts of things you might need, automatic payroll tax filing, simple direct deposits, even things like health benefits, commuter benefits, workers' comp, 401k, whatever it is, Gusto makes it simple and has options to fit nearly every budget. They have all these tools built right in, things like automated offer letters, onboarding materials. You can get direct access to certified HR experts if you need any support with any kind of tough HR situation. So it's a great platform for your payroll and benefits needs. Switching to Gusto is quick and simple. You can just import your existing data, get up and running quickly, and you don't pay a cent until you run your first payroll. Gusto has been trusted by over 400,000 small businesses and was named number one payroll software according to G2 just a few months ago in fall 2025. So try Gusto today at gusto.com slash ATP and get three months free when you run your first payroll. That's three months of free payroll at gusto.com slash ATP. That's gusto spelled G-U-S-T-O dot com slash ATP. Thanks to Gusto for sponsoring our show. Let's talk about Claude Code and AI ethics. And I don't know, John, how you want to introduce this, how you want to talk about it, but what are we talking about here? Well, so I think the past couple episodes I've been talking about my experiments with Claude Code. And if you've listened to the show for a while, you you would have heard in the past, I guess, year or so or two, maybe. Well, when we first started talking about LLMs and chat GPT and stuff, a lot of our early episodes about that were like on the topic of how did they get this training data? What is the legal and ethical and moral ramifications of training on the world's data and then charging people to use it? I think we talked about like the Miyazaki like image generation thing in ChatGPT. It's like that that product cannot exist and has no value without first stealing all the works of Hayao Miyazaki. But then it's like, OK, but is it stealing? Is it fair use? What are you know, what are the courts going to say about it? We spent a long time talking about that. And I think if you've tuned in recently, you might be like, oh, they keep talking about AI, but they never talk about any of these other issues, including like what does it mean for people's jobs and worker exploitation? and what about the bubble and possible economic crashes? And what about the environment and all the power of these things? We've talked about all these issues before, but it was a little while ago, and that's kind of what we mostly focused on before we actually started using the products and before the products sort of came to be broadly useful instead of like a technical curiosity. And then recently I was talking about cloud code. So a lot of people wrote in to say, hey, what about all these other factors? Like, yeah, it's great that you enjoy cloud code, but what about all the data they're still going to train it on? What about the environment? What about eliminating jobs? What about the bubble bursting? You know, what about these tools being used for propaganda and oppression? You know, like you guys talk about that. And part of the answer is we have talked about that a lot in the past. But the second part is this item here, which is we should continue to talk about it. And what I want to say on this topic is that none of those things are resolved. Like, yeah, we did not come to any conclusions. And it's not like the world figured it out. And now we have a set of laws and rules and everything is fine and we're sure it's. Nope. None of that is resolved. Like there are various court cases that are happening, but it remains a completely open question. How bad is it going to be for jobs? Are they going to work out legalities? Are the legalities going to bankrupt things? Remember we talked about like people got like $3,000 per book that was stolen, but it was only because those books were stolen. But the judge otherwise said the training on books that you didn't steal is fair use. these laws there's there are many cases still winding their way through the courts about various big companies like the new york times and disney or whatever suing the ai companies we don't know how this is going to turn out and that's just in the u.s so uh you know we are talking about you know using these products and trying them out or whatever but it's not as if all those other issues went away nor have they been resolved and by resolved i mean my personally mean resolved i I mean, like figured out in a way that is sustainable, that we say, OK, we have a set of rules and guidelines around this. And we think if we continue along this path, people will be fairly compensated for their work. It won't disincentivize creation. It won't cause the end of humanity or whatever other. But no, we haven't figured out the right set of those things. We don't know. It could be that letting all these companies take all the world's work and then sell it and become, you know, they could become huge, powerful companies that dwarf any company before known by man. And we let them steal all of our other content. And it turns out that was a terrible idea. We don't know yet. So I don't have any answers here. And this is not, you know, we're not going to like dwell into any of these topics. But you'll be hearing more stories about how this court case turned out. And are these people paid? And what is happening with OpenAI? And, you know, like, and it has the bubble burst. And are they really using all those GPUs? And should it be stealing everything? And how, you know, how bad is it for the environment? And what is XAI doing in Memphis, putting giant gas generators outside their big data centers? Like that is all still happening. And you may know there are people who are on the extreme side of this to say, OK, that means you should never use AI at all. And you should not use these products and you shouldn't talk about them and you should be an abolitionist and you should throw yourself into the gears of the machine or whatever. Everyone's got to find out where they think is the right place to be along that spectrum. And I think we're all kind of figuring out, especially since we don't actually know how bad it may be. Right. We don't know what will the consequences be of putting all this and stealing all this information and selling it into product. Everyone is kind of used to the set of rules that we're used to that you were born with. Like, how does copyright work and how does it work for music and books and movies? And like you just accept those as like that's just the way things are. But anything new is scary and weird. But if you look, if you look into any of the other things like for television and music and writing and you name it, So the systems we have now are very imperfect and are not particularly any more just or well adjusted than than, you know, the perfect thing you have in your mind. It's just what you're used to. So that's my personal answer on this is that I don't know how this is going to turn out, but I am not an abolitionist and I not totally against looking into it because I recognize that it not actually particularly different than all the other forms of technology that have come before it And I hope we figure it out. And I'm fighting for this to be something sustainable. But I don't think saying that we should never, ever use this stuff is the right solution for right position for me personally. I would go a little further and just say, like, we don't have a choice. This is here. This is happening. It's happening with or without us, and it's happening to all of us. You can decide to sit it out, but what that's going to mean is sitting out the technology business for the foreseeable future. And that's a decision you can make. Like that's entirely up to you as a technology user, as a consumer. Like you can do that. But if you sit out AI, if you condemn the whole thing, and as John said, like there's lots of tricky like moral arguments and legal arguments. Like I could see why somebody would condemn the whole thing. But this is the tech business for the foreseeable future. You can be a part of it or you can fall behind. like if you are in the tech business and you want to continue to be in the tech business you kind of have to get on board in some capacity or at least you have to be okay with it and recognize like this is a massive force that is transforming not only our industry but many industries you know soon or currently um like it's here it's big it's going to keep being big. And no matter what happens in the short term, like if, you know, if the bubble pops, you know, people are people are saying that and look, there is a bunch of really, you know, massive amounts of money and ordering and speculation and finance tricks going around. But at the end of the day, if what we have right now is all it ever is, this is already incredibly useful and incredibly valuable to solve lots of people's problems, even if it never gets any better than what it is right now today. And that's not the case. Of course, it's going to get better. So this revolution is here. You can choose not to be part of it, but what that will mean is sitting out the technology business for the foreseeable future. Yeah, and you can choose to not be in the technology business because you're against it, just like choosing not to be in the cattle business because you don't want to eat meat, right? That makes perfect sense. And also, I think there's an equally valid position to stay in the tech business and lobby against it or lobby for, you know, better fit, more fair rules, just as people in the technology industry currently lobby for more fair rules for, like, paying artists for their music streams on Spotify or whatever. Like, it's possible to be, to stay in it and fight for a more just rollout of this technology. Like, I was, my parents were asking me about AI recently. I think it's finally penetrated the boomers. And they wanted me to explain the whole situation to them. And the analogy I use for them is one I've also used in a couple of emails that I've responded to from ATP. is like it's a lot like the Industrial Revolution. Whether it was like, is it a net good or a net bad? The Industrial Revolution caused incredible short-term harms and long-term harms. Short-term, filled the cities with smog, exploited workers, people got cancer from all the chemicals it was putting into the environment. It was just terrible. Like people who was out in the fields working are now like children are in factories with their faces covered with soot. You know, all of London is in a giant black cloud. People losing their jobs left and right because machines are replacing them, right? And then long term, there was incredible long term harms from the Industrial Revolution. Climate change that you mean, even though people were kind of picking up, that was probably going to be a thing very early on. It's like, oh, but it'll be fine. Look, the factories are getting cleaner. We have child labor laws now. I'm sure it's a smooth sailing from here. And then, you know, like, it seems like the planet's getting warmer. It's like, no, it's fine. It'll be fine. It wasn't fine. right um but on the flip side electric lighting indoor plumbing like the ability to manufacture things like the industrial revolution had tremendous downsides that are actually very similar to the downsides of the computer revolution the internet revolution and now the ai revolution and that doesn't mean that it's like you should just accept it because there's good parts of it too no you have to fight against the bad parts you have to get the pollution out of the cities You have to come up with cleaner energy. You have to implement child labor laws. You have to get the lead out of the pipes. You have to not let people build with asbestos. Like you have to do all those things. And I feel like a lot of the people who are fighting against AI, quote unquote, against AI, are simply trying to make it so that the children aren't in the factories. And then we stop putting soot into the air and chemicals into the rivers. Right. And it's the tricky thing with with AI is that because it's a it's so new and weird, it's not as clean cut as like, hey, don't put the runoff from your factory into our river like that was straightforward and it still took like decades to make any appreciable progress on because rich people are mean um but like and we're in the same situation here billionaires controlling this ai right but it's not entirely clear like is it possible to build a sustainable society where they steal all the well they they train on the world's work and then profit from it but don't give anything back is that can we build a sustainable system out of that? Or do we have to, you know, have lawsuits where people get paid for the stuff and they have to work? Like, we don't know. We don't know how it's going to turn out. It's not as straightforward as child labor and pollution, but it is very similar in that it's unclear whether it will be a net good or a net bad at any given point. Because I would say the Industrial Revolution was a net evil for a long time before it was like, okay, well, I guess having, you know, like machines and factories and manufactured goods and, you know, all that other stuff is actually pretty good. But for a long time, it was really bad. And then all of a sudden, now with climate change, like, oh, we thought we were over the hill, but we're not. We did it. We made some terrible choices. And now their chickens are coming home to roost generations later. Right. So who knows? AI could be like that as well. But all this is to say is that it's complicated. It's not resolved. And just because we talk about it in a nice way and say we had fun working with Claude Code and we're going to talk more about AI stock now doesn't mean we think everything's fine. There's no reason to look anywhere else. There are no other issues. That's not true. All the issues remain and are difficult. Yeah, I was going to say earlier, and then I think, Marco, you beat me to it, that it would be irresponsible for at least the three of us not to explore this because of both of our professions, both the clickety clacking and the yakety yakking. And that, you know, it is expected that we are going to be more efficient with our with the work that we do on our apps. And I think that leveraging these tools is kind of table stakes, which is what Marco was saying. And additionally, for us to stick our heads on the ground and not be aware of this and try it at least some is, I think, irresponsible for the show. And I think that it is literally our jobs to at least kick the tires and see what this is all about. And like John said and like Marco said, it doesn't make any of this like a perfect situation. It doesn't mean that we're excusing all of the ills. But, you know, the genie's out of the bottle and we just got to at this point, I feel like we have no choice but to at least participate in it to some degree. I don't see it as like we have no choice and there's no fighting it. I think there is fighting it. And I think it's equally our responsibility to continue to highlight the problems. Right. Yeah. Just as much as our responsibility to know about the technology just for our programming jobs and for our tech podcasting job. it is also a responsibility to continue to acknowledge and fight against the worst of the excesses and try to, you know, do the equivalent of stopping the factories from putting their runoff into the river and, you know, get the smog out of the skies and all the other stuff there, because there's so many equivalents to that in AI. And there will be like, we don't even know what the long term harms are going to be like. We don't know what the climate change equivalent is, but we know the job loss, pollution, exploitation, propaganda, like the printing press. The printing press was a pretty good invention, but boy, did it empower propaganda. so many harms came from the printing press but also good but that doesn't mean to say well the printing press is here there's no sense fighting it propaganda is going to be everywhere no fight it fight it but you're never getting rid of the printing press is not the solution but you know what I mean like anyway how do you feel about CDRs well luckily we don't have to worry about them because they're gone baby that's all I'm saying it's worth us noting and I think John you would put a link this is kind of tangentially related but you would put a link to a friend of the show, even though I don't think he knows any of us. I don't think any of us know him. But Alec from Technology Connections put together a 90-minute video that is ostensibly about renewable energy. And I definitely learned a lot from this. And that's the first 60 minutes. And the last 30 minutes, continue with renewable energy, and then things take a turn. And I beg of you, if you can give 90 minutes to this, I genuinely think it is well worth your time. If you can only give 30, though, start when he's kind of rolling the credits, so to speak, and he shows all the Patreon supporters and all that, and just pick it up right there. And you will, well, I don't think you'll be disappointed, and you will either be shouting, hell yeah, or hopefully you will reconsider some of your priors. But this video is incredible. The only reason I put it in here is because it's a good, it's one of the things that I think about when considering the environmental impact of AI. Okay. As I said on past shows, I'm perfectly happy to expend energy production of humanity on things that are useful for humanity. So I'm good with power plants heating my home, right, and providing lights because I think those are good things, right? Keeping lights on the roads at night for safety and stuff like crypto, stupid, but AI, valuable. Yeah, we generate power to do things that are useful for us. And it's a question of, okay, well, but AI is taking too much power and it's not useful enough or whatever. Uh, the, I view it mostly as the energy generation problem is a separate thing. No, we should not be burning coal. Yes, we should be using solar panels and renewables. And we're doing a terrible job of that in our country. And we apologize. Um, but this video is good at addressing like technologically speaking, uh, things actually do look vaguely hopeful. If you don't, not if you don't look, if you look at the U S things look miserable, but if you look at the rest of the world, renewables are kicking butt. And this is, if you haven't been keeping up with this, this video from technology connections This is a great overview of just how good a renewable's got. It was another one I forgot to get the URL for, but it was a recent story that a wind farm in Ireland or something was being shut down. They had 21 turbines. Oh, yeah, yeah, yeah. They had 21 turbines that had been put up in the 90s. They were shutting the wind farm down, and they were replacing those 21 turbines with new ones, and every single new turbine they put in produces more power than all 21 of the old ones combined. That's progress. Since the 90s. That is progress. Anyway, watch this video. So the political rant at the end, I don't think we'll convince anybody that doesn't already agree with it. But if you do already agree with it, it can be cathartic. We are sponsored this episode by Masterclass. With Masterclass, you can learn from the best to become your best. It is always great to broaden your horizons and to learn new things and to pick up and hone your skills. And Masterclass really is an amazing gift either for yourself or for other people because you can give the gift of knowledge and get better at things. So you can do amazing things at work. For instance, you have a course called Win at Work with Professor Jeffrey Pfeiffer and his Power Playbook. You can apply CIA-tested tactics to everyday life with the art of intelligence. You can negotiate your next raise with lessons from Super Agent Rich Paul or FBI negotiator Chris Voss. And then you can do other things like design your dream home on your budget with Joanna Gaines. Apply the principles of improv to your life with Amy Poehler. Develop good repeatable habits with Atomic Habits author James Clear. and so much more. There's so much in Masterclass. And their plans start at just $10 a month bill annually. You get access to over 200 classes taught by the world's best business leaders, writers, chefs, and so much more. So you can turn your commute or your workout into a classroom. There's audio-only modes. You can listen to Masterclass lessons anytime, anywhere. And so no matter what your schedule is, no matter where you are, they have apps to support you on your phone, laptop, your TV, so Masterclass goes with you wherever you can learn. There's no risk every new membership comes with a 30-day money-back guarantee, and memberships come with bonus class guides and downloadable content to help you get even more out of each lesson. Right now, our listeners can get an additional 15% off any annual membership at masterclass.com slash ATP. That's 15% off at masterclass.com slash ATP. Masterclass.com slash ATP. Thank you so much to Masterclass for sponsoring our show. All right, let's do some topics. And we have to reenter Vision Pro Corner. And now I have a buddy. And so, John, did you watch both episodes of Top Dog? I did. I watched the dogs. I've watched the dogs as well. That's the whole thing, right? It was just those two episodes? I thought we were getting like a sample, but it was just two and done, right? Yeah, I think it's like a sum total of around 30 minutes, give or take a little bit, if I recall correctly. But, yeah, it's about the, what is it, Crufts or something like that? Yeah, so CRU is like the thing in your code where you got like some old code. You said you got a lot of Cruft in this code. It's CRUFTS. It is the world's biggest and oldest dog show. That's right. It takes place in Birmingham, I believe. and this was a immersive two episode show about that dog show and let me I know I've been through this a hundred times please just indulge me 2D is you know your television 3D is a rectangle that has depth and immersive means not only is there depth but you can actually look around and change your perspective which you can't do in 3D every time you say that I bristle because it's the wrong it's fine just go on I am happy John for you to give me a I know what you mean. I know what you mean. I would just phrase it in a different way. Well, I'll work on that. I'll take that as a homework assignment. But anyway, this was entertaining. I mean, I am not dog obsessed. I really enjoy my dog. I generally enjoy dogs as a point of order. But nothing about this, like I don't watch dog shows that I don't really care about. But this was, to my eyes, very clearly either directed or edited or both by someone who has only ever done 2D content and has no freaking clue how to do immersive content. And Marco and Ben Thompson, front of the show, have banged the drum over and over again that you cannot cut or if you do, you have to do it extremely sparingly. We also spoke about this with regard to basketball. Generally speaking, I think that Marco and Ben are a little too far in one direction and that I think a cut here and there is fine. This, to me, was brutal. It was all the things that Marco and Ben are always talking about. It was a cut every few seconds. The camera moved a lot more than I think it should have because that's kind of, and I've said this before, I don't really ever get motion sick, but when the camera is moving, it kind of gives you the feeling. feeling um i i really feel like this was basically a three it was it was created and cut such that it was more 3d than it was immersive literally speaking it was immersive as i turn my head i can see different parts of the scene in front of me but in terms of the way it was presented it was basically 3d there was very little that was like fun and cool that you could get by just turning your head to one side or the other. And that was really disappointing to me because I really love the immersive stuff. And I keep I keep banging the drum of like the Metallica concert was incredible. I thought the NBA stuff was incredible. And that you can look around and you have the time to look around and change your perspective and look at something that maybe the camera isn't wanting you to focus on, but you find interesting for whatever reason. This was none of that. And I honestly, I was pretty disappointed by it. And by the way, the language I would use for this is small FOV versus large, small field of view versus large, but field of view in two dimensions, not just left and right, but also up and down. And, yeah, I did notice that on this thing as well, because in any Vision Pro, you know, quote, unquote, immersive content, you can turn your head far enough where you see the edges of the thing. Just often the edges are just so like the field of view is so large that you're looking at like half a sphere or a quarter of a sphere. And so it's really you turn your head way to the left, way to the light, way up, way down. You know, you can see your feet. You can see the sky. sometimes there are even 360 where the field of view is a perfect spear and there's no place no edge but usually there is an edge and this thing the edges top and bottom were real close like you could look up and down a little bit but you couldn't see the people's shoes and you couldn't see the boom mic that was inevitably hanging overhead maybe why they cut the field of view because and in that respect that's why you're saying in the case it felt more like watching a tv show type thing because in the tv show type thing you can't look up and see the boom mic and You can't look down and see the people's sneakers. You just see what they frame. And so this was much like that top and bottom. Left and right, there was more freedom. I felt like, was it 180 left and right? I don't know if it was 180, but it was pretty big. Maybe, you know, I don't know. The field of view was pretty wide horizontally, but narrow vertically. The cutting didn't bother me at all. I thought the immersive environment was a good place to see dogs, because dogs are cute when you can see them in 3D. I would agree with that. i think whoever made this is not very good at making a stereotypical reality tv show thing having watched a ton of reality tv like this is just a gimme man it's like a dog show there are personalities there people want that like you got to have you got to give people the the villain edit and the the hero and the underdog and you know you got edited together knowing how it ends to make and they just it wasn't it was like they're not very good at their job in terms of making it sort of chewing gum for your mind kind of like fluff uh reality show like you don't have be mean to the participants you don't have to make fun of them or whatever just develop drama and they tried but it was so so yeah um the other thing i'll say is like the part i was most interested in i was hurt so much by their poor uh filming of it probably because they didn't have enough cameras which was the uh fly ball competition oh i know where the dogs have to run and get a tennis ball and run back i'd never seen that before saying i wanted to see it and they wanted to show you the big dramatic finish and it's like you didn't you basically didn't even get it on camera i mean you kind of did but it was so far away it's such an awkward angle it's like this is not how you film a sporting event they also had a bunch of times where they had straight up 2d content and i thought that they handled it well it was kind of like projected onto like a screen i'm not doing it justice i actually talking about like when they they would have just straight up 2d content that they would show you you couldn't really move your head and change the perspective it was just 2d content presented on a screen you're never changing your perspective you're just looking at different parts of the image. John, just bear with me here, all right? And your perspective means you'd be able to see around the back of something that you couldn't previously see, but you can never do that. All right. You can't change your particular focus or field of view. You're just looking at different parts of the FOV. Yes. Anyway, the point is there were moments, several moments, where they had just 2D footage because I guess they didn't set up the immersive cameras or what have you. Like for past years or just in the show? No, it was in the show. That's how flat it was that I don't think I even noticed this because there's like... And that's what I was going to say is they did it really, really well. Like it was presented in a way that it wasn't off-putting or anything like that. But ultimately, to your point, and what made me think of this is that they didn't capture everything in Immersmith. They whiffed. Oh, it's still very cute. Another thing that I think is interesting, I mean, you're a rant on the cutting. This is one, I'm not sure if this is just that I'm not used to it, but just be aware this is a thing you ever watch Immersmith. In regular television or movies, they will do things like, they will have the camera at varying distances to people. So let's say someone is sitting in a chair for an interview and like a documentary. You'll see them like from the waist up and they'll be talking, sitting in the chair. And then like for some dramatic moments, they'll have a close up and you'll be seeing them just like their head and shoulders filling the same size frame. That's common, you know, a wide shot, a close up of, you know, people or whatever. In immersive, because it's 3D and because you like feel like you're there, like you could reach out and touch them. It messes with your sense of scale. Like they had this lady with like a dog on her lap and it was like close up, like maybe like from her chest up with the dog on her lap. And then they cut back to like a more sort of like a shot where the person in the Vision Pro was the size the person really was. So if they're five feet tall, they look five feet tall in the Vision Pro. And suddenly the dog looked like way smaller because before I was closer to the dog. But I just thought it was a giant dog because I had no way to like compare scale. Like you see this when they get close to people's faces. It's unnerving because it looks like a gigantic dog or head or whatever really close to you. It doesn't like it's just that I don't know if they can do the same thing with scale because when they go close up, you don't think that person is huge. You just know that they're taking a tighter shot. But when you can see their 3D, I kept thinking the dogs were bigger than they were until I saw them like walking around. I'm like, oh, no, that dog is the size of a rat. It just looked huge because I was three inches away from it and it was in 3D. And it felt like it was very strange. So maybe that's just me having to get used to stuff like that. But it is a weird experience to not know the size of things because they seem like they're right in front of you. But they're not. It's just a picture on a screen with a big FOV. Yeah. All right, Marco, you did an experiment recently. Tell me about that, please. It's kind of an experiment in progress, actually. Oh, OK. I think this might surprise you. Oh, gosh. Me personally or just both of us? Well, OK, just as I'll try to make this quick. I decided, you know, kind of in like the cortex yearly theme arena this year, I wanted to focus generally on efficiency in my life. Like there's a bunch of, you know, just kind of cruft I've built up over the years, friction I've tolerated over the years that I'm trying to just kind of find inefficiencies and consider whether I want to keep them that way or not. You know, I'm not like being ruthless, you know, trying to cut every single efficiency, but just kind of be aware of inefficiencies. Think about them. Consider whether I want to make a change. And then if I want to, then, you know, then make a change. I jump in for a second here. I'm thinking in my head, what inefficiencies does Marco have in his life? And I'll be honest, not a lot is jumping to mind. Like, you don't strike me. Like, you're not the kind of person to tolerate inefficiencies. So I can't wait to hear. Do you have any ideas before he tells us what he's talking about? Casey, what inefficiencies are in Marco's life? Like, it seems like there's not a lot. I'm struggling. Like, there's more inefficiency in my life for sure. I feel like. I mean, like, I have two houses. Like, that's never efficient. But you have so much less stuff than I do. Oh, I don't think that's necessarily the case. Well, I've got nothing. Your life always seems like the model of efficiency to me. Oh, my God. Then yours must be a disaster. It is. I've seen the basement of the beach house, and I wouldn't say that there's unreasonable things down there, but in the basement there was a lot of things down there, as an example. We're thinking of things, though. That's not just the – like inefficiency is like, you know, every day I come home, I have to do this thing because the knob sticks, and I just deal with it, and the lights don't turn on. And Marco doesn't tolerate that. He gets it fixed. That's the kind of thing I'm thinking about. All right. Well, maybe there's things in your life that we don't know about, but it seems – I mean, your computer stuff and your technology stuff seems pretty efficient. Well, except the only thing I can think of is something to do with cellular and travel and hotspots and things of that nature. Of course your mind would go there. Well, I mean, Mark, when I... I mean, he can get rid of the Vision Pro, but he was the first of us to get rid of his Synology or just essentially stop using it because it didn't fit into his ruthlessly efficient technology life. You were giving me a lot more credit than I deserve for efficiency. All right. Well, go ahead. All right. Let's see what you got. So this is... I'm looking at a bunch of different things this year. I mean, I'm looking at Overcast, like, you know, just like, you know, the hosting stack of Overcast is, you know, pretty inefficient. Okay. I should have thought of that. You're right. That is inefficient. Yeah, there's stuff there. I'm looking into doing more with S3-type things with Cloudflare, different things there that I'm going to look at throughout the year. I mentioned very briefly last episode that I was looking at some financial planning stuff, and there were some inefficiencies in how I was managing investments, and I've been working on those recently. But one other thing that just has been bothering me a lot recently, is how badly 1Password and Safari work together. Oh, you're still using 1Password? So that's, and I started looking at like, okay, what, should I just go all in on 1Password or go all in on Apple passwords? And, you know, because they've been kind of fighting each other for years now. You should do one or the other for sure. now something else has changed recently is that i started using ai a lot i started using chat gpt i many times a day just asking small questions basically where i used to do more web searches or asking siri for things i've been doing a lot more asking chat gpt and that's been going very well another thing that happened recently is google gemini got better and when i actually compare factual answers where I know the correct factual answer between ChatGPT and Google Gemini, Gemini was getting it right a little bit more often. And there are certain things about Gemini that I don't like as much as ChatGPT, but the sophistication of the model is hard to ignore. It is seemingly a little bit ahead of ChatGPT in a lot of things, not in everything, and it has its own quirks and annoyances. But Gemini has been pretty good recently. And I also started thinking about AI, and I'm like, you know, Google, I think, has a really bright future here. Because when you look at, you know, how these AI things are advancing and, you know, what's the next step? Well, the next step for ChatGPT is figuring out their business model and, you know, making their ad system. and as you integrate more into things you need data sources you need deals with other vendors or things like business and directory providers mapping providers different affiliate deals with everybody to try to monetize through that if you want to be able to book hotels or whatever you need that Google is in a really good position for this world they had a bit of a slow start But they make the models. They have their own in-house models. OK, OpenAI has that, too. That's great. Google has their own cloud infrastructure. OK, well, OpenAI doesn't really have much of that. Google makes their own chips to run it. OpenAI doesn't have that. and Google has all these existing properties. They have data sources. They have the entire content of the web that they already have in their possession through means that most people don't want to turn off and I know they try to separate it but it's not that separate. They have Google Maps and they recently just integrated Google Maps pretty well into Gemini and so I was able to do something like I was in the city yesterday and I said make me a walk. I want to walk about four miles. get me a walk through Central Park that's about 4 miles somewhat scenic and it generated Gemini generated this walk and it was stupid I asked for a 4 mile walk and it said here's a 3.1 mile walk and I'm like hold on I asked for a 4 mile walk please make it 4 miles and it's like okay here's a 4.6 mile walk numbers are hard for them whatever but using Google Maps it showed me the route and then it said you can send this to your phone, to Google Maps. And it had a link, so I could send it to someone else walking with me, and I could save it if I wanted to. And that showed me, okay, this kind of integration, I think Google has a really big edge here over everyone else. All the integrations that they're going to have with all the data, I think they're going to really build a big edge in the near term, and long term probably as well. One of the reasons why I was interested in sending the link to my phone from Google Maps through Gemini is for all these years, I've been bouncing between three different Maps apps on my phone. Google Maps, Apple Maps, and Waze for different purposes. Waze was my driving directions app because it's good with navigating traffic on Long Island. Google Maps, I would look up business info and stuff like that. And Apple Maps I would occasionally use for something else, you know, maybe like walking directions. But Apple Maps drives me nuts on the phone in one particular way, that when you have active direction following going on in Apple Maps, it takes over the lock screen of the phone in a way that seems to significantly and reliably slow down unlocking the phone for me. Because it's like you unlock the phone so you can see, like, so you can navigate the map, really, and then you can extra swipe to go to the home screen. I can't find a way to turn that off, like the way Apple Maps takes over the home screen, the lock screen rather. And so that actually drives me crazy when I'm trying to use Apple Maps for navigation. So that alone is a reason why I mostly didn't use it for that. And then Google Maps, I would look up businesses, and it has by far the best business data with like, you know, find me coffee shops and you're here. And then are they open right now? What are their hours? Are they good? Give me ratings. like Google Maps destroys Apple Maps in that. And Google also owns Waze, and so they have access to all the same traffic data. And so I decided this is ridiculous to have three different Maps apps the year of efficiency I going to try to use just one Maps app for everything And the only one that made sense to do that with is Google Maps So for the last month or so, I've been using exclusively Google Maps. Okay, so when I had Gemini generate me my Google walking directions, it was great. Send it to my phone. Okay, done. And it was, you know, perfect. It was there. You know, I could pull it up instantly. And it's a walking route through Central Park that had like six different stops and they all transferred perfectly. And I could view it again. Like when I actually went on the walk, like the next day I viewed it, it was still there. It hadn't like fallen out of memory or whatever. Like the link still worked. It was still sent to my phone. It worked really well, honestly. And Google Maps for driving directions, there's a lot of small ways in which it works better than Waze. I know this is a confusing thing to hear. like the M5 Max Max. There's a lot of ways that Google Maps is a little bit better than Waze. Not in every possible way, but in most Waze. It's a little bit nicer, a little bit more advanced. It's also a lot faster at everything. One of the downsides of Waze is if you ask for directions somewhere, it kind of spins for a while before it generates them. I don't know why this is different, because Google Maps and Waze are both owned by Google and both seem to operate in most of the same data. But getting directions in Google Maps is almost instantaneous every time. That is not the case with Waze. I've used Waze for years. It isn't my connection. Like, I know Waze is slow to generate directions, and Google Maps is instant. Google Maps also had better voice search. It's better for, like, oh, find me a charging station along this route. It's better for all of those things. So, so far, I'm very happy with Google Maps as my only mapping app. Again, it's not perfect, but I think it's way better than using three different apps for the same thing. That was not the efficiency I expected you to say. That's a minor efficiency, but I'm glad to see you drop down from three to one. Yeah. I mean, I occasionally will use Google Maps, although very rarely. And typically, if anything, it's for business information, like you had said. I almost never use Waze, but I'm almost never in places that have traffic. And I actually find that Apple Maps is good for me for 98% of the things that I use it for. It was not good at first, for sure, but the last few years, it's been quite good. The issue with Apple Maps, again, besides my irritation with the way it takes over the lock screen and slows down and locking up my phone a lot for some reason, my issue with Apple Maps is the same issue I have with Waze that if I'm going to only have one mapping app, neither of those is good enough to be the one. I get it. You know, because Apple Maps really falls down with business information. I didn't even know it had business information. I always use Google Maps for that. It's not even there. And it's not even like, I know that's part of Maps, but like Google has like reviews and like pictures of all the food and the menus. And like that's a whole separate like product. And it's like, oh, and by the way, this point on the map brings up that info. But Apple, I mean, I guess Apple has some business of it, but it doesn't compare. But I use Google Maps and Apple Maps and Google for all the reasons that you said. But I use Apple for driving because they're much more attractive and larger for my old eyes. viewing of the road and the lanes I find really helps. Although I will say that sometimes it still makes me do very stupid things and I get angry at it and switch back to Google Maps. So I'm down to two, but I'm glad you're down to one. All right, so also in the same vein, again, thinking about the whole picture here, AI is a big deal and it keeps getting better and it keeps getting bigger and it's taking over what used to be web search for me. I realized that so for my web search engines, I for years used DuckDuckGo. Same. And one of the things I got hooked on with DuckDuckGo is the bang shortcut syntax, especially for searching Amazon. So when you have DuckDuckGo, you can type in exclamation point A space and then whatever you want. And that's a shortcut to Amazon. You can customize those. There's a bunch built in. There's G for Google if you want to quickly jump over to Google search. There's bang eBay for an eBay search. Like there's a bunch built in. I mostly use the Amazon one a lot and occasionally the eBay one. And by the way, if you use Safari like I do, you can do that with a Safari extension where I just type az space and then a search in my address bar and does the same thing. Yeah, there's a few different extensions, some of which I found via chat. But yeah, there's a few different extensions that do this. But Safari also supports DuckDuckGo as a built-in search engine. So it's very, very easy to switch to it. But in recent, in the last, I think, two years or so, I've been using Kagi. Same. I heard it on the talk show because the CEO has actually been on the talk show, I think, twice with John Gruber. And it sounded interesting to me. And I've been using Kagi for the last couple of years. And it's been pretty good. Like, as a web search, if you actually want web search, you want to, like, find a review of a new fridge, Google is garbage for that. It's just it's been gamed like crazy, and it's really hard to find good info on Google. And part of that is because everything's been gamed. Part of that's also because there's increasingly less and less good content to find because the web is dying in large part due to AI. But also it was dead before that. It was dead because social networks killed it. So the web is really not in a great place right now content wise. Hasn't been for years. And Google hasn't really seemingly worked very hard on their web search in that time either. So, you know, it's not like they're, you know, without fault here, but the web's in a tough spot and web search has therefore been really bad. And that's part of why I have so much value these days in AI, because it's just it's better at giving me information than web search has been for many years. So I've been using Kaki and I've been, you know, but I realize I've been less and less actually going to the search results page. Usually I'm using Kaki to just search Amazon. on so what you know maybe in the era of all of this ai stuff i'm like i'm trying google maps i'm trying gemini why don't i just try using google as my search engine again what am i doing fighting all this like i fought i fought against google i've raged against the google machine for years for decades actually and why like let's give it a shot so i have switched back to google as my search engine oh i don't like that at all i never left although the ai search results at the top really annoy me yes i know i know how to get rid of them with the little yeah to be clear like the the ai summary on top of google search results is horrendously bad it's very bad sometimes it's funny so it's so often comically wrong and when i say gemini is good that's not what i'm talking about the worst part about them though is because they are at the top of the google search results everyone in my life anytime they quote unquote look up something on their phone i have to ask them to just scroll past the ai search result i'm not like just i want to know the actual answer the ai search results at the top of google are hilariously bad they're so like i think google is giving ai a bad name with those i honestly think google should they should either make them use a better model which i know i kind of understand why they're so bad yeah because like the scaling of that it would be incredible right like they don't want to get like it's too expensive to give everyone a good ai search result but like i feel like almost like they should go back to the old days where they just have like canned responses to common queries that they would actually like look up and have like you know just the dumb old sort of like hard-coded if someone asks for this thing run this thing instead of just like oh check it to gemini but one of our tiny little models and it will give total bs so it just annoys me to the top at the top you have to scroll past and i forget what the query parameter is that you can put into your google search query so it doesn't show that but it's just such a hassle because it's not so much me i can scroll past them it's everyone else in the world who now quote unquote look something up in their phone and they read the total bs it's the top of the google search results and i say no keep scrolling keep scrolling i really think google's doing way more harm than good for both themselves and for the for the perception of ai with those i mean but the well i was going to say the good news, but it's maybe not the good news. But like as the cost of inference goes down, presumably those answers will get better ish. But I will never accept them as the result of quote unquote looking something up. You didn't look something up. You just took a stew of words that came out of a probability machine and read it back. If you're going to take the time to type something into Google, look it up at a source. So get some kind of source, just sources from somewhere, not the amalgam of all the words information run through a blender. Right. So anyway, tell us how you really feel, John. So, all right. I'm trying out Google with my search engine. And so far, I've said it that way a couple of days ago. I've barely noticed any difference because I the first thing I did was, oh, let me figure out a way to get my Amazon shortcuts to work again. and so I you know there's these extensions and then something else happened a few days ago I saw somebody filling out a form and I noticed their autofill worked a hell of a lot better than mine okay every time I place an order for something you know Safari will offer me like you know to fill the information and it's it's not good Oh, you switched to Chrome, didn't you? Yeah, that's what I was thinking. The auto feel is a little bit better sometimes. I am trying Chrome. Oh, my God. You're going to have an Android phone before the end of the year. No, that's definitely not going to happen. I'm not so sure. So that's the final shoe here is that I have tried switching to Chrome. I'm only like a day in. You're becoming more like me. You're using Google Maps. You're using Chrome, Google Search. I've been using those things all along. So here's the thing about Chrome. Chrome is as annoying as Google has always been with Chrome, with the very heavy-handing pushing the logins for everything. And so, yeah, you know what? Fine. I logged in. I'm doing it the way they want me to do it. The same way, like, if you want to use Apple products, you can, like, rage against the machine and, like, you know, not sign into an Apple ID and, like, not use the app store. You can do it. It's just harder. And, again, my thing here is, like, I have been fighting this fight for so long. no one cares it's not affecting anyone but me meanwhile when i see certain things the way they work in chrome look i've been using safari full-time on all my devices for well over 10 years and i have to bounce over to chrome pretty often because whatever i'm using doesn't work in like some like you know oh i have to fill out like some state tax website thing oh it doesn't work right in Safari. First I try, okay, well, what if I disable all my content blocker? Maybe it's my fault. Nope. Try again. And it still doesn't work because it just doesn't work in Safari. And you can make an argument for like, is this, you know, is this people not following the standards? Is it Safari not following the standards? Is it web developers not testing things right? And the answer is yes, it's all of those things. But my problem is that sites don't work in Safari sometimes and I have to go redo the same action in Chrome. That's Apple's problem to fix. I don't want to make it my problem. And whatever the reason, whoever's at fault, it's Apple's problem. Why am I continuing to battle this unless there's really, really good value there? Okay, so if Safari's way better than Chrome in a bunch of other ways, then I'll take that. I'll keep fighting that fight. But whenever I pop over to Chrome to do something, I notice a couple of things. Number one, everything works. Number two, it's so fast at everything. Now, part of that is, you know, I think there's a few animations and stuff. It also still looks the way web browsers should look, unlike Liquid Glass. But okay, everything in Chrome is faster. Everything. Opening and closing tabs is faster. Loading websites for some reason is faster. Everything is faster in Chrome and not a little bit. Switching from Safari to Chrome feels like when I switched from Intel to Apple Silicon in terms of performance. It's like everything is just baseline faster by a lot. And also, 1Password works a thousand times better in Chrome than in Safari. Logging into websites, auto-filling forms, it's so good and so fast. I have been wasting years of my life typing my name and address into web forms that didn't auto-fill correctly and filling out password forms because it didn't auto-fill my password or having to paste in my 2FA code because that didn't auto-fill. Chrome works so much better in those ways. And 1Password works so much better in Chrome. And it's all so fast. So that's what I'm trying. I'm trying Chrome as my default browser on my Macs. I downloaded the iOS version of Chrome. I don't know if I'm going to stick with that, but I wanted to give it a shot. No, don't bother with that. You're not missing anything there. Yeah. It's the same engine. That's the problem. Well, but there will be other benefits if sync and everything are good, which I don't actually know yet. So obviously when you switch from Safari to Chrome and you're otherwise an Apple person, you're going to give up certain things. You're going to give up certain integrations. Now, one thing I've noticed is that Chrome on Tahoe has the autofill for messages, which is nice. I don't think I'm going to switch to Tahoe on my main Mac yet because I still hate everything else about it. But I was using it on my laptop that has Tahoe, and I noticed that it did the autofill for messages thing, because I think Apple added an API for that in Tahoe. So that kind of made me jealous of Tahoe briefly. The one biggest downside to Chrome, which luckily you're avoiding, is if, like me, you are an iCloud keychain user and not a 1Password user, the Apple's keychain plug-in extension for Chrome, it's fine. I understand why it works the way it does, but there's some additional friction because of their, you know, understandably paranoid security where if you don't use it often enough on a particular site, you have to end up unlocking it by typing in a six digit code before you can do the autofill. And that's a step you don't have to take with one password. And that's a step you don't have to take when you're using iCloud Keychain in Safari. So there's a little bit of friction there. And then the other bit is if you've been using Chrome since the beginning, like I have and never stopped, you end up with tons of like stuff in Google's password manager and migrating off of that. Because it used to be you couldn't use iCloud Keychain at all directly within Chrome. And then they made that extension. So some of my passwords are in Chrome and some of my passwords are in iCloud Keychain. And doing that transition has been a pain, especially since basically the most expedient way to do it is to not disable Chrome's autofill. So now I've got two autofills fighting in the same Web page. and i'm i'm still i'm almost done with that transition but uh one password this is one of the the beauty and former sponsor of the show whatever but one of the things about one password is that it is in fact cross platform cross everything so it is works in safari it works in chrome it's the same wherever you put it it's not like apple's thing which works way better in safari than it does in chrome yeah and so i at this point like so far my experiment with chrome Again, it's only been a couple of days, but so far, it feels amazing to be this fast and this automated in so many ways and to just have fewer problems. And this is the kind of thing that I'm hoping to do more of during my year of efficiency. It's just like I've been tolerating certain friction and doing a lot of manual work and shuffling things back and forth. And I know this isn't going to be a panacea, again, because I'm not going to have Chrome probably on my iPhone because it is kind of weird there. But because I'm not going to do that, I'll miss out on, like, you know, certain shared histories and things. Like, if I'm using Safari on the phone, but Chrome on the desktop. So, you know, I know there's going to be some additional friction there. And maybe, long term, I will realize that's not worth it. And maybe I'll switch all the way back to Safari again. But right now, like, the initial impression of just having Chrome as default, it's just so fast. And I'm looking forward to things not breaking as much or at all. Because I've just been dealing with so much of that. I have a different name for your yearly theme. It's the year of basic. You're becoming basic. You're going to use Google for search. You're going to use Chrome for your browser, which, by the way, go look at any kid. And, like, they want to use Chrome as their browser. My kids refuse to use Safari because they've just grown up on Chrome. Yeah, you're using the things that everybody else uses. You're not using DuckDuckGo and Kagi and sticking with Safari. you're just going to use Google Search and Chrome and Google Maps, the year of basic, but in a good way, in parentheses. Yeah, because, like, again, it's just I'm tired of fighting all these fights that don't matter to anybody but me. Like, all these years, I've been raging against this stuff, and you know what? In the meantime, like, you know, Chrome has some downsides, and Google's really creepy, and it used to have a pretty big battery cost on Apple's computers. Well, I think Apple's computers are so good now that the battery cost, You know, I'll take it. Yeah, the battery cost is still there. Safari is still better in power. I'm sure. But you know what? If it gets me this much speed, fine. I accept the battery cost. You know, these laptops have great battery life these days. I can spare a little bit to have my time back. And so that, I think, like, that's a major shift here. I think AI is a big shift here and integrating Gemini with all your stuff. I think it's going to be increasingly valuable. I'm not going to switch to Gmail or anything, although I did briefly consider, like, should I start, like, forwarding all my stuff? email to gmail so we can maybe integrate stuff there you're turning into me yeah you really are god help us all try using an intel mac uh that's not gonna i mean let's not be too ridiculous um so this this is where i am like and i i want to i think the ai revolution is blowing a lot of stuff wide open in tech and it's going to keep doing that as we were saying earlier and so i want to actually challenge my my assumptions that i've had for like the previous eras of tech my My assumptions of I should avoid Google because they kind of creep me out and I don't like the way they do some stuff. It's like, okay, well, Apple has their own BS too. None of the tech giants are particularly clean in that way anymore if they ever were. And so there are certain areas that I think deserve reevaluation sometimes. And in the wake of significant disruption, which is what AI is doing and will continue to do, that's a good opportunity to re-evaluate some of your assumptions and and long-held habits and allegiances and it's like is is this still best for me is is this are my assumptions still the case and are the choices i made 15 years ago still the choices i should make today uh or are certain things worth re-evaluating and so that's that's what i'm doing i'm heading into this year uh doing some of that re-evaluation and uh hopefully save myself a bunch of time along the way and hopefully making some good stuff for the show, whether it succeeds or not. Chaos is a ladder. Casey? Nope. Anyone? Nope. Absolutely not. I'm not asking you, Marco. Come on. Yeah. Anyway, chaos is a ladder and Marco's climbing it. What is that from, John? Game of Thrones. I've never seen it. No. Jesus Christ. I didn't expect Marco to see it, but all right then. We are sponsored by Factor. Now, look, everybody tells themselves they want to eat better. Oh, I should eat healthier. I should cook more at home instead of getting takeout so often, etc. Look, I've been there. This is what I tell myself all the time. The problem is we don't have time to cook every single night necessarily. Cooking takes a lot of time, you know, prepping and shopping and everything. Factor doesn't ask you to meal prep or follow recipes or anything like that. It just removes that problem from your place. So you get two minutes and you get real food and it's just done. Factor is made by chefs, designed by dieticians, and delivered to your door, and you just heat it for two minutes and eat it. It's great. It's nice, fresh food. You get lean proteins, colorful vegetables, whole food ingredients, healthy fats, the stuff you'd make yourself if you had the time. But, of course, that isn't always the case. We know that. There's no refined sugars, no artificial sweeteners, no refined seed oils. It's just good, healthy food. I've tried Factor. My family's tried Factor. Everyone likes it. It gets positive reviews from guests, from family members, even the kid, like he likes it too. Everyone likes Factor Meals. Whatever your needs and preferences are, they have things like high protein, calorie smart, Mediterranean, GLP-1 support, ready to eat salads. There's a new Muscle Pro collection for strength and recovery. So whatever your goals are, whether it's healthy eating, calorie management, more protein, or just easier meals, Factor has you covered. They're always fresh, never frozen, ready in two minutes, no prep, no cleanup, no mental load. So go to FactorMeals.com slash ATP50 off. Use code ATP50 off to get 50% off your first Factor box, plus free breakfast for one year. Offer only valid for new Factor customers with code and qualifying auto-renewing subscription purchase. Make healthier eating easy with Factor. Thanks to Factor for sponsoring our show. All right, so breaking news as we're recording. Apple has just released the last 24 hours. Is it a beta or it's a beta, right? It's RC, release candidate. Release candidate for Xcode 26.3, which adds agentic coding. Reading from Apple's newsroom, Xcode 26.3 introduces support for agentic coding, a new way in Xcode for developers to build apps using coding agents such as Anthropics, Cloud Agent, and OpenAI's Codex. Agents can search documentation, explore file structures, update project settings, and verify their work visually by capturing Xcode previews and iterating through builds and fixes. In addition to these built-in integrations, Xcode 26.3 makes its capabilities available through the Model Context Protocol, an open standard that gives developers the flexibility to use any compatible agent or tool with Xcode. I have not personally had a chance to try this. I have been busy trying to get ready the next release of Call Sheet. But, John, you've dabbled for quite a while, right? Yeah. Well, so first of all, I think it is notable that Apple has rolled this out in 26.3 of Xcode rather than like saving it up for a big reveal of WWDC because they see the writing on the wall, which is everyone is using these things. It's not just because they saw that I was using it last week. But like, you know, like people have been using these things for months and months. Oh, crap. They got Syracuse. Yeah, that's right. By the time it's like AI penetrating to the boomers, it's like by the time I'm using it, really, Apple's behind the times that they're not integrating. And they had like the AI coding assistant in their models and stuff like that. But everyone is using things like Cloud Code and has been for ages. And it's like, why is this not integrated with Xcode? And we've heard rumors about that for ages. I mean, one of the original AI rumors is like that Anthropic, Apple is doing a deal with Anthropic to integrate Claude into Xcode or whatever. Anyway, they released it. It's here. I've tried it. It's nice that they are. It makes them seem nimble. It's nimble for Apple. I'm not saying it's nimble because it's not because they're behind. Right. But it's nimble for Apple to actually roll this out. And it more or less works. And I wanted to try it. So I'm still on Xcode 26.0.1 because of the Tahoe icon issue, where that's the last version of Xcode that will let me have the new icon, the new OS, and the old icon on the old OS before Apple broke it and refused to fix it. So I installed 26.3 on one of my development machines, the one that's running Tahoe, in fact, and I wanted to try it out. And the problem I gave it was I can't use the Tahoe icon on Tahoe and the old icon on old OSes. Fix that for me. This is your hotspot. Yeah. And so I tried Claude Code on it, and Claude Code said, oh, you know, you can't do that, and here's why. I'm like, I know. I know Claude Code. But I want you to try and see what we can do. And I went back and forth for a while. It was heroic. It tried a lot of things. It was like writing Python scripts to parse the assets.car file and doing all sorts of stuff. And we tried all sorts of, whereby we, I mean it, and me just, like, telling you what to do. or all sorts of stuff. The integration with Xcode is pretty good. Like it is, you know, I've only used the command line cloud code, but I know that there's lots of GUI stuff out there, like the new codex thing from OpenAI. And anyway, the integration Xcode in the sidebar, it feels kind of cramped, but it's like, it's fine. Cloud code can do lots of stuff in Xcode, but not all things. Occasionally it would say, hey, put this as a new build phase, copy this build script. And it would have nice affordances for like you just click on and it will copy the full contents. You're not like dragging out, you know, But it couldn't add a build phase itself. It needed me to copy and paste the shell script into the build phase and do all this stuff. So I was like, oh, that's kind of cruddy, but whatever. Anyway, it failed. It couldn't get it to work. So I'm like, okay, next up, let's try the OpenAI's codex. And for Claude, again, I was using whatever that expensive plan I have now. And for OpenAI, I forget which plan I was using. But you get to use the good model. You get to use 5.2. And I tried with codex, and it said the same thing. Oh, I just looked this up, and that's not a thing you can do. And I said, I know, I know it seems like you can't do it, but let's find a way. And I went back and forth and basically I rubber ducked with it a lot because, you know, when you're explaining rubber ducking is when you're facing a programming problem and you explain to a rubber duck. Rubber duck's not going to help you, but the act of you having to explain it makes you, you know, verbalize things that you hadn't thought of before. And so me with rubber ducking, I'm typing into the prompt like, I know it seems like you can't do it. Like, look, Xcode 26.0.1 does it. The app is on the App Store right now. If you download it, it has the Tahoe icon in Tahoe, and it has the pre-Tahoe icon. Like, I know it's possible. Just because the version of AC tool, the command line tool, that came with all versions of Xcode after 26.0.1 can't do it? The old one could, so whatever it's doing, let's do that here. And going back and forth with rubber ducking, I was like, wait a second. Why don't I just take the assets file with the icons in it from the version I built, like from the App Store version, from the version I built with 26.0.1 and copy it into my Git repository for my app and just say, hey, add a new build phase. When all the building is done, your final step before code signing is to chuck out the assets.car that you just built with Xcode 26.3 and replace it with the one I just copied from my app store build. And so that's what I told them to do. I had rubber ducked it, and I'm like, this is terrible and stupid, but it should work. And then rather than dirty my hands doing it, I just said to Codex, here, here's it. slash application slash switchglass.app slash blah, blah, blah. That asset star profile, put it in my Git repo, make a build phase, and by the way, Codex can make build phases by itself. Codex didn't ask me to do anything. Codex seems to have much deeper hooks into the MCP. I don't know if it's MCP or whatever it is, but it has much deeper hooks into Xcode where it never asks me to do things myself. It's just like, I can make a build phase. I can do this. Although Codex was so needy with asking permissions, no matter how many times it said always allow, always allow, always allow, there was always another prompt coming. So So these things are a little bit fuzzy. But anyway, Codex did this for me. I now have a terrible, hideous workaround. And, of course, if I ever want to change my app icon, I have to do it in Xcode 26.0.1. And, yes, I have filed a feedback against this as an enhancement request asking Apple not to be this dumb. We'll see if it ever gets this. But in the meantime, I use the coding agents. I think for a first implementation in an RSE, it's pretty okay. Lots of people are doing even more impressive things with it. but I was happy to see that it served its function both as, you know, letting me try these integrations but also letting me rubber duck with it because I think that is actually one of the roles. I mean, Casey, you talked today about talking to it as like kind of a co-worker, but rubber ducking is another thing that co-workers are good for. They ignore you. They have their headphones on. They're just nodding politely, but you explaining the current thing you're fighting with can make you realize, oh, what am I doing? Why don't I just copy the good file out of the app on the App Store now and check it into my repo? Done and done. that is a very weird way of going about this but when you don't have a whole lot of other choice it makes sense yeah it's terrible like to be clear it's hideous like it's i should never have to be done but you gotta do a lot of hideous things in tahoe and then the other the other angle on this is uh so uh kyle hughes uh posted a graph i'm not sure where it came from but his wording on it was 2025 was indeed the dawn of drop shipping apps as the prophecy foretold uh I'll put a link in the show. It's a drop shipping, but it's like by making it easier to become like essentially a merchant online where you don't have any inventory and you just take orders and then take the thing from somebody else and make it ship to you. I did increase the number of people selling things online. Well, the graph is how many iOS apps have been released per month. And you can put a dotted line on the graph where like agent of coding arrived and the lines going up much faster than it was before. Like before it was basically flat and now it's taken off and it's not shocking because if you put a coding agent to xcode and you can just type words in the sidebar in english and get an app out of it and you can totally do that uh yeah there's going to be more stuff submitted to the app store probably more junk but uh it's a concern for all of us because it's not like app review was uh particularly uh it's been reasonably speedy compared to the bad old days but like this increase is not going to help things yeah i mean i think this is this is the least of our problems in terms of... That's Apple's problem, but yeah. Yeah, I mean, I think the area of, like, how much is agentic coding going to affect both us as developers and also the software environment on the iPhone? Those are huge questions. And I think the answer is going to be it's going to affect us a lot in both ways. It already is affecting us a lot. That's the thing. Yeah. It's already happening. Yeah, this is happening today. And we didn't even cover the whole Clawbot, Maltbot, OpenClaw thing. There so much happening right now driven by AI that it hard to even hear about it all let alone try it all or or become proficient in any of it it you know we in such high motion right now but will software developers like us will we even be that necessary going forward or like you know what what will junior developers do like maybe you have one senior developer you know controlling a bunch of agents like that themselves are controlling a bunch of agents like there's so many ways this can go but it's it's looking like we should not underestimate the capabilities of these systems uh for the foreseeable future because there's there's so much that they're that they are already doing today this is an area like i personally i feel like i'm way behind because i have not yet used much much ai coding at all i haven't done any of the agentic stuff i've only been like you know asking the chatbots for code snippets here and there like you know I was writing something for Overcast today, and I was like, oh, I have this set of data here. How do I write me some Swift code to take this set of data that I have and make it into this other type of format? And it's like, oh, well, the algorithm you want is called this, and here's an implementation with Swift that uses generics. And I plopped it right in, and it was great. But even that, is this just like me typing in my name in a web form? Why am I wasting time copying and pasting Swift code that is generating into my project and integrating it myself when I could just be using these agents to say, just make this feature work. And I don't think we're that far from that, even for very complex things. We're there already for a lot of simple stuff. And this stuff has just been born, basically. We're so early into this, and it's this good already. This is getting better rapidly, and I think we are in a real inflection point for many industries, including our own. And I don't know that it's necessarily super destructive, but it's certainly very different. And we, again, we've got to get on board or be okay with being abandoned. Yeah, and comparing this, I know it's the same underlying tech, but comparing this to the early optimism about how LLMs are going to lead us to AGI, the real AI, the thing that used to be called AI before we took the term and perverted it. You know, like HAL 9000, you know, human level intelligence. Like all we need to do is add more parameters. And the experience with like, you know, OpenAI going from ChatGPT, you know, two and three and four, you know, going up is like, well, they kind of have plateaued and how good those things that you can talk to work. It seems like that's not a direct path to AGI. But the coding engines using the same underlying technology, we're not asking them to be HAL 9000. We're asking them to do one very specific thing, which is right code. and they are getting better at that, kind of like how fast LMs are getting better between GPT-2, 3, and 4. Like they're in that phase. Now, maybe they'll plateau out as well, but we are in the upward curve of coding agents. They are getting rapidly better, much faster than the, oh, ask me a general question and I'll tell you something and hopefully it's not total BS, right? Because again, I say it's because code, you know, it can test it itself. Like you can't do that with facts. You ask something and it gives you, well, here's a plausible answer. But is it right? It can't help you with that, right? But code, you can run it. You can write tests against it. You can iterate on that until the tests pass. You know what I mean? And there's tons of code out there, both open source and whatever. You know, like code is tractable in a way that general intelligence is not because you immediately run it and see if it does something. Whether you're running it or the agent is running it, you can tell to a much higher degree whether it is right. And I think that is helping these agents increase much more rapidly than currently the sort of like, I'm just a chat bot, you can talk to me. Like they seem to not be getting better as fast. Certainly they're not getting as better as fast as they were between chat TP2, 3, and 4. I feel like we're in that phase now. So I don't even know how good these things are going to be in like five to ten years unless this curve levels off. But right now it's not leveling off. it's getting better like every day seemingly so it's an exciting technology yeah and and again even if it is already leveled off and even if it never gets significantly better than how it is right now it's still amazingly useful and is changing everything like and yeah and it will probably keep getting better i mean i i think i would bet that it's continuing to get better but yes it already is useful but like i talked a lot about this in the directives i just recorded yesterday like it is a skill to know how to use this just like it is a skill to be a senior developer that leads a team and talks to junior developers that's not it's not a given that people will be good at that and it is work and it is difficult it's just a different kind of work and so there are people out there who are currently way better at using agents to get work done than other people like i'm just playing with it just because it's fun like but i i recognize how much of a novice i am at playing with these things and that it is in fact a different skill than having an actual human team and it's definitely a different skill than writing the code yourself And these are all important skills. And I don't think any of them are going away, going away. But don't think, oh, now anyone can do this because you can talk to it. It does make it open to more people. But the skills required to use an agent well are actually fairly difficult to master and require lots of expertise and are not sort of evenly distributed through the populace. So it's really just more of a a change in who might excel in a world where most coding is done by agents. But it's not easy. And I don't think it will actually be the case that, you know, depresses salaries or whatever. But I do think, you know, again, like the Industrial Revolution, lots of people will lose their jobs and then people they'll die. And new people will get different jobs. And it's going to be disruptive. It's going to be harmful and disruptive. And if done carelessly, it's going to cause a lot of problems. But I think it is it will eventually be seen as progress, hopefully done in a way that is more humane than the Industrial Revolution and maybe also more humane than the PC revolution and the Internet revolution. It's moving so fast, it's hard to tell for sure, like how much of an impact this will have to how many things. But so far, like that estimation, as we learn more about these and as we use them more, that estimation is going up, not down. Yeah, like it is very clear this is changing more than we think. This is reshaping a lot of industries, but it definitely is reshaping software development and using it like I'm saying it is its own skill. It's it's like the birth of Google, speaking of which, like the birth of Web search. When Web search became a thing in the 90s, if you were good at using Web search, you had like superpowers. And imagine how that translated to almost every part of life when no one else was using web search that much, and you were the nerd who knew – who had access to the internet, who knew how to use web search, and maybe even were, like, the person who got good at using, like, search operators to actually find better information. Before Google got rid of all this. Yeah, right. But, like, you know, in the early – like, that was a superpower. And eventually then that power became – you know, everyone had it. And if you were like the one neophyte back then who was anti-computer. Wait, is that neophyte? What's the other one? No, that's not neophyte. The opposite. Yeah. You're trying to think of Luddite, but Luddites get a bad rap. Luddites get a bad rap and they were mostly just arguing not to have reasonable labor laws. But anyway, that's what turns out. Oh, okay. Well, if you were anti-technology or ignoring technology throughout the 90s and 2000s and you missed the entire Internet revolution, you could have gone on living life just fine. Millions of people did, but they were increasingly being left behind by that world. We are now in the point with AI, we're around 1999 in terms of search engines and the Internet. Like this is booming and it's radically changing a lot of things and you can get good at it and you probably should get into this world because it's going to change. It's already changing a lot. It's going to keep changing more and more stuff. And you don't have to. You can just ignore this, but you will be left behind. You are being left behind already. Soon, like most of the workforce is going to be young enough that they don't care about whatever copyright things. you might be worried about. They don't care about whether AI is sometimes wrong or whether it's making stuff up. I hope they care about that, please. But sorry, like, they're not... I think being... Correctness will still always be a measure somewhere. Otherwise, planes are going to be falling out of the sky and we're going to be in idiocracy pouring soda on our planes. We might have some challenges in that area. But look, like, you know, didn't we have the same challenges with web search? Yeah, I know. I think this is definitely more... There is more of an upside, but there's also more of a downside. In web search, like, in terms of a skill, it's not that difficult of a skill, whereas this is more like learning how to use Photoshop, where Photoshop is this amazingly powerful tool, and learning how to use Photoshop to its fullest potential makes you extremely valuable, but there's a lot to learn. Like, I know, like, one ten-thousandth of Photoshop, and already I have more skills than most people I will meet. And someone who really knows how to use Photoshop, Photoshop made them incredibly powerful and there's a lot to learn. And I feel like coding agents are like that, where there is a huge depth of stuff to learn. And the problem with coding agents is they're changing every single day. So if you waste your time becoming an expert in some particular coding agent, like all your skills are going to be obsoleted tomorrow, but it's just like learning Photoshop 1.0 and then 2.0. Then they add layers. You're like, what the hell is a layer? Like in the early phases of powerful tools, there is an advantage to learning to harness them to maximum capacity, but expect your knowledge to be trashed periodically because things are changing so fast. Yeah, and that's going to be the case for a while. But like this train is moving. Like this is where things are and will be going for the foreseeable future. And so and like, look, I am very often slow to adopt new technologies. And I recognize that I am behind in adopting AI coding. even you know i'm like you know two months behind but like but i'm behind and i need to i need to get on that again like if you want to keep working in this business this is it it is going to be a lot of skills that we're going to have to develop you know all of us old people who who grew up without this we're going to have to do a lot of relearning the same way when the internet came up a lot of people had to do a lot of relearning who were already doing things a certain way for 10 20 30 40 years you're a graphic artist before photoshop you're like what the hell none of my skills seem to transfer over. I know I use all these papers and tracing things and exacto knives and tape and like you want me to use a computer? What? Yeah, but that's that is where we are. Jump in because this is it. Like this is not being put back in the bottle and if you think copyright or moral arguments are going to save you, I'm sorry, I have bad news for you. I don't think that's going to happen. You know, there were also copyright concerns over web search engines because in order for Google and other search engines to index content, they have to read it. And that's making a copy, even temporarily in the server's memory, that's making a copy. And this was litigated, whether or not it was fair use to make a copy of copyrighted stuff from a web page to just like read it, like if from from a program from either a search crawler, or even a web browser, like you have to make copies in memory to do anything with anything. That's copyright violation at the purest form. And that had to be litigated. Guess what? Technology won. The massive, massive utility of technology won out over nitpickiness over copyright law. And I think the same thing ultimately is going to happen here. I mean, there might be a slightly bumpy route to get there, but copyright's not going to kill AI. AI is here, and the value, the utility is so high, you're not going to be magically safe. Nothing's putting this genie back in the bottle. Nothing's going to save us from this. Well, you really have a lot more faith in our judicial system than I do. Just because it doesn't make sense to you doesn't mean it won't become law in this country. Have you not learned anything? Well, there's a lot of money behind it. Don't worry. There is. But still, even that's not a guarantee. Anyway, I'm not as I don't I don't agree that it's the same situation. I think we do need to make some adjustments for reasons that we've discussed on past episodes. But I just don't know how it's going to turn out because trying to use like logic and reason to predict what laws will be upheld in this country. It has not worked for decades, and it's really depressing. All right. Thank you to our sponsors this week, Gusto, Factor, and Masterclass. And thanks to our members who support us directly. You can join us at atp.fm slash join. One of the many perks of membership is ATP Overtime, our weekly bonus topic. It's usually about 15 to 25 minutes of one more topic for the show that we just – it kept falling down the list, and we didn't actually get to it. This week on Overtime, we're going to be talking about apparently Apple is rumored to be making basically an AI pin. We're going to see what that's about. We're going to talk about that in Overtime. You can join us to listen at tv.fm slash join. Thanks, everybody, and we'll talk to you next week. John didn't do any research. Marco and Casey wouldn't let him. Cause it was accidental. It was accidental. And you can find the show notes at atp.fm. And if you're into mastodon, you can follow them. At C-A-S-E-Y-L-I-S-S. So that's Casey Liss. M-A-R-C-O-A-R-M. and T. Marco Armin. S.I.R. A.C. U.S.A. Syracuse. It's accidental. Accidental. They didn't mean to. Accidental. Accidental. Tech podcast. So long. We have one of our favorite segments, which is let's make fun of Tesla and Elon Musk. Tesla is killing off the Model S and the Model X. This is from TechCrunch from a couple of weeks ago. Tesla's ending production of the Model S sedan and Model X SUV. CEO Elon Musk announced Wednesday during the company's quarterly earnings call. The company will make the final versions of both electric vehicles next quarter, he said, adding that his company will offer support for existing Model S and Model X owners, quote, for as long as people have the vehicles. Another quote from Elon. It's time to basically bring the Model S and X programs to an end with an honorable discharge because we're really moving into a future that is based on autonomy, he said. So if you're interested in buying a Model S and X, now would be the time to order it. Sales of both models have flatlined in recent years. Despite interior and exterior refreshes along the way, Tesla has faced increased competition in the luxury EV space from legacy automakers as well as upstarts like Rivian and Lucid Motors. Also, their CEO is a complete piece of sheet, which might have something to do with it. A little bit. Well, I mean, in the case of the S and the X, though, like, what is the S from? 2014? I think 12. 2012, 2014? Like, in the car industry, there are generations of cars, usually numbered, like whatever generation of Honda Accord they're on. And a generation will last multiple years, but not over a decade. Well, they had like a half generational update or two throughout. Right, but they just, like, they were like, I guess we can just keep making the Model S forever, and the answer is no. you need and i know they did like tweaks and they're the current model s is very different from the previous one but not that like it within the car world it is the same generation it is it's i know they've changed so although they've changed so much like they change all these different pieces of the chat but it's like it's kind of ship of theseus like i know they're a different kind of car company but the point is they didn't keep up with their competition so the model s was an inarguably the best electric vehicle ever created when it was produced and for years it was that and then it just didn't keep up with everybody else and still it was really good for a long long time but it just didn't and everyone else got better and better and it just stayed the same and got worse in ways when they got rid of the stocks and all that crap or whatever um and so like oh now now we have to bring it to an end well if you rolled out a new generation of model s maybe your sales would increase but as you pointed out there's there are other factors here, the main one being Elon Musk himself. So whatever. Well, also, like, you know, so what I mean, look, I owned two Model S's and Models S. They for the time, especially they were amazing cars. But that was, you know, my first one was 2016. It was on a lease. Got another one, I guess, 2019 or whatever. They were amazing for the time. And they really did do what what Elon said he was setting out to do early on, which was like advance electric vehicles, basically, like make a big splash and force the rest of the industry to start electrifying their vehicles. They succeeded. They did that. And that was in very large part due to the Model S. But that was a very different time. This is a very different company now, a very different world, a very different competitive landscape. Elon is a very different person now compared to how he was in 2016. He successfully lobbied to get the $7,500 tax credit for electric vehicles eliminated, which I'm sure really helped his company. Good job. Well, they probably already passed it with, like, the allocations. Because, like, it's like once you sold a certain number of vehicles in total as a company, like, yours wouldn't apply anymore. I think they were still getting it in some way. But anyway, if he cares about advancing electric vehicles, getting rid of that was stupid. Yeah. But, I mean, God knows what he cares about these days. He's gone a very different direction. But, you know, when the Model S came out, it was, you know, we had to justify EVs' high prices somehow. and so EVs were all like these high-end luxury cars like the whole reason it's called S is he was ripping off the Mercedes S-Class in terms of like market position because it's like the battery was so expensive you had to make the car like a hundred grand and so it's like okay well how do we sell a hundred grand car we'll make it compete with the Mercedes S-Class or at least we'll position it that way in practice it was you know a fairly inexpensive car interior on top of a very expensive car battery with an amazing drivetrain. They somewhat succeeded with the competition. But what really set the company on fire was kind of two things. The Model 3, which radically changed everything, and Elon promising all this self-driving robo-taxi stuff that has not quite really ever made it and may someday possibly... We'll get to that in a little bit. But now, Now, there's so many EVs in the market now from everybody else, and they're not only the super high-end $100,000 cars anymore. Now they've come down in price, including largely Tesla's own Model 3 and Model Y. And where Tesla has driven their product line was towards volume and towards lower prices, which they do. They're very competitive in that area with the Model 3 and Y compared to the rest of the market. They are very competitive, but what they've done is they've stripped down, they've decontented their cars to bring the price lower and lower and lower. And meanwhile, dealing with their own profitability along the way, and that's been kind of a roller coaster. But like the first Model S that I got was fairly luxurious. It had a lot of luxury features. It had, you know, nice. It was like the highest end Honda interior quality, you know, with a really amazing drivetrain that beats supercars. Like it was very lopsided. But it was a very nice car. Over time, as they've gone with the higher volume models, the models 3 and Y, Those were a lot less nice inside than the S and X are slash were. And then when they redid the Model S to add the stupid half steering wheel and other things, they kind of gave it the same treatment. They kind of decontented it. They took a lot of stuff out. They simplified or whatever, but it made it actually feel like an even cheaper car. over time as tesla has become more about just selling the biggest volume of the cheapest evs that they can they've totally ignored the s and the x and maybe that's because the market has too i don't know the the current model s is not nearly as nice as the old ones uh tesla is not nearly as nice of a car company um they're they're really no longer that luxurious of a car company and certainly elon himself and his politics have have taken a lot of a lot of the damage out on the company and so now it makes sense like as much as it pains me to see that the model s a car i used to love a lot um is gone or it's about to be gone i hardly ever see them on the road anymore the new generation i've i've hardly seen any ever so it does seem like nobody was buying them it does seem like most people were just buying the Model 3 and why. And also Tesla has changed as a company so much that they no longer care to even be a luxury car maker making, you know, $100,000 sedan. So it is a shame, but I think that the era for that car has passed and the era for Tesla to make that car has passed. And now other companies have made a bunch of really good EVs. we no longer need Tesla to address that market, which is good because they're not going to. But now if you want a luxury electric sedan, you have lots of options now, and lots of them are very, very good. So it is the end of an era, but I think that era kind of ended up a while ago. I don't think it's a great business plan to just switch to low-margin products and volume. I mean, that's the reason people have diversified product lines is you want to fleece the rich people with the low volume, high margin fancy products, even though most of the cars you sell are the cheaper ones, right? That's a reasonable model. But yeah, sedans, nobody likes them. Car companies just stopped making them entirely. The Model X should have been great for them, but they had to do the stupid Falcon wing doors. Like SUVs are popular. Big SUVs are popular. Big electric SUVs are popular, right? Why is the X not successful? A, they didn't update it, and B, stupid Falcon wing doors. Yeah, the X was very expensive. It was always a bit of a range sacrifice compared to the S. but it is baffling to me that Tesla still doesn't really make the most popular kind of car in America. Yeah, the Y is as close as they get, which is like a small SUV. Yeah, the Y is in the ballpark, but not that much in the ballpark. But the Y looks like an inflated 3 instead of the 3 looking like a shrunken down SUV. Yeah, they're a little bit old school in that way, which is one of the reasons I like the S. But the X, that was just a fumble. That's their own fault. They fumbled that themselves by demanding those stupid doors. And it's just too weird for people. And so even their entrant in a fairly popular segment, which is like expensive big SUVs for rich people, they've screwed that one up somehow too. And yeah, but if you don't update your products for a long time, it's like not updating the Mac Pro and then complaining nobody buys it. Like there are a lot of things conspiring to be a problem with this. But like the idea that we're only going to sell Model 3s and just go cheaper and cheaper. Like don't forget about the Roadster. Remember that? which was announced. I think I have a note in here somewhere how long ago the Roadster was announced. I'd forgotten. Oh, I'd forgotten about it. Yeah, we do have a note on it later, but it was eight years ago. That presumably will be a very expensive luxury car if they ever ship it as well. So I don't know what to do. We should continue with this story because there's a few other tidbits to talk about Tesla's future in a post-S and X world. Yeah, so their most recent earnings, this is reading from The Verge. In the quarter that ended in December 2025, Tesla reported a 3% decrease in revenue and a staggering 61% decrease in profits over the fourth quarter of 2024. So you want to sell more low-margin products, you say? The earnings report comes a few weeks after Tesla lost its title as the world's best-selling EV company to China's BYD, which sold 2.26 million vehicles last year. Tesla reported selling about 1.6 million vehicles in 2025, an 8.5% decrease year over year. Additionally, from the earnings call transcript, Elon said, But I really think long term, the only vehicles that we'll make will be autonomous vehicles, with the exception of the next generation Roadster, which we're hoping to debut in April. Sure, you mentioned this was revealed in November of 2017. To be clear, I only had one child when the Tesla Roadster was announced. You could have put a deposit down eight years ago on a car that still doesn't exist. Mikayla did not exist in November 2017. She was still cooking. And that's when the Roadster was announced. And to be clear, they didn't just announce it. They showed a car that looked like the Roadster driving away really fast. So they had something on four wheels that moved, but it was obviously not the car that they intended to sell. And I also think it is not going to be inexpensive like the Model 3 if and when they put it out. Also, it's hilarious to hear Elon being like, oh, everything's going to be autonomous, because as we're about to find out, they're not doing well in that department either. So any day it'll be the day. Tesla's own robotaxi data confirms a crash rate at about three times worse than humans. This is Fred Lambert at Electric. New NHTSA crash data combined with Tesla's new disclosure of robo-taxi mileage reveals Tesla's autonomous vehicles are crashing at a rate much higher than human drivers. And that's with a safety monitor in every car. According to a chart in Tesla's Q4 2025 earnings report showing cumulative robo-taxi miles, the robo-taxis experience roughly one crash every 55,000 miles. For comparison, human drivers in the United States average approximately one crash every 500,000 miles, according to NHTSA data. Waymo has logged over 125 million autonomous miles and maintains a crash rate well below human averages. Now, it's worth noting that not everything gets reported. So when I said a moment ago, human drivers do one every 500,000, that's crashes that are reported. And so there's a little bit of like thumb in the wind. Not every robotaxi thing gets reported either. Well, fair. but Fred Lambert did a little thumb in the air thumb in the wind calculations and said yeah basically robotaxis are three times worse than anything else additionally and we'll talk about this in a second as well I don't remember if it was this article or the next one we're about to bring up but they noted as well that whenever Tesla has a robotaxi accident they basically redact everything whereas Waymo's robotaxi or Waymo's whatever autonomous vehicle reports will say the Waymo is traveling at five miles an hour north on first street and then it impacted a pedestrian or whatever the case may be and like has extremely detailed reports whereas teslas are hidden behind oh this is nd8 or whatever the case something like that like corporate secrets blah blah blah yeah they're they're very very shady about it super shady uh waymo safety record reading from the new york times in december of last year uh jonathan slokkin writes when compared with human drivers on the same roads waymo self driving cars were involved in 91 percent fewer serious injury or worse crashes and 80 percent fewer crashes causing any injury if waymo's results are indicative of the broader future of autonomous vehicles we may be on the path to eliminating traffic deaths as a leading cause of mortality in the united states while many see this as a tech story i view it as a public health breakthrough yeah waymo's approach to this is like the exact opposite of tesla tesla has been promising the moon for years and years and failing to deliver and doing stupid things And Waymo has been so incredibly cautious and careful over so many years. Everything has to be totally controlled. Roads mapped down to the smallest millimeter. Just roll everything out slowly. Do everything slowly and carefully because the most important thing is to not hurt people and to not be unsafe. And, yes, they clog traffic in San Francisco and stop emergency vehicles, and it hasn't been all perfect. But, like, I always get the impression that Waymo is trying as hard as they can not to screw up. And I get the opposite impression from Tesla, which is they're trying to say, we're here. We've done it full self-driving. Your car will be making money for you while you sleep. And we'll just keep saying that for the next decade. Please give us money. And by the way, we're taking LiDAR out of the cars because I think that's a good idea, but I'm stupid. So it's not shocking that like RoboTaxi is like he just wants there to be a RoboTaxi and they put a human safe driver in every car. And there's still three times worse than like this is just like this is one of the main things. reasons that I will never buy a Tesla is I do not trust a company with that person at the head of it because every instinct he has and everything he wants his company to do is the opposite of what I want in a thing that I put my life into right I understand like they have at various times been good cars and may be good cars but like his incredible control of that company it filters down into all levels like every story you hear about the terrible fact the terrible racist factories which you know in hindsight should have been a big sign about him um and like the things he makes people do in the assembly lining buying a brand new car and finding parts of it held together with zip ties and the removing lidar and saying cameras are good enough and promising self-driving and it's just like i don't want to be anywhere near that with my life or anyone else's life uh and then waymo like i'm i'm not sure what their path to success is there but everything i read about them is like they're always doing a little bit better than they did before and they're trying to be as safe as possible and i'm glad to hear that their stats show that they are actually exceeding human safety in the in the scenarios where they choose to drive which again are still very limited but like that's appropriate like that they're not saying it'll drive drive your this car will drive you across country while you sleep like waymo doesn't say that because they can't do that and they're being careful and i hate elon musk i will fight you for which one of us hates him more.