0:00 I decided not to like push that for C26 because I want to see what we could do with reflection. 0:04 And what we get is like really cool. 0:07 And part of the reason it's so cool is that like you just have your type, you implement those three operations, like I was saying, if you want a forward iterator. 0:15 You're not inheriting from anything, so there's just the type itself, no base class. 0:18 The base class before was CRTP base, so it was an it was a template, there's no templates. 0:22 Okay. 0:23 And then you open up this constant valve block and you say iterator interface, and then you say what category of iterator it is. 0:29 In this case it'll be forward, and then that's all you have to say. 0:31 And so he said something along the lines of, yeah, man, having the first kid was really difficult. 0:36 I was not ready for the first kid. 0:38 It was like so much work, and it just it felt like it just kept getting worse and worse. 0:43 And then you just kind of trailed off. 0:44 I was like, that's it, that's your advice. 0:45 He was like, oh, well, when she turned seven, it got way better. 0:49 I was like, that so seven years? 0:52 That's what you're telling me. 0:53 Seven years. 0:56 Seven. 0:56 And I was like, 1:07 Welcome to ADSP the podcast, episode 313, recorded on October 28th, 2026. 1:15 My name is Connor, and today with my co-host Bryce, we finish part three of our three-part conversation with Zach Lane. 1:22 In today's episode, we chat about AI, C26 Reflection versus CRTP, having kids, and more. 1:30 Quick announcement before we hop into today's episode. 1:38 Finally, the Pod God Podcast Player is out in beta on Android. 1:46 If you are currently listening to this podcast on Android and you're interested in trying out a new podcast player, feel free to pause now. 1:55 Hit the link in the show notes, which will bring you to the Google Play Store. 1:58 Download the app, then subscribe to ADSP over on Pod God and listen there. 2:06 When do we get that, Bryce? 2:07 When do we get to not deal with your loud team? 2:09 Yeah. 2:10 You gotta leave that in, though, so that her episode, so that the number of episodes that Ramona appears in. 2:17 This is a great transition because so what Bryce is referring to there is that at a certain point, I think it was episode like 267 or something, the platform that I host these podcasts on just added automatically generated transcripts. 2:31 If you want, you can click a button and it it does it for you. 2:33 It's not perfect, but you know, for free and they didn't charge me anything extra for it. 2:37 Very nice. 2:37 But in order to get the backlog of all 260x episodes transcripts, you have to go in and manually click the button. 2:46 So obviously I'm not gonna do that. 2:48 So I got uh I got my codex to go and I like I actually wanted to use computer use, but apparently OpenAI's computer use only works on Windows and Mac, not on Linux. 2:57 And uh so I was very sad about that. 2:59 But then of course Codex was like, yo, I mean, we don't need computer use for you this. 3:02 Just give me your credentials or whatever. 3:04 It actually didn't even say give me your credentials, it just said, I'll give you like a login thing so I can keep some browser state alive, and then like I can just go and hit all 260 whatever's. 3:13 But on top of that, because I got all the transcripts, I didn't perfectly tag every episode with like guest and then the guest name and whatnot. 3:23 But now that I had the transcripts, I basically could go through and say, like, fix all the guest tags and whatnot I've missed. 3:30 But the thing that I came out of that was uh, you know, perfect, you know, episode counts, like appearances and recordings. 3:38 But then on top of that, every once in a while, my wife or Bryce's girlfriend pops on the mic and says something, and the transcript captures that. 3:45 Like I had to go label it myself. 3:47 But like, so not only does if you've ever been, and actually I forgot to uh that reminds me. 3:51 So if you've ever been a guest, you have now a tag and it ranks you by like the number of episodes you've been on and the number of recordings. 3:59 So I've mentioned Zach, you were number six, but after this, you're probably gonna pop up to number four, number three. 4:04 And uh, but because of that, Ramona, Bryce's girlfriend, and Shima, my wife, also now have tags. 4:11 And uh I think they break like the top 15 because they you know they each have like three or four. 4:16 So now that's what Bryce is saying is make sure you leave it in there so Ramona's count goes up. 4:20 And I think actually Bryce's mom makes an appearance once. 4:23 Uh speaking of which, I did that all with AI, and we asked you earlier, but I I cut Bryce off, or more accurately, Zach off before you could respond. 4:32 We don't need to make this a full episode, although we can if you want, a little a little digression. 4:36 The name of this episode should be Old Greybeard Yells About AI. 4:41 Listen, back in my day, we didn't have no AIs. 4:44 Um, so look like I mean, this is the thing. 4:47 Like, I feel like the AI stuff is an incredibly useful tool, and I've used it a lot for like uh Google searches about especially stuff like I was working on like a just a for fun like game project for a while, and I was using the Unreal engine, and you know, Unreal is kind of its own world, right? 5:03 So it's all this like domain expertise that's incredibly widespread. 5:06 Everyone knows it who does Unreal stuff, but as a new person to Unreal, I didn't know any of this stuff. 5:11 And the AI searches were great for that. 5:13 It's like how do I do this? 5:14 It's like here's the class or function or you know, set of things you need to know to do that particular thing, and then so it's incredibly useful of that kind of stuff. 5:22 It's made search great, it's even given me like a lot of really simple Python scripts and stuff, but ultimately, like if it's something in C, I want to do it myself because that's why I do this, is because it's fun, right? 5:33 I mean, I don't like you know, I didn't get in this field for the pay, like the pay is great, but like I that's not what motivates me. 5:39 I'm motivated because it's like a fun thing to do with my time, you know. 5:42 And AI takes all that away, like it becomes like everything's just a code review, and I don't want to just code review some, I don't know, kind of half-assed intern's work, you know, like like yeah, I the the thing is I I find it it actually rewarding to to code review an intern's work because then you're actually teaching someone something and they get better, but the the AI is kind of static. 6:03 I mean the met the model gets better, but not because of something I did, right? 6:05 So so I I uh I tend not to use them and eventually it's gonna be such a problem that I don't that I'm gonna have to start. 6:13 But uh for now I'm just like doing things myself because it's way more fun out of 6:16 it. 6:16 So that was gonna be my follow-up question because we've mentioned on this podcast before an elusive unnamed individual that works at NVIDIA, and when we asked, we actually asked them to come on the podcast, and I'm gonna use them to keep the the gender a secret so we give no revealing details about this individual, but they they said that if at any point they're forced to like use these tools, they would just retire because 6:41 did they say that they w that they were they said no when we asked about if they wanted to come on the podcast? 6:45 Correct, yeah. 6:46 Okay, I just want to say to that person, you would be such a good guest. 6:50 I I would love so much to have you on. 6:52 I think you'd be a really good guest. 6:53 There's a lot of fun stuff we could talk about, not just AI. 6:56 There's all sorts of other fun things that you do that would be fun to talk about. 6:59 Well, I mean, but like we would love to talk, because that's the thing, is we uh between Bryce and I, we don't get a lot of contrarian takes because we're like drinking the Kool-Aid or whatever, you know, uh thing thing that you want to say. 7:09 Anyways, it's it's 7:11 unlimited tokens now, buddy. 7:12 You better be nice to me. 7:13 I'll give you some unlimited tokens. 7:15 For for the purposes of most people's work, you have unlimited tokens at NVIDIA, as long as you're not spending like trillions of tokens, basically. 7:23 Like I don't hit like, you know, at one point when I was doing the recursive self-improvement stuff, I would, you know, my token spend would go up. 7:30 But if you're just doing you know regular programming work, you're never gonna come close. 7:34 Anyways, the point being is that this individual said they would retire before they would basically do this kind of let the AI write all your code, you review stuff, and then that's it. 7:45 So you're saying that you're avoiding that, but it you just mentioned at some point it'll become a problem, but you didn't mention retiring. 7:51 So I'm curious, like yeah, you're so you you definitely, you know, and that that's the thing, is there's a ton of folks that I talk to that basically say this kind of thing, is that they love the the craft of writing code and whatnot, and that is to some extent going away. 8:05 So I'm interested, what do you expound on uh you know how you feel about that? 8:10 Because we haven't had a lot of folks on this podcast talking to that, like you know, uh, you know, being in the industry and watching the shift while trying to still write the code yourself. 8:20 And like, do you think that you're gonna have to is is it gonna become like necessary, or do you think you'll probably be able to work the rest of your career writing code and it won't become a problem? 8:29 Well, I mean, you know, I don't have that many years left of working as a professional programmer, probably. 8:35 Like, I mean, maybe ten, but you know, maybe I'll keep going. 8:38 I mean, you know, certainly like Beeman went until he was in his mid-80s or late eighties or something, like still contributing stuff. 8:46 So, you know, maybe I'll keep going for for a long time. 8:49 But but I think I can probably dodge it for ten years, but maybe not. 8:53 But but the thing is that you know, there's also this aspect that like so when I first started programming professionally was the end of 2002, I was still a student. 9:00 For last semester I was in school, I was working as kind of a part-time thing, and then I transitioned to work at that same job uh full-time, and that was my job for almost nine years. 9:10 And in that nine years, so if you think of like end of 2002 to like beginning of 2011 when I worked there, you know, Herb wrote the no no more free lunch article, and then all of a sudden everything became multi-core and concurrent, or multi-core and parallel, or both. 9:27 And so I really had to change how I thought about programming. 9:30 I like, you know, literally I had a conversation with someone at work in that in that earlier part of that era where someone was like, Oh, this is not fast enough, it needs to be at least twice as fast, and someone else said, like, well then leave it be because in 18 months we'll respec the computer that it's supposed to run on, it'll be fine. 9:46 Because in 18 months the computer will be twice as fast. 9:48 Like, well, why are we talking about this? 9:49 This is pointless. 9:50 And that attitude, like that that you don't need to care about cycles, you need to care about product programmer productivity that led to things like, you know, Java being so popular for such a long time, and and other things like that in that era when I first started programming. 10:02 So to me, programming was all about the like mathy parts, the s the you know problem solving parts and and that sort of thing. 10:09 And the low-level stuff was something I had to kind of put up with to get to the fun mathy bits, right? 10:13 And um and then over time it became more and more important to do everything as efficiently as you can. 10:19 And now basically everyone who uses C ⁇ still does it because a very it's it's a language you have a large amount of control over where the efficiency is, right? 10:28 Like if something's inefficient, you can just change it in a way that you can't in an interpreted language and in certain certain other higher level even compiled languages. 10:36 And so, like, you know, I was kind of dragged into that kicking and screaming, but then I I started really enjoying like the optimization problems that I that I had to deal with, and now I work in stuff that's very low level, and I'm very comfortable with it. 10:47 Like today I was working on like wire protocol stuff, right? 10:50 So like very low-level stuff. 10:51 So, you know, the idea that like I would just get habituated to using the new AI stuff, and you know, I2 welcome our new AI overlords, that that's perfectly reasonable that that's the outcome, right? 11:03 But for right now, I haven't used it very much, and the stuff that I've used has also been probably a little bit out of date and kind of crappy compared to the latest Frontier models that are sounding like way, way better. 11:14 And certainly like, I mean, I've seen how you know people like Barry are getting really interesting stuff done with it. 11:20 So I think probably it's in my future. 11:22 I just haven't sat down and sunk the time into it that I would need to really, you know, come up with a workflow that works for me and and kind of understand how everything works and so forth. 11:31 I have a very kind of cursory understanding of of how it works, and like I said, it's probably a little bit out of date as well. 11:35 But I think it's going to become inevitable. 11:37 Like, you know, we have a lot of stuff in just about every workplace these days saying like, hey, you know, this is the latest model that we've got, go use it. 11:46 It's very important for productivity, everybody starts using these things. 11:48 You try it at least once a week. 11:50 These kind of like signals from from upper management. 11:52 And the the only thing that I think is is maybe not going to leave it inevitable is if people cannot find out how to get these models, the operational cost of these models down, because right now they're they're just hemorrhaging money. 12:04 And if they can't stop that, if they can't find a way to optimize that, then maybe this is actually not going to be sustainable for most companies. 12:11 And there's going to be a few companies where you have like one team with unlimited tokens and everyone else doesn't get very many. 12:17 And so it's not a part of everyone's workflow. 12:19 But if it becomes part of everyone's day-to-day workflow, it's just going to be something I have to adapt to, I think, with everybody else. 12:24 Hey, and that's why Brest is going to work on a company to focus that focuses on bringing those costs down. 12:29 Yeah, yeah. 12:30 It seems like that's a big, big, big deal. 12:33 Yeah, I think I think costs need to come down by like 10x to 100x, and I think that they will. 12:37 I think that that's j just naturally because there'll be new hardware will come out that'll be more efficient. 12:42 There's there's still a few more, you know, rounds of of efficiency gains just for new chips that that'll come out. 12:48 And then like we'll just get more efficient in how we design in different systems. 12:52 And yeah, I think the cost will come down. 12:54 And also like more open models will get on the market, and that's gonna create a a race to the bottom in terms of token pricing. 13:01 Yeah, yeah, that's definitely true. 13:03 And then like also like more capacity eventually becomes available. 13:05 Like every every time token prices have have gone down, the issue is every time the token prices go down, it usually goes way up. 13:12 And so that's gonna continue. 13:13 That's gonna continue. 13:14 That's true. 13:15 But but if people can't keep building out capacity, because right now the only thing that unites people politically in the United States is that they hate the idea of data centers being built near them. 13:22 Right. 13:23 Well that well, that's the thing. 13:28 So like we have tons of data centers all over the country and they haven't been a problem because they've been building them at a smaller scale. 13:34 But now these things are such large scale that you cannot ignore them. 13:38 And so maybe the solution is to build a bunch of smaller ones. 13:41 But for for training, you need big data centers. 13:44 For inference, it's fine to have small data centers. 13:47 And in fact, for inference, the main thing you care about is latency. 13:50 And so you may actually want smaller data centers that are more localized to people. 13:56 Um, yeah, absolutely. 13:57 But but 13:58 what I'm saying is like if if if the political reality becomes every politician gets re-elected by saying, I promise no new data centers, then like you don't have a choice to build the big ones. 14:06 You have to like build smaller ones and like have them talk to each other in some way because that's just a political reality. 14:12 Right. 14:12 Yeah, but and I I I think that smaller data centers is like a combination of smaller data centers and bigger data centers is almost certainly the future that we'll see more localized and and more like edge type computing for for various inference applications. 14:28 Yeah, that's what I mean. 14:28 Like I think most everyday folks don't realize that's why Google is fast, is that there's a data center like really close to you at all times when you do a Google search, you're searching that, you know. 14:37 And that I'm sure the same thing is true of like trying to get latency down in in AI stuff. 14:42 So Zach, are you do you ever come up to New York now that you're working at Elseg? 14:46 No, no, never. 14:47 I mean, there is definitely well, there's one per person on my team that works uh, you know, Eddie Nolan, he works in the office up there, but like no one else is anywhere near there, and like like the the only thing that we sort of go to that like we have people from all over the company attending is like CPP on C because it's near the London office and there's way more people in that area. 15:07 So yeah, I don't really have any reason to go to New York. 15:09 I was gonna berate you for not not not coming to visit me, but uh 15:13 No. 15:13 No, I mean you know how it is. 15:14 I usually look people up when I'm when I'm in town, but uh yeah, I I don't really travel much anymore. 15:18 Like I don't need to, I'm fully remote and we I I joined in a narrow window when they had everybody doing remote. 15:24 Now it's it's forbidden. 15:25 But we're grandfathered in. 15:26 I'm gonna have to run because Ramona came in here to uh tell me that she's very hungry and wants to eat, and I I told her sh I told her I'd like to have dinner with her, and then she said uh, well, how long is this gonna take? 15:37 And I told her ten minutes, and that was about six minutes ago. 15:40 So 15:40 Alright, okay, so let me just before you go then, one one thing that I just want to talk about because I've been working on it lately and it's super exciting. 15:46 Okay, and that is like this thing I was talking about before, which is like um the iterator interface thing that used to be a C CRTP template, right? 15:54 So if you want to have, let's say, a forward iterator, you just implement equality, dereference, and then pre-increment, and then you you inherit from the CRTP base that like fills in all the other stuff you need for a forward iterator. 16:06 It's very nice, but I decided not to like push that for C26 because I want to see what we could do with reflection. 16:12 And what we get is like really cool, and part of the reason it's so cool is that like you just have your type, you implement those three operations, like I was saying, if you want a forward iterator, you're not inheriting from anything, so there's just the type itself, no base class. 16:25 The base class before was CRTP base, so it was an it was a template, there's no templates. 16:29 Okay. 16:30 And then you open up this constant valve block and you say iterator interface, and then you say what category of iterator it is, in this case it'll be forward, and then that's all you have to say. 16:38 Because now, since you can see the code in the class, you don't need to know what the what the reference type is, you just look at what the dereference operator does. 16:46 You don't need to know like if it has an equality operator because you can see if it has it or not. 16:50 So you can not only like deduce the types of the type depths you used to have to specify as template uh arguments, but now you can also check like, oh well you said you're a forward iterator, but you didn't implement dereference, let's say. 17:02 Like it can check that you actually implemented the right operations for the thing you're you're doing. 17:06 And um like the fact that there's no template anywhere and that like I can actually use the end definition of what the class has in it, instead of just like being given this thing, which is an incomplete type that I'm basing everything off of and using template magic to like deduce whether or not this operation makes sense much later on. 17:26 Like there's it's gonna be once you generate the code, much faster to use this thing in all the places where you use the the iterator. 17:33 Because you've you don't have to instantiate templates anywhere, right? 17:36 Like you just stamp out the code and there's no templates anywhere. 17:39 It's like crazy. 17:40 It's it works so well. 17:42 And um and like I said, like Barry's been doing all this work. 17:46 I'm sure those other folks that are involved in those papers are doing a bunch of work too, but like he's the one I've been talking to in specifics, so I know the work he's doing, and it's like I think the version that we're gonna get in 29 is gonna be like just as much of a game changer as the stuff we got in 26. 17:59 It's really, really cool stuff. 18:00 It's very exciting and very fun to work with. 18:02 And and I would say like the biggest difference between doing this and doing template fuckery is that it's easy. 18:08 Like if you make a mistake, you can figure it out pretty simply, and it's just like straight line regular code. 18:13 So if you make a mistake, you could be like, oh, this if has the wrong condition. 18:16 That's kind of that's the kind of thing you're debugging instead of like, oh, which overload matches the best, or is this a specialization that gets picked up or isn't it? 18:24 Like that kind of thing. 18:25 Is this uh because it sounds like you're building this on stuff that Barry's you know giving you a LLVM branch of in order to do the so does this live anywhere on GitHub or uh or like if people want to play with this, can we point them at something? 18:39 Yeah, so if you go to Barry's GitHub, which I think is well, let me just look it up real quick. 18:45 It's possible that Barry is actually a frontier LLM uh just uh you know the jump trading has had for many years. 18:53 Just just based on the volume of output that he seems it seems like we should 18:56 put Barry back on, because we had Barry on, I don't know, how long? 18:59 Oh my god, we 19:00 did, I remember. 19:01 That was like But 19:02 yeah, we only had him on once. 19:04 That was when we were young. 19:05 That was when we were young. 19:07 Were we young when we started this? 19:08 I don't know if we were young. 19:09 We were young 19:10 when we started this, Connor. 19:11 That it was we were definitely young. 19:13 We were not old when we started this, and we're old now. 19:16 Alright, I'm gonna I I'm gonna have to run cheese. 19:18 Alright, well, 19:19 Zach and I will keep chatting for a bit, but uh 19:22 bye Bryce. 19:23 So if 19:24 you go if you go to Brevzen as his GitHub handle, slash LVM dash project, it's just like the regular LVM project, but it's a little bit out of date because he's been making changes to it and not keeping up with all the all the latest commits. 19:38 And then he's got a branch on there that says what's it? 19:40 The exact branch is something like oh it's compiler dash explorer slash berry. 19:45 And so that is, if you go on compiler explorer, that is the version that's up there. 19:50 And so if you want to play with it on compiler explorer live, this is the thing that it comes from. 19:56 But I'm I'm doing a thing where like I'm making lots of changes locally, so I need to be able to compile locally to, you know, it wouldn't make sense because there's so many more files than just the one I'm working on that I would need to be in place for me to try to do it on compiler explorer. 20:09 But yeah, uh it's really nice. 20:11 It's like r it's really a challenge to configure LVM. 20:14 It took me about two days of of evenings like trying to get it com configured, but once I get it configured and built, like it's been amazing to work with. 20:22 Okay, yeah. 20:22 We will definitely we'll put Barry on our list of folks to uh to chat with and I will uh track these links down and uh I'm not sure if you have too. 20:29 Like if you said that this branch is living in compiler explorer, if you've got some like uh what do you call it, demo example that you want people to land on. 20:38 Oh, yeah, no, definitely not. 20:40 I mean I eventually I want to do that, but for right now, it's um you know it's part of a boost library that I'm adding some I'm adding a d a new implementation to. 20:48 And so it's like a ton, there's a whole tree of code that it requires, and that's the kind of thing that's like really hard to put on get on in on Internet Explorer. 20:56 I mean uh compile explorer. 20:57 Yeah. 20:57 I know they have they technically have the the branches of compilers and they have the libraries, but it sounds like this is uh unreleased boost library, so it's not as not as simple like uh but down the road, it sounds like definitely people will be able to uh play with this stuff. 21:12 Yeah, yeah. 21:12 I think that's one thing that's kind of missing from from compiler explorer that I mean I understand it's very hard to to implement, so I'm not I'm you know, no shade on them for not having this, but like it would be really cool if you could say like here is a tree on uh on GitHub, please like pull in any include that's rooted at this tree, and then like all the relative links inside there will assume that they're coming from the tree itself, if that makes sense. 21:38 Right? 21:38 But like, you know, that requires maybe you to modify your compiler front end or something like that for it to to understand that like you know it's missing a dash I for basically the the part of the tree that you care about. 21:49 So it's it's not an easy problem at all, but like the fact that you can't do that means that like you kind of have to jam everything in one file to to make compiler explore examples. 21:58 And it's great for like small Medium sized things, but if you get to a big enough thing, it just kind of becomes unwieldy and it is what it 22:04 is. 22:04 Yeah, 22:04 I was gonna say I 22:05 know 22:05 Compiler Explorer supports like raw single file uh like HDCs or whatever, but 22:10 yeah, and that's super useful. 22:12 Yeah. 22:12 Okay. 22:13 Well, we'd be sure to link this in uh in the show notes or description of however you're listening to this. 22:18 And uh I will say thank you so much for coming on, seeing as Bryce dipped. 22:23 Um be sure to check these show notes either in your podcast app or at adspthepodcast.com for links to anything we mentioned in today's episode, as well as a link to a GitHub discussion where you can leave thoughts, comments, and questions. 22:36 Thanks for listening. 22:37 We hope you enjoyed and have a great day. 22:39 Low quality, high quality, that is the tagline of our podcast. 22:45 That's not the tagline. 22:46 Our tagline is chaos with sprinkles of information. 22:49 I even you know, will I leave this in? 22:52 But at one point we, you know, Bryce was wearing uh we'll call it a tank top. 22:56 I think that's the politically correct term for for that thing. 22:59 But he was moving around a lot, and let's just say he was exposed. 23:04 And 23:05 there was too much pepperoni on the pizza. 23:09 And he noticed at some point, but uh before that I screenshotted it, and now I've got the best blackmail ever. 23:16 Uh if if something, you know, something happens and I need to blackmail Bryce. 23:22 Um, you know, I have I have the goods. 23:24 No. 23:24 Anyways, we'll have to have you back, and uh maybe we'll do it for I'm not sure if we're gonna do a holiday special this year, but those were I've always wanted to do a like uh holiday special game show where I'm the host and uh I've talked to I've talked to Ben about this, Ben Dean, where like uh you know they have the countdown like British game shows, but anyways, where I just have some like you know absolutely like uh chaotic mayhem and I just like throw out categories and uh and then I get a few people. 23:52 So you know, Bryce would be there, you'd be there, maybe get Tony, you know, some fun people from uh that we know from conferences and and uh the C. 23:59 And uh but that takes a lot of upfront work, which is something we're not great here at ADSP. 24:07 We're good at emailing people sometimes two days before we want to record, and uh which is what happened with you. 24:14 I think we we pinged you, hey, you free on Wednesday, but you just hadn't checked your email in a while. 24:19 So 24:20 Yeah, yeah. 24:20 Yeah, that email, like it is so frustrating that they cut off Gmail at work. 24:24 It's like I didn't realize that was the only time I really checked my email until they cut it off, and then I never respond to emails now, and it's really quite bad. 24:31 It's 24:31 crazy how defaults work like that. 24:33 I I turned off Instagram notifications and removed it from my home screen, and now I just never I just realized it's one of those like things where you get a notification and then you end up in this like you know, what do you call it, black hole of reels and uh and stuff, and I don't actually really care about any of that stuff, but if if it's if you're getting notified, that's a whole other I gotta talk I gotta talk to people about that, is that I've realized this like the access that we give corporations to just ping us, like 24:57 it's insane. 24:57 Yeah, it's crazy. 24:58 It's crazy, yeah. 24:59 Like, because I've been thinking I'm a little bit out there with my ideas, and like, you know, we're we're on the precipice of like augmenting physically and whatnot. 25:07 And so I've been thinking about like what it would be like to have like a little you know neural enhancement. 25:11 But like, could you imagine like insert any large corporation with like a social media platform and they have a device and then like you put it and now in your retina or whatever you're getting pinged? 25:22 Like that is literally the most dystopian thing I could ever think of. 25:26 Yet we already kind of do it with ourselves. 25:28 Um yeah, I think it's almost inevitable. 25:30 I mean you 25:30 know 25:30 there's a black mirror about it. 25:31 Yeah, yeah, yeah. 25:32 That's what I was thinking about is that they live in the box, and then like you know, they they if they close their eyes, you know, the the the sirens start going off. 25:38 But it's like 25:39 Well, I was I was thinking of the one where like there there's someone who dies, right? 25:43 And then or sorry, did they die or was it it's either they died or they had such bad spinal damage that they were gonna be paralyzed, but either way, like I think it was they actually died, and so what they could do is they could repair enough of their brain, they basically put part of their brain's functionality in the cloud, right? 25:59 And then these people couldn't keep up with the payments, they're like, That's right, we'll just put you on a lower tier. 26:03 Lower tier was like they would just freeze and like say an ad in the middle of talking, you know, it's just like the most horrible thing ever. 26:12 It's so it's so brutal. 26:13 But that's that's what's crazy is that like a this mental exercise of being like, oh, what would that be like? 26:17 And then realizing that like these cell phones that we carry around with us that are like the primary way we communicate, like you know, they're already inundated with this stuff. 26:25 The defaults for notifications are all on. 26:27 It's like when you think about this as like an exo kind of augmentation, it's just like I'm like, why you know my hygiene of notifications is terrible. 26:37 And so now I've just started like all notifications. 26:40 I don't need a LinkedIn notification on my Garmin while I'm going for a run, like enjoying nature. 26:44 Like it's some person I don't know who I've never met in my life, is adding me as a connection, and they have access to my wrist. 26:50 That's insane. 26:51 Like yeah, 26:53 it is really wild. 26:54 Yeah, and 26:55 like we don't, we don't there's no I've never heard on any tech podcast a conversation about this, like the absurdity of like what we're giving these corporations. 27:02 Like, I you know, I'm not sure LinkedIn's not or you know, is it an evil corporation? 27:06 Nobody knows, you know. 27:07 Uh but like I mean, I'm not sure they're trying not to like ruin this aspect of my life while I'm out on a run. 27:13 But you know, they kind of are a little bit. 27:15 And so now no more notifications for LinkedIn. 27:17 And uh Instagram's just basically completely gone. 27:20 So and I guess I mean Gmail sounds like it would have been useful to have access to your Gmail though, because um 27:26 Yeah, and that's the thing is I think I turned off, I think I used to have notifications on for for Gmail on my phone, and then I just got you know, there's so many automated ones that I get from like things I'm subscribed to. 27:37 Right. 27:38 That I was just like, this is intolerable, I have to turn this off. 27:40 Yeah, and so then and and so now I don't like I probably if I went through the effort of like setting up a whole bunch of filters and getting stuff that was actually for me and not anything automated and so on, then it you know I could have the notifications turned on, but like I just I just try to look at it more frequently, and sometimes I do, sometimes I don't. 27:58 But 27:58 that's that I mean that honestly, like, and the AI has caused me like these kind of like we're the precipice of like technology changing has made me think about all this stuff. 28:06 But like that's another example of like email, and I I've mentioned this guy a couple times, and I always get in hot water when I mention his name, DHH. 28:12 And anyways, one of his companies is uh Hey Email, where he he always talks about how email is just terrible, and we give everyone access to it if they have it. 28:20 But like I guess the way his service works is that you have to like allow an email into like your garden, basically, so everything is blocked by default, but over time you can let people in. 28:31 But I've realized that like a part of the problem is that you need your airplane ticket when you buy it from like you know, Air Canada, and it's just that like 99% of the time they're just badgering you with like, oh, here's a vacation, here's a deal, here's a thing. 28:43 And it's like, well, I need my ticket that one time, and so like, but it comes from the same like you know, no reply at aircanada.com or whatever. 28:52 Right. 28:53 And uh, anyways, there's gotta be fixes for this stuff in the future, but 28:57 I don't think there are. 28:58 I mean, that's why this insidification thing has caught on so much, is that like when you you know, so the so the early internet was mostly just people just hobbyists, right? 29:07 Like I saw so many amazing, crazy things on the internet in the early days, like you know, 95 era, right? 29:14 Before like all the e-commerce stuff took off in 97 was when that started. 29:19 And then everyone who was on the internet, they were just like, you know, some weird, you know, graduate student who really loved like the Church of the Subgenius or some insane you know, insane countercultural thing like this. 29:29 And so they had some they had some uh crazy website that was just fascinating. 29:33 Like, I don't know what uh most of this means, but it's like someone spent like hours and hours of their life putting this together, and I find that fascinating. 29:40 And then and now everything is just about like retaining viewership, and they the content is literally just that, it's content, it's just something that they're gonna serve to you, and they're gonna tweak which exact things are served to you until they find a recipe that gets you to stick to that app longer, and then that's all it's about. 29:57 And so you get served stuff that is almost never interesting, but is like just interesting enough that you wanna look at it, right? 30:05 Which is like totally different from like the kind of stuff I would stumble upon on the old internet and just be like falling down this rabbit hole of like just insane counterculture or just subculture things that I didn't know anything about, and that was fascinating to me. 30:19 And ultimately it's because you know the when when you drive everything by commerce instead of by just people being social, then that's how you end up with the social media landscape we have now, right? 30:28 I I think it's ultimately inevitable. 30:30 Um there's no fix. 30:32 That's it, because that's the thing, is that there's certain places like the YouTube algorithm, you know, there's many problems with YouTube, but I I gotta say, the YouTube algorithm is phenomenal. 30:40 Like it'll serve one time it served me a 2014 Strange Loop video, which is like a conference that's like now, and anyways, it's like quasi changed the direction of like my intellectual interests, and like just in the last couple months, it served me a video on one of my favorite runners, Grant Fisher, that at the time had like a couple hundred views on it, and somehow YouTube just knows that my favorite runner is Grant Fisher, and like this one YouTube channel made a 50-minute like documentary on them, and it was like the best YouTube video I've ever seen. 31:11 Anyway, so when it's like done, and don't get me wrong, YouTube's trying to make money, but like when it's done right, like there's a balance of whatever, it's amazing. 31:18 But like so many places now I feel like TikTok and Instagram, you go there, and you just know that like your brain is being like programmatically like you know, screwed with, and you get off of it like an hour later when you should have just been listening to an audiobook or reading, heaven forbid, a physical book before you went to sleep. 31:35 But instead, you were asking yourself, are these raccoons dramping jumping on the trampoline that's been caught by this like security camera at night? 31:42 Is this AI or is this real? 31:43 Because you know, this it's pretty cute if these raccoons jumping up and down are real, you know. 31:48 But I just I don't know. 31:48 And uh is you think that we're it's just inevitable in a capitalist system or like there's no fix for this? 31:54 Well, I th I think I do. 31:55 I mean I mean I think it is because like you know, ultimately, like so let's say you find some service that that works really well for you, right? 32:02 Like in this case, it's it's YouTube. 32:03 Like YouTube works well for me, I think, in the same way it does for you. 32:06 I do discover things by the via the recommendations that are that are I think good, but what I find is the the recommendations that are much better are when it does do one of these recommendations for me, and I find someone that then kind of cross-promotes other people that do similar things to what they do, and then those people, I'm like, oh wow, this is really interesting. 32:23 Whereas the vast majority of the things it recommends to me are just like, you know, very vanilla, very kind of like modest moderately in uh engaging, but like I'm not that interested in it. 32:35 I watched the video one time, I never want to see it again. 32:37 And then there's like maybe five or ten percent that are like really stuff that I I'm I'm glad that it showed me. 32:43 And then my way of finding other things like that is almost never through the algorithm itself. 32:46 So I think even really good exemplars of it are not that good. 32:50 Yeah. 32:50 And if you had like a a recommendation engine that was designed to not increase engagement, but like, you know, there's this thing that like um I don't know if Pandora still does this, but it definitely started out as like a pretty decent recommendation engine for music that used a lot of like particulars of how music is uh is uh perceived. 33:09 And I think it worked pretty well for a long time. 33:11 And I I stopped using that service a long time ago, so I don't I don't know if it kept up with that. 33:14 But that was specifically designed towards like the way we get people to use this is we we advertise that it's gonna pick stuff that you like based on stuff you already like, and to the extent that it works that way, people use the service. 33:26 And maybe it stopped working that way, and that's why I stopped using it, but um but like I think that like even if you have something that starts out that way, I think it's inevitable that if it doesn't, you know, you're you're sort of what what do they call it, the the the TAM or whatever? 33:41 What's that was a business term for your like total total addressable market? 33:44 Addressable market. 33:44 Yeah, 33:45 addressable market, yeah. 33:46 So if if the number of users can't go up, then the amount that you get out of each user has to to maintain growth, right? 33:52 And so ultimately you're gonna end up having intense pressure on the developers of those algorithms to like make them stickier. 33:59 And the way you make stuff stickier 34:00 is 34:00 like, you know, Rage Bait and things like this that are that are not like they're not fulfilling in terms of a consumer of that. 34:05 It it just makes you angry and makes you want to engage because of the anger instead of like you know, like something that you're getting out of it. 34:11 What what I've found is like the you know, I used to kind of like fall down Reddit, you know, rabbit holes of like not even rabbit holes wrong term, I would 34:20 just 34:20 like look at Reddit for way too long, right? 34:21 And um what I find is that as long as I'm currently reading a book on the same device, like I I use the Kindle app for books, so as long as I'm currently reading a book on the same device, I will just uh I will like I'll get one of these notifications from Reddit or I'll get a link from some CPP Reddit thing that's going on from someone else. 34:38 I'll follow it, and I'll look at that for a few minutes, and that is something that I want to see. 34:42 And then as soon as I kind of like swipe out of that and I'm left with the other Reddit, I'm like, oh, what's that? 34:48 And I click on it and like you know, the the next shiny thing will engage me. 34:51 But what I find is that it'll only do it for like two or three iterations. 34:55 I'm like, what am I doing? 34:55 And then I'll just like literally tab over to the book that I'm reading, and that that that kind of disengages me from that loop. 35:01 And I end up reading an actual novel for way longer than I read the stuff on Reddit. 35:06 But but without that, like in the loop somewhere, I do get stuck. 35:10 Fortunately, Reddit is the only version of that. 35:12 I don't I don't get stuck on Instagram or whatever the others are. 35:15 But yeah, I can see how people do, and it's really interesting. 35:18 I wonder if like I I kind of have this pet theory that you know in the in this way that like many generations always rebel against the the previous generation's norms. 35:27 I really wonder if if like Gen Z later on in life or maybe Gen Alpha from the beginning are gonna 35:33 be 35:33 like, yeah, these apps are garbage. 35:34 Yeah. 35:34 Like 35:35 I don't care that you lived your entire life on them, I'm not gonna do that because like I I don't wanna I don't wanna be that way. 35:40 And we're gonna hopefully it'll be like a you know two or three generation, you know, bump that that people are so completely owned by their devices and not the other way around. 35:51 Yeah. 35:52 Yeah, it's I don't know. 35:53 I find myself in the last like year of my life, ever since AI is that I've I've started thinking like what where society's going, and a lot of sometimes I get very sad because I'm just like, you know, I'm gonna I I I've I've talked on previous episodes of like the potential like death of communities, because you know, if if if if you're you know say what you will about coding and whatnot, but like I think one of the most important things is like you get together with people, and especially you know, you're working remotely, I'm working remotely. 36:20 A lot of people work remotely these days, like going to conferences is like one of the highlights of my professional like year, you know, it's you get to meet people in person, and like I come away so energized, even if I'm exact absolutely exhausted. 36:33 And it's like, well, what if like people aren't writing code? 36:37 And then like, you know, like you have these constantly running autonomous systems that are like self-healing, and like it's like, okay, well, you know, most half the podcasts I listen to are all about writing code and stuff, and it's like, well, are those gonna all evaporate? 36:50 You know, and people have criticized ADSP that we don't talk about algorithms as much anymore, and we talk more about high-level stuff, and anyways, and then uh yeah, just thinking of like uh what is it, anthropics about the IPO and they're a public benefit corporation, and and then like just when you I hear you talking about the you know the app addiction and whatnot, I'm just like, you know, wow, how do we get here? 37:09 You know, it's like uh people used to have like you know houses filled with multi-generational, like you got grandma and blah blah blah, and it's a it's an insane amount of more work like raising a kid. 37:20 Yeah, it really is. 37:21 And like it makes sense to have a multi-generational home when you've got like aunts and uncles and grandparents that are constantly like it's so exhausting when you are just handing between two people fussy baby that that that's just how babies are. 37:36 They're just fussy. 37:37 And uh it's you know, you get to 3 p.m. 37:39 in the afternoon and you're just like, Man, it is uh it's a lot of work. 37:44 I get why people used to live closer to home and like have like a but like that's a lot rarer these days, you know. 37:50 Yeah, yeah. 37:50 Yeah, I mean in in North America it's almost unheard of. 37:53 Yeah, right. 37:54 I mean, it's like w very much the minority case, right? 37:57 Yeah. 37:58 So I I will tell you like the something that no one ever told me there's two things no one ever told me about parenthood that I wish someone had told me. 38:04 Perfect, love it. 38:04 One is look that you know, you hear all kinds of advice from people, solicited or not, there's books and stuff. 38:09 These are two things that never heard from anybody. 38:12 One is you're gonna clean shit out of a bathtub on a semi-regular basis. 38:16 When the baby gets in the bath, it's nice and warm and they just let go. 38:19 And so, like I never thought that would be my life, and I did it multiple times, and so that's something no one ever tells you. 38:25 But the other one is when you have a when you have one kid, it's like a lot of work when they're really little, and then after a while it it kind of settles in, and then it's like having it's like each of each of you, you and your partner, you both have a part-time job that you added to your schedule, right? 38:39 Like you can you can do things you used to do, but you just can't do as many of them. 38:42 You have less time in your week, right? 38:44 Yeah. 38:44 When 38:44 you have your second kid, your old life is over. 38:46 That's when no one tells you. 38:48 It's really you have the second kid that there's no doing the things you used to do. 38:51 There's just there's just the kids. 38:53 Like, that's it. 38:54 Like it it's it's bizarre. 38:55 It's like the step function from one kid to two is so nonlinear. 39:01 It's like a part-time job to like now I'm working three jobs. 39:04 It doesn't make any sense. 39:05 And I've heard people say that the third kid is kind of like that as well, and then after that it doesn't matter because as soon as you can't play man-to-man defense and you have to play zone defense, it doesn't matter. 39:14 Like you can't, you can't eat there's there's too many kids, they outnumber you. 39:17 And so three and more is all the same, is what I've heard. 39:20 That's actually I've star I've I first of all, thank you. 39:23 That is great. 39:23 This is definitely staying in the podcast, whether the ADSP listeners care or not. 39:28 Uh but I I've actually tried to tell my wife that even though it's a lot of work right now, like this is actually it's unique in that we don't have a second kid yet, right? 39:38 Like, yeah. 39:38 So when the when we do finally get uh baby to sleep, we can go and watch two episodes of whatever show we're watching right now. 39:45 Right. 39:45 And like, you know, right now we figured out how to get the rap sling, and so we go on a lot of like you know, 60-minute, 90-minute walks because baby likes that. 39:53 Yeah. 39:53 And then we can we can chat, you know. 39:55 Sometimes we'll actually we'll be listening to a podcast together, so we each get a headphone. 39:59 And it's like, I keep telling her, it's like, even as hard as it is right now, this is like the last time because like once however many years we wait to have a second one, like there's gonna be no going for a walk with like a three-year-old and a newborn, and like, oh, let's listen to the latest, you know, New York Times podcast together. 40:18 It's just gonna be full-time entertaining. 40:20 And I've seen this because of my my sister has two kids that are like four and one and a half, and they're like full-time entertainers, you know. 40:27 It's like, well, we we gotta is there a library nearby? 40:30 Is there a splash pad nearby? 40:31 Like, what's nearby? 40:33 And like right now, even though we have a newborn, we still actually do get to watch quite a bit of TV for better or for worse, and uh and and do stuff while the baby's asleep. 40:44 And uh so I wonder, yeah, how much of the your old life is over when the second kid comes is a function of of that that you've even if you settle one and then one sleeping, oh we've got it we've got a second one that that is talking and walking and uh trying to jump off things. 41:01 Yeah, I think that's definitely that's definitely a huge part of it. 41:04 I I think there's also like you know, I had mine pretty close together, they weren't even two years apart fully. 41:09 Oh wow. 41:09 And so like they were like both kind of rambunctious the same way, but but there is a good thing. 41:14 I remember right before right before I had my first kid, I was at lunch with a bunch of people from work, and this one guy, Brian, he was because it came up that my wife is pregnant, we're about to have a kid kind of any day. 41:24 And so he said something along the lines of, yeah, man, having the first kid was really difficult. 41:29 I was not ready for the first kid. 41:31 It was like so much work, and it just it felt like it just kept getting worse and worse. 41:35 And then you just kind of trailed off. 41:37 I was like, That's it? 41:37 That's your advice. 41:38 He was like, Oh, well, when she turned seven, it got way better. 41:42 I was like, that so seven years? 41:45 That's what you're telling me. 41:46 Seven years 41:49 seven, 41:49 and I was like, Jesus Christ, but yeah, but but almost almost exactly that happened with me. 41:55 When the youngest one turned seven, the two of them could kind of play, they're both boys, so they could kind of they're interested in more or less the same stuff. 42:02 They could play together reasonably well. 42:04 I had to be like nearby to like, you know, kind of officiate. 42:08 You know, like they needed they need a ref sometimes, but but like mm you know mostly they were kind of on their own thing. 42:16 Like I like I couldn't just completely ignore them. 42:18 Like I couldn't be involved in something where I was not gonna hear them if something happened, but but like I didn't have to be directly involved in everything, and that was like a massive change in my life. 42:28 But it happened almost exactly when the second was seven. 42:31 Interesting. 42:31 We have a we have we have one boy right now, and I predict we're gonna have a second boy, but that's kind of wishful thinking because of I've got a good second boy name that I think would go for completely unimportant reasons. 42:44 Uh but I am a little bit worried because I I've heard stories from a coworker of mine that he's got two boys, I think, well, at least two boys. 42:52 There might be other kids in the picture, but he says it's it's just nuts. 42:55 Like anytime they're getting in the van, you know, it's like they open the door and it's a race to see who can get to the van, and there's like dents, yeah, dents in like the van door because they have no concern for their well-being. 43:05 And they say, How many times do I have to tell you don't slam yourself into the van? 43:10 And uh I'm not sure if that's similar to your experience if you got when you 43:14 Oh absolutely they were competitive about absolutely everything. 43:17 Really? 43:18 Yeah, yeah. 43:18 And and it was made worse by like, and I think this is like actually adaptive that if it like I don't have any experience with kids except my own. 43:25 I grew up basically an only only child, like I had two sisters that were like ten years older than me, so like I I didn't grow up with other kids in my household. 43:32 And so like I didn't really know how siblings were gonna be or how radically different they were gonna be, but it's adaptive that second kids are bigger because they need to basically compete for resources with the first kid. 43:44 Really? 43:45 Right, like if the first kid eats all the food, then the second kid starves, so the second kid needs to be big enough to like elbow them out of the way and get some food, right? 43:51 And uh and it's also adaptive that the first kid is a picky eater and the second kid is will eat anything. 43:56 Same reason. 43:57 Really? 43:57 The the the first kid finds the stuff they like and they eat it, and the second kid fills in the gaps with whatever the first kid doesn't like usually and eats that. 44:07 Because again, you want to, you know, thrive, and the way you thrive is by getting the right amount of food, and so those two things are both adaptive. 44:13 But so I had this one situation where the two kids were like really little, I think they were like just like twenty no, like almost two and a half, and then the other one was like eight months old or something really little like that. 44:25 And the little one, who was big, was basically weighed within a pound of his older brother, was almost two 44:30 years 44:30 older than him. 44:31 Okay. 44:32 And he had him by the back of his shirt, and he was frog marching around the kitchen laughing, and the other one was like, ah, trying to get him off. 44:39 He didn't know what to do. 44:40 Like he just he's a toddler. 44:42 He's got this other toddler like manhandling him. 44:44 And it was like so hard not to laugh at that. 44:47 But you can't you know you can't reinforce his behavior, but it was so hard not to laugh. 44:51 But yeah, so like they were because of their size and stuff, they were always like super competitive and always trying to do everything to one up each other and whatever. 44:58 It's kind of mayhem in the house. 45:00 Wow. 45:00 Stuff to look forward to. 45:03 But it's it's very likely if you have one kid, I don't know about very likely, but it's likelier if you have one kid of a certain sex, then you're gonna have another kid of the same sex. 45:11 Really? 45:12 Yeah. 45:12 Because it's not fifty-fifty. 45:15 There's things that determine the sex of the child that are not that make it not fifty-fifty. 45:20 And so it's more likely if you have one that you're gonna have another one of the same kind. 45:23 Really? 45:23 So if you Yeah, if the if the man finds himself to be extremely high status, uh, then he will have more more male children because that's more ad advantageous for the male children to survive. 45:35 This is not a pet theory of mine, this is like evolutionary biologists have established that this is this is something that happens. 45:40 If the man believes his wife to be in i very attractive, then he'll have more female children. 45:44 I'm not sure what the difference is. 45:45 I'm not sure how it balances out. 45:47 If you think you're if you're the king and you have a very beautiful queen, I don't know if that I don't know how that works, but um but but but the idea is that uh you you are because you're selecting the the generation of your sperm cells, not consciously, but because that's something your body can do, it does it to make your offspring more advantageous according to your perceptions of the world. 46:06 Interesting. 46:06 So I really gotta get like a self-inflated ego going into the second kit here and feel like, you know, men are the best, I'm I'm the best, uh, in order to get like that extra point two of a percentage point. 46:18 Just do that Superman stance in the morning every day when you wake up. 46:22 She just sees me uh, you know, staring at myself in the mirror. 46:25 Uh 46:26 right. 46:26 You're you're a god. 46:28 Nothing, nothing. 46:31 But this has been a blast. 46:33 Why did my recording stop? 46:35 When did it stop? 46:36 Oh no. 46:38 Oh no. 46:38 There's a little spinning wheel thing. 46:40 It's on a Mac. 46:41 What is going on? 46:42 Tell me it's okay. 46:43 Oh no. 46:44 I push the stop button, so it's you know, just a the normal audacity thing. 46:49 I had the recording going and then I pushed the stop button and now it's just a spinning pinwheel. 46:54 Uh okay. 46:56 Hopefully this got saved. 46:57 Hopefully it's not like a loss that would suck.
Community discussion