[00:00:00] Nathan: Hey, we are live.
[00:00:02] Andrew: We are live, allegedly
[00:00:04] Nathan: my, my [00:00:05] signal is when you say hello.
[00:00:08] Andrew: know, right? There's, [00:00:10] there is... I don't know. I, I, I keep sa- I say this half the time when I'm like, "I just need to finish [00:00:15] doing the proper studio live setup shit." doing it remote's kinda [00:00:20] weird, but we can wire all that in, but I got
[00:00:23] Andrew: Too much other AI stuff going [00:00:25] on.
[00:00:25] Nathan: And some other things potentially related to AI.
[00:00:29] Andrew: [00:00:30] Yeah. I don't know. It's good. There's a lot to cover. I feel like we didn't even record last week. My life has [00:00:35] been fucking crazy, so it's like I I got a lot [00:00:40] moving and I'm still speeding up somehow
[00:00:43] Nathan: Nice. [00:00:45] Nice?
[00:00:45] Andrew: I don't know.
[00:00:45] Andrew: I, I was... It's funny 'cause it's, I we, we'll, we'll [00:00:50] talk about AI and then maybe segue.
[00:00:52] Nathan: Sure. Yeah.
[00:00:53] Andrew: anybody tuning in, [00:00:55] we are gonna talk about aliens and we're gonna talk about [00:01:00] AI.
[00:01:00] Nathan: Yeah.
[00:01:00] Nathan: Who would But
[00:01:01] Nathan: That they so I,
[00:01:02] Andrew: uh... like this?
[00:01:04] Andrew: Oh boy, [00:01:05] I I think like at the top of my mind is like I had a pretty good week last week building [00:01:10] and, and reviewing some stuff with a buddy and, I continuously get [00:01:15] feedback from folks that they're just completely amazed about what I'm building and how much and [00:01:20] how fast, and how I seem to be like not going crazy.
[00:01:24] Andrew: And [00:01:25] I think partly like this is just how my brain works and it's [00:01:30] fine. But also it, it's how I build or don't [00:01:35] build. Secret is still don't build a bunch of shit that doesn't matter.
[00:01:39] Andrew: But, yeah. [00:01:40] Okay. It's, it's yeah, and, and how I'm using like agentic [00:01:45] systems and how I'm pushing through some of the, the problem sets that people [00:01:50] have, maybe experienced or maybe not.
[00:01:52] Andrew: And so I'm actually have a, a [00:01:55] not... I'm gonna say a lot of shit that I'm like not intending to actually be a plug, but will sound a [00:02:00] lot like a plug
[00:02:01] Nathan: Okay.
[00:02:02] Nathan: Disclaimer,
[00:02:03] Andrew: no, I it [00:02:05] it's like I like to talk about what I'm doing, but it, it... I don't love when people on [00:02:10] podcasts are just casually "Oh, I have this thing that you should check out," or, "I'm [00:02:15] gonna talk a- I'm gonna talk my book."
[00:02:17] Andrew: I hate that
[00:02:18] Nathan: So maybe this is a side tangent, but [00:02:20] i- in some weird way you could, you could argue that that is the point of podcasts. It's like the content marketing [00:02:25] angle, right? Where they provide the content, but also they're plugging And like anytime a guest [00:02:30] has come up or anytime someone comes out with a book, there's a guest that would go on [00:02:35] 50,000 different podcasts and
[00:02:36] Andrew: yeah.
[00:02:36] Andrew: That, that's like the original podcasting thing, right? It's all just an [00:02:40] ad. It's
[00:02:40] Andrew: The morning shows on, on network television or even the nightly news [00:02:45] anymore. It's just a giant ad. Yeah, I, I get that. I, I like that the podcast isn't [00:02:50] monetized in the sense that, you can donate, I think, still.
[00:02:54] Andrew: I don't even know if it's [00:02:55] real. Used to be able to go to Patreon or, or
[00:02:58] Nathan: [00:03:00] Subscriber.
[00:03:01] Andrew: subscribe- hold on, I have a, I think I have a subscribe. There we go. [00:03:05] Subscribe. Yeah, I don't really care if people do or not. I, I've said multiple times I would [00:03:10] love for the podcast to basically pay for itself.
[00:03:12] Andrew: It's close, and it's not that much. It's I don't know, [00:03:15] it's 30 bucks a month or 40 bucks a month to edit the software and then I do it quick. That's [00:03:20] why it looks like shit. But you know what? I'm okay with that. But I, I don't like the... I think that [00:03:25] you lose the authenticity if you're trying to hawk shit all the time.
[00:03:28] Andrew: And I don't have a problem with people who do it [00:03:30] as long as you understand that that's what's happening. The, the problem I have with it is most people have no idea what that is. [00:03:35] As far as I can tell, Instagram is just like a giant ad reel.
[00:03:39] Nathan: Oh yeah, of course.[00:03:40]
[00:03:40] Andrew: And I don't know that people really internalize that.
[00:03:42] Andrew: They're like, "Oh, this is really cool." I'm like, "Yeah, it's product marketing." [00:03:45] You're just, you're literally opting in to having somebody steal your data and then [00:03:50] advertise to you. So anyway. Why we're here. So I I, I do a lot of [00:03:55] I've been doing a lot more agentic work and agentic building and agentic [00:04:00] systems 'cause it's to scale a company in 2026, like why not?
[00:04:04] Andrew: Why [00:04:05] wouldn't? I don't know why you wouldn't do it. And I've come to a few different realizations [00:04:10] about working with AI that I think people don't seem to have. I've been ta- [00:04:15] I've talked to some people about it, and it's like kind of 50/50, even people who know me really well. And, and so my buddy and I [00:04:20] were j- we've been chatting about this stuff for years, right?
[00:04:22] Andrew: We used to work together once upon a time. And I, I [00:04:25] don't know that he believed me, and then I s- I sit him down and start showing him shit, and he's "Oh, okay." "That, [00:04:30] that's real," and "Damn," " That's a one, one person and some AI?" And I'm like, [00:04:35] "Pretty much."
[00:04:36] Nathan: Hmm.
[00:04:36] Andrew: And so I, I've come to, to conclude, and [00:04:40] I this is where like I have a...
[00:04:41] Andrew: I'm starting to blog again, right? I have a [00:04:45] blog that I am starting. I don't know. I've had a blog on and off for a long time. I find [00:04:50] that there's just a lot of s- yeah I find that there's a lot of stuff that, I wanna write [00:04:55] about or wanna talk about that just doesn't really fit for the podcast or I'm sick of [00:05:00] getting locked into other platforms
[00:05:02] Nathan: Okay, sure
[00:05:03] Andrew: So like I've been using LinkedIn more for this kind of [00:05:05] stuff, and it's like you have like pages or posts or something.
[00:05:09] Andrew: I don't wanna write [00:05:10] a Substack. I'm not trying to recruit you to my fucking blog. But [00:05:15] I, I have a blog, and then the other piece of this was like I'm sick of trying to promote. And [00:05:20] so I built a blog in a hour with some software that [00:05:25] I'll be able to, to do a blog post and then it'll go to the world all [00:05:30] the different socials and shit.
[00:05:31] Nathan: Yeah, yeah, right
[00:05:32] Andrew: But I own my own blog and I don't even think I have [00:05:35] like a... I, I support RSS and Atom, so if you wanna subscribe and get [00:05:40] notified, you can. But there's no fucking e- in email list. This isn't some MLM [00:05:45] scheme,
[00:05:45] Andrew: And then it, it also works with the Fediverse, so the blog itself [00:05:50] is like a social account for folks who aren't familiar, so you can talk to it, and so I, I... But I [00:05:55] have two kind of, posts that I work on, is one I, I call thinking out loud, [00:06:00] which is like me just like idea fragments or like kind of half-baked things that are [00:06:05] not super polished and more like, like what tweets used to be and then people like tweet [00:06:10] thread when they're trying to like actually try to get out an idea, but not a f- a proper like [00:06:15] novel or like essay
[00:06:16] Nathan: an, that's an interesting, yeah, that's an interesting rabbit hole to go down. [00:06:20] But yeah, yeah. They, they used to do that
[00:06:22] Andrew: Yeah. And be- and then I guess on X now you can like pay for [00:06:25] longer, and I'm have no interest in doing that. I'm... I actually might if it wasn't... It's [00:06:30] like ridiculous to get... You have to pay eight bucks a month or something. I'm like, "I'm not gonna pay $80 a month to [00:06:35] use more than 200 characters." the original limitation was because it was for text messaging, and that was [00:06:40] all you could text message.
[00:06:41] Nathan: Yeah. Yeah,
[00:06:42] Andrew: So anyway, yeah.
[00:06:44] Andrew: yeah. and [00:06:45] that was a nice limitation. And then the other kind of piece is just essays, where it's [00:06:50] more polished or longer form, more thought out, that kind of stuff. So I, I've been working on a piece for this, [00:06:55] which is why I bring it up. Again, I'm not trying to plug, but
[00:06:57] Nathan: This mean, meaning how you work with [00:07:00] AI or
[00:07:00] Andrew: Yeah.
[00:07:01] Nathan: like the idea of... Okay.
[00:07:02] Andrew: Yeah, 'cause what I've, what I've been finding is [00:07:05] that, Read the essay when it comes out. I had a a,
[00:07:08] Nathan: Sorry
[00:07:09] Andrew: I had a [00:07:10] sync malfunction earlier. I wrote a bunch of stuff, was working to refine [00:07:15] it, and then it sync destroyed my document, so I have a fragmented [00:07:20] copy that I need to redo, otherwise it would've been done this morning. It's mad. That's a whole thing. [00:07:25] Moving data is hard. Yeah, so I... Oh, [00:07:30] fine. Okay. I wanna talk about aliens too. No, no, no, no, no. So [00:07:35] the blog, the blog... So I wanted a very simple way to write a markdown [00:07:40] file and then just push it
[00:07:41] Andrew: somewhere
[00:07:42] Andrew: and then have it do stuff, right? And
[00:07:44] Nathan: [00:07:45] yeah,
[00:07:45] Andrew: there's a whole, yeah, there's a whole markdown
[00:07:48] Andrew: Theory of people out there.
[00:07:49] Andrew: [00:07:50] I just want something very simple and, and something that is flexible enough that I don't have to [00:07:55] process it. It just "Oh, here's a markdown." So the markdown directly gets turned into a post. [00:08:00] And, the, the way that I interact with the server is I just make a, a [00:08:05] document in my Nextcloud, push it to the right folder, and then the server picks up on it and the rest is [00:08:10] history.
[00:08:10] Andrew: Pretty simple. And I the reason I wanted that is I didn't wanna build a bunch of complex [00:08:15] systems bullshit. I didn't wanna have fucking FTP, SSH. There's all these [00:08:20] different ways to do this. SCP. I was just like, "Yeah, but I just want a really simple [00:08:25] off-the-shelf quasi commercial way to, to do this so I don't have to deal with all this stuff."[00:08:30]
[00:08:30] Andrew: And I have Nextcloud for some other stuff anyway, and been playing more with it. And [00:08:35] it works pretty well. There was some kind of bug though in the Nextcloud notes app that I was [00:08:40] using didn't push my local copy
[00:08:42] Nathan: that kind of sync. Oh, [00:08:45] S-Y-N-C.
[00:08:45] Nathan: Okay. I
[00:08:47] Andrew: yeah. I was like, "Really, Nathan? Do [00:08:50] we need to have this conversation?"
[00:08:52] Nathan: that's why I was so confused. You mentioned sync and I thought you [00:08:55] were, like, handwriting documents and then your sink overflowed and it
[00:08:59] Andrew: Oh, [00:09:00] yeah, yeah, yeah. No, no, no.
[00:09:01] Andrew: No, s- yeah, s- so no, no. no. So any...[00:09:05]
[00:09:05] Andrew: I don't know. I, I might open source it. I, I just, I don't know. It's such a weird kind of use case thing, but I [00:09:10] didn't wanna write an editor. I didn't wanna write all this stuff.
[00:09:12] Nathan: Yeah. Okay
[00:09:12] Andrew: And I'm trying to minimize the like just because I [00:09:15] can, I should always.
[00:09:17] Andrew: But I've, I went through the fucking everything, [00:09:20] like Gyro and, and all these other WordPress, and I'm like, I want something
[00:09:23] Andrew: super, super, super [00:09:25] lightweight, and all I wanna do is then have it pushed to the universe. That was [00:09:30] surprisingly hard. I was very surprised where a lot of platforms didn't...
[00:09:34] Andrew: Or I would have to stand up a [00:09:35] WordPress and it's a big heavy service that I I don't have to maintain. So anyway, it all [00:09:40] screwed up and now I'm using Obsidian with the Nextcloud sync plugin and it seems to be working great. [00:09:45] So
[00:09:45] Nathan: Nice. Yeah, yeah,
[00:09:46] Andrew: save yourself some headache and don't use... And there's a bunch of known issues on [00:09:50] the Nextcloud official notes app f- with this kind of [00:09:55] problem, so just be aware I guess.
[00:09:57] Andrew: I don't wanna... I don't like to shit on people too much, but come the [00:10:00] fuck on guys. Like
[00:10:01] Nathan: Not
[00:10:02] Andrew: that's pretty bad. Yeah. and I pushed [00:10:05] it and I had a problem with my phone and I, 'cause I, or like I wrote it... I think I wrote it on the [00:10:10] web. I was testing it, right? And then pushed to my phone fine. I made some changes.
[00:10:13] Andrew: It didn't reflect and I was like that's [00:10:15] weird." And then it finally reflected and then it went and I wr- I've spent like an hour. Like I'm [00:10:20] mad, right? I, I lost like probably a good 20 minutes of work. So anyway. So I'm writing [00:10:25] this and the the conclusion I think effectively to sum it up in a [00:10:30] couple words, it's a really long tease for my new book.
[00:10:33] Andrew: No, I'm kidding. I'm not writing a [00:10:35] book.
[00:10:35] Andrew: Uh, been awesome. You sure it was straight-faced?
[00:10:39] Andrew: Just troll the [00:10:40] entire audience. They're like, "When are we talking about aliens after we talk about my new book?"
[00:10:44] Andrew: [00:10:45] Player coach. Yeah. Yeah so, effectively they'll get summed up in the idea of [00:10:50] like player coach. And I think a lot of people get this wrong as well, but when you're working with AI [00:10:55] systems, especially agentic systems, you have to be a player coach.
[00:10:59] Andrew: You have to know [00:11:00] how to manage, and you have to have a, a certain set of managerial skills. You also have to [00:11:05] know your shit and be really deep in your domain [00:11:10] or, you can hit limited success, but you don't actually unlock the like crazy [00:11:15] shit that like I see every day and I know we kinda talk about and you see a lot of it.
[00:11:19] Andrew: And [00:11:20] like people will do keynotes at certain conferences and proclaim certain [00:11:25] industries are completely dead and then get ridiculed all over the internet. Yeah, [00:11:30] there's a lot of hyperbole, however, it's fucking real. I don't... Like I'm, [00:11:35] I, I just, I don't even defend it anymore and people are like, "Yeah, you can't do all that."
[00:11:38] Andrew: I'm like, "You're right. I'm just clearly [00:11:40] lying. Here's the product. Go fucking try it. If you don't like it, then I don't know what to tell you." you, but [00:11:45] this took, this must have taken you months." It's like, "No." And so it's why am I [00:11:50] able to do that and other people aren't? I have a strong skill set in tech.
[00:11:53] Andrew: I've been doing this for too [00:11:55] long, frankly. The plumber route looks great,
[00:11:58] Nathan: Like fishing around. Hey[00:12:00]
[00:12:00] Andrew: Yeah, I know, right? But I, I think it comes down to a f- a few different things. I, I should [00:12:05] probably have just read the thing that I wrote 'cause I have it much more clear. But, and I'll probably link [00:12:10] it in the podca- I don't know.
[00:12:11] Andrew: It's it's not hard to find it if you're really interested, and it's... [00:12:15] I'm not, again, I'm not trying to plug it. I should, yeah. We have a Discord. I keep forgetting to tell people [00:12:20] we have a Discord.
[00:12:21] Nathan: Yeah.
[00:12:22] Andrew: But yeah so it, it's the player/coach [00:12:25] thing, and then you, you... The, I think the problem that a lot of people get hung up [00:12:30] on is they overestimate their ability and they overestimate AI's [00:12:35] ability to bring them through things that they themselves don't have a deep experience on.
[00:12:39] Nathan: Pretty much there. [00:12:40] Yeah
[00:12:40] Andrew: And I explain some rationale and use some, some examples in [00:12:45] that. And I think there's a lot of super capable, very deep knowledge [00:12:50] folks who have a hard time working with AI because they have no managerial skills at all. They're just an [00:12:55] excellent IC or they're, like, a researcher and have to manage weird shit, but not [00:13:00] really.
[00:13:00] Andrew: You're not really managing a team probably, and even if you are, they're all highly [00:13:05] capable excellent entrepreneurially-minded people and it's [00:13:10] that's not really managing a team. It's like the dream team in a way.
[00:13:14] Andrew: Or you have [00:13:15] people with exceptional management skills who grossly overestimate their knowledge in a, [00:13:20] in a field or their ability to actually do the thing.
[00:13:24] Andrew: [00:13:25] And we've talked about that a lot. to do it. Yeah, yeah.
[00:13:27] Andrew: Yeah. And AI can help you, right? [00:13:30] So I use it all the time to push me further in, in places I don't know a lot in [00:13:35] or refine. But the thing that I see consistently is [00:13:40] that when I hit the edge of human knowledge or when I start doing novel stuff, it's effectively [00:13:45] useless.
[00:13:46] Andrew: And even on executing tech stuff and I won't pump another [00:13:50] thing that I'm building, but, it's been the thing that I've been teasing for two... [00:13:55] Probably about two years now. 18 months at least. It's had many iteration. I'm actually looking really forward [00:14:00] 'cause it's close. This it's, we are within hours of it coming out.
[00:14:04] Andrew: And I, I would [00:14:05] like to talk about it 'cause I, I think it's you know the whole story and it, it's been
[00:14:08] Nathan: yeah. I've been waiting for this ever since you [00:14:10] teased it the first few
[00:14:10] Nathan: times.
[00:14:11] Andrew: pivot, right?
[00:14:14] Andrew: [00:14:15] goddamn, that thing has pivoted, we went from circles to irregular [00:14:20] polygons and then back. But
[00:14:22] Nathan: Interestingly enough,
[00:14:23] Andrew: but it, it [00:14:25] can't do it, right? Yeah
[00:14:27] Nathan: The, the con- the concept's still the same. Your, your form factor's all [00:14:30] different though, right?
[00:14:31] Andrew: Yeah, no, that's the core of it. But a lot of people don't understand that. I even tell [00:14:35] them, right? And, and they're like, "But you're still working on that?" I'm like, "Yeah, 'cause it's still the right thing to be [00:14:40] building." It's just there's a lot of weird tangent-y question stuff that needs to be answered, [00:14:45] and I found a really clever way, I think, around one of them, which we should maybe talk about [00:14:50] offline
[00:14:50] Nathan: Okay. Okay, interesting. Okay
[00:14:52] Andrew: But, a- a- and so [00:14:55] folks are hitting that, and then, The... You have to be [00:15:00] constantly evolving how you work with the edges. And I think that's where anyone who's [00:15:05] like a decent player coach is seeing like 10X performance, let's say. And [00:15:10] then they hear about people in the 1000X or, and all these numbers are just made up, but it's, [00:15:15] it's pretty incredible. And I'm very confident that a single [00:15:20] person who's an excellent player coach and can learn very quickly and evolve can, radically [00:15:25] build things that used to take 10 people to do. And again, you might not hit [00:15:30] everything well, it might not be absolutely perfect, but it's doable.
[00:15:33] Andrew: And if you have the [00:15:35] knowledge and/or you have enough knowledge to get started and can keep going, like AI is gonna lie to you, you cannot give [00:15:40] it judgment capability. But, if you can iterate fast enough, you can [00:15:45] make the wrong mistakes that you and I have made over a lifetime and figure it out, and you can do that in a [00:15:50] few days or maybe weeks, and then unbuild it and rebuild it a [00:15:55] better way than it-- I remember I've built entire [00:16:00] systems where once you hit scale, you're like, "Oh, shit.
[00:16:03] Nathan: Yeah
[00:16:03] Andrew: Now we have to literally [00:16:05] basically scrap and rebuild because it's just we can't keep going this way. The [00:16:10] technology doesn't work," or we hit physics or whatever. So that-that's been a realization that I've come [00:16:15] to. And the second plug that is not technically a plug is I built software for [00:16:20] this.
[00:16:20] Andrew: I don't remember if I... I don't know if I even talked about it on the podcast before. Couldi is out there, [00:16:25] C-O-U-L-D-I.xyz if you wanna take a look. It'll be in public open [00:16:30] beta soon, like
[00:16:32] Andrew: hours, days. It's been in private [00:16:35] beta. There's two pieces. Is agentic building and agentic-- and the management of the agentic [00:16:40] building, and I think the management piece is, is more critical.
[00:16:44] Andrew: And I've [00:16:45] reframed this even in the last few days in that the folks who don't really [00:16:50] know are gonna probably not see a benefit and are gonna be lost. I think the, it really starts to show benefit [00:16:55] for the folks who are in that kind of player coach role and are like they're just [00:17:00] overwhelmed, right?
[00:17:00] Andrew: "I'm sick of managing the goddamn agents." There's you get all these weird kind of scaling problems with [00:17:05] agentic workflows, and so I think the tool really starts to enter there. And then [00:17:10] I'm still testing it myself, but it's... I think I've unlocked the [00:17:15] What did I say earlier when we were talking, calling it multi-agentic swarm [00:17:20] management or some-
[00:17:21] Nathan: you, you put, put, put words together.
[00:17:24] Andrew: It sounds [00:17:25] insane, and it
[00:17:25] Nathan: Yeah,
[00:17:26] Andrew: insane, but it's effectively man...
[00:17:28] Andrew: It's, so it's if you can think [00:17:30] about it in like a company structure, it's like if you're the manager, you're player-coach, you're on the team, [00:17:35] you know what your team's doing, you're managing people, you're, you're bringing shit forward. At the [00:17:40] director level, you have multiple managers, and so that's a different skill set.
[00:17:43] Andrew: And maybe you know what [00:17:45] you're doing, maybe you've come up through that and you're an excellent IC and you just have developed that skill set. But like when you get to the [00:17:50] executive level, it's a totally different world. Like you probably have no idea what's really going on [00:17:55] day-to-day, but you might think you know.
[00:17:57] Andrew: And y- you don't realize how slow [00:18:00] everything actually moves, and maybe you do, but it still moves too slow. And so [00:18:05] that's where I think like you graduate into the... And again, what do I mean by multi-agent swarm [00:18:10] management? It's like the idea of a swarm is you have multiple agents doing stuff. I have [00:18:15] agents manage swarms and spin them up, and then I have multiple different agents doing multiple [00:18:20] swarms doing multiple things.
[00:18:21] Andrew: And I don't, I don't know of an actual term for this [00:18:25] yet. I'm sure people have been talking. I'm like, I'm not, I'm not trying to claim I've invented this, but I think I've [00:18:30] invented a better way to deal with it and to scale it and to make it so that I'm not just [00:18:35] sitting in front of my computer 10 hours a day, cracked out, fucking with AI.
[00:18:39] Andrew: And so [00:18:40] anyway, so I've, I've rolled some of those lessons into the product, but it's also like I do [00:18:45] think there's a limitation where like you, the human, also you have to be continuously learning. You have [00:18:50] to learn how to manage better. Like I've managed teams, I've, I've been an individual contributor, I've [00:18:55] been a high-level individual contributor, I've been a new employee.
[00:18:58] Andrew: I've been everywhere across the [00:19:00] gambit. So I... And that's not necessarily unique, but I also have [00:19:05] very deep knowledge in technical system design, in systems design period, in [00:19:10] coding, in running teams who build shit, and so a lot of people will hit one [00:19:15] of those or maybe two of those and it's not that you can't learn it, and it's not that you can't get better [00:19:20] at the skills.
[00:19:20] Andrew: A lot of people don't want to. I talk a little bit about that in the essay that got [00:19:25] deleted. But, but I think those are the questions that people are posing with that [00:19:30] are either intellectually or not intellectually understanding it. Yeah, we're on a [00:19:35] AI tangent, but I think... Like it does, it really comes down to player-coach-ness right [00:19:40] now, and like you can get a lot of value by just asking it general questions, and a lot of people do that.
[00:19:44] Andrew: But I'm like, that's [00:19:45] not... When I talk about AI or generative AI, that is not even close. And "Oh, I generate [00:19:50] code with it too." I don't actually at this point. Oh, the second [00:19:55] piece of player-coach and then, and then when it comes specifically to software development, and a lot of AI too, but like [00:20:00] specifically software development is outcome driven.
[00:20:03] Andrew: You have to be outcome driven [00:20:05] That's the secret. And I think that'll be perplexing for a lot of folks, or they think they're outcome driven, [00:20:10] and
[00:20:11] Andrew: You're not. I don't think there's a fuck ton of easier. [00:20:15] Sure, sure. Yeah
[00:20:15] Andrew: idea. Most people don't do software development outcome driven anyway, and then they wonder why [00:20:20] nothing works or why they have people they can't, we can't align the teams.
[00:20:24] Andrew: Then you're not outcome [00:20:25] driven.
[00:20:25] Nathan: Hmm.
[00:20:26] Andrew: And
[00:20:26] Andrew: You I And you're reaching your outcomes, why do the [00:20:30] teams need to be aligned? It doesn't matter.
[00:20:32] Nathan: I, I think it does in some sense. I think this is, this is [00:20:35] the, the weird paradox that a lot of people run into, right? So if I can dig into a little [00:20:40] bit, the player/coach idea just to spell it out, someone who used to be a pl- usually [00:20:45] sports, right? Someone who used to be a player, and then they either hurt themselves or [00:20:50] they
[00:20:50] Nathan: move on in their career, and then they, they go into coaching, right?
[00:20:53] Nathan: And, and the reason that's so [00:20:55] valuable is because over that individual's career, they've developed this sort of [00:21:00] intangible expertise about that space. And that's very much what I [00:21:05] think we talk about with AI is people who actually know, like [00:21:10] coders, the process behind doing it, are very well equipped to use them in that way [00:21:15] to assist because they actually know the low-level details. This is something we've talked about before on the cast. [00:21:20] It's
[00:21:20] Nathan: nothing particularly new. But I think that's the, the reason why the [00:21:25] player/coach is so interesting is because that person has developed this working [00:21:30] experience over that period of time, and the parallel to someone using AI to [00:21:35] learn it, I think is, in a way, trying to fast-track that learned experience, right?
[00:21:38] Nathan: If you can get into the [00:21:40] mindset of using the AI to teach you the thing, the process of you going through [00:21:45] it, the, the quote-unquote, "learning process" that you're doing is effectively that, the player part of the player/coach [00:21:50] thing. That's why I think a lot of people don't understand this idea of being able [00:21:55] to learn using the AI but be outcomes-driven, because if you can't... [00:22:00] It, it's like trying to do two thing, two different things at once. If you don't know what the outcomes are, you can't [00:22:05] know what it is you need to learn and vice versa,
[00:22:07] Andrew: 100%. And, and that [00:22:10] it's, You can use AI to push your knowledge forward. You can use AI to [00:22:15] learn. The problem becomes the further away you deviate from [00:22:20] common knowledge, even if AI is really exceptional, it's, it's [00:22:25] still stats, right? It's still giving you a standard de- distribution [00:22:30] of knowledge, and a lot of times it gets to be wrong, and this is where I know a lot [00:22:35] of researcher friends who are, like, really lukewarm on it.
[00:22:38] Andrew: And I'm like, my [00:22:40] buddy and I actually had a really long debate about this, sometime this weekend, [00:22:45] and he's doing a lot of, I hope to get him on a podcast, but he's doing a lot [00:22:50] of really interesting things on the edge of human knowledge. [00:22:55] And, a one of the projects that I've talked about a little bit and, but it's also [00:23:00] there.
[00:23:00] Andrew: And what I find is... And, and I was actually, last night I was... This is a g- a great [00:23:05] example. So I was trying to improve, an AI model pipeline, and I was like, "How [00:23:10] do we improve it?" And it gives me some kind of bullshit stuff, and then I was like, "Okay, let's test it." And then it's oh, we've te- [00:23:15] we've tested this, we haven't tested this.
[00:23:16] Andrew: We test it. Oh, it doesn't work. I was like, "What the fuck?" And so then I w- I was doing some [00:23:20] other research with actually AI, funny enough, and that model was able to [00:23:25] come up with, A new model had dropped like, few days [00:23:30] ago, and it's in a different variation. It's open weight, so it fits in, [00:23:35] in the shape.
[00:23:35] Andrew: And I was like, "Cool, let's check this out." And the, the AI that's running this pipeline design was just like, [00:23:40] "No," "We don't need to. It doesn't fit," blah, blah, blah.
[00:23:42] Nathan: Hmm.
[00:23:43] Andrew: I was like, "Okay, we're gonna do it [00:23:45] anyway," and use these settings, 'cause I'm using AI to execute. I don't give a... I'm [00:23:50] not gonna shit, sit there and fucking write code like a monkey.
[00:23:53] Andrew: And, we tested it [00:23:55] and it improved processing throughput by, almost [00:24:00] 1,000%. With no meaningful degradation in [00:24:05] performance of
[00:24:05] Andrew: Output.
[00:24:06] Andrew: So if I would've not known anything and just said, [00:24:10] "Hey, I make it better," and it said, "Nah, I can't," then I wouldn't have gotten that.
[00:24:14] Andrew: I'm [00:24:15] dubious that it would've made it better.
[00:24:17] Andrew: I didn't send you... I was gonna send you a screenshot, 'cause [00:24:20] it's unbelievable.
[00:24:21] Nathan: Yeah Tensors are very strange things. [00:24:25] But, it's it's, it's just, it's an optimization piece, so it's cool. And then I did... A- overall, the [00:24:30] pipeline eh, I don't know, it's maybe a 10% increase or decrease in time run or [00:24:35] whatever.
[00:24:35] Andrew: But faster is always better, and less energy and less compute is always better in [00:24:40] my mind, and there was, like, a very negligible... I spent an hour testing it. [00:24:45] Th- that gets even worse for things like biology or chemistry or other [00:24:50] human sci- subjects. But the problem I'm, I'm running into with a lot of folks deep in those [00:24:55] fields, though, is that they also overestimate their understanding of their own domain and field, [00:25:00] and they are not trying to use the [00:25:05] technology in other ways that can help.
[00:25:07] Andrew: And again, my buddy and I were talking a lot about this, [00:25:10] 'cause I, I've been, I've been trying to use it to further some stuff I'm doing [00:25:15] along, which is as far as I understand, it's not been done before. Maybe I'm [00:25:20] wrong, but I can't find any reference of it. AI can't find any reference of it.
[00:25:23] Andrew: It's true what is it? [00:25:25] Greenfield research or whatever.
[00:25:26] Nathan: yeah. Yeah, yeah
[00:25:27] Andrew: And AI's just not helpful, [00:25:30] but you can u- still use it in ways to do some research or do some work. [00:25:35] And I talk about it, it's almost like you trust it like a Wikipedia article, right? I'm like, Wikipedia is a great [00:25:40] source of information,
[00:25:41] Nathan: you check the sources
[00:25:42] Nathan: and, yeah it's a getting started and it's...
[00:25:44] Andrew: You'll [00:25:45] find it's wrong. The more you learn about a topic, the more you're like, "Oh, actually there's a lot more information and [00:25:50] details and stuff that may or may not be relevant." And so anyway. So [00:25:55] yeah. So I, I, and that's the AI thing. We had a much more interesting [00:26:00] cast, I think, planned
[00:26:01] Nathan: We were talking before it. Yeah, yeah. I
[00:26:03] Andrew: But I wanna talk about [00:26:05] aliens.
[00:26:05] Nathan: how do you want to... segue this?
[00:26:07] Andrew: I don't fucking know. I I mean, so it's inter- [00:26:10] okay the reason that the, that we kinda came up with the topic, we were talking about some stuff, [00:26:15] and I think you made the observation... I know, right? Yeah. You made the observation, though, [00:26:20] that, I guess we were talking, and one of my things is I, I get a lot of...
[00:26:23] Andrew: A lot of people I talk [00:26:25] to have an irrational hatred of AI. I don't get it. I [00:26:30] they think it's gonna kill everyone, they think it's gonna destroy the environment, destroy humanity [00:26:35] whatever. I'm like, "I, I don't know, maybe?" There's a lot of stuff that can do that. And so I look at it, [00:26:40] and, and maybe I'm a little biased 'cause I do literally do this every day, seven days [00:26:45] a week, but I see it for what it is, and it's like an in- incredible [00:26:50] tool for coding i- in the right contexts.
[00:26:53] Andrew: It's an incredible tool [00:26:55] for, synthesis if you're okay with accuracy not [00:27:00] being perfect.
[00:27:01] Nathan: Sure. Right
[00:27:03] Andrew: Outside of that, s- [00:27:05] it's not super useful as far as I can s- I there's always little deviations here and there, [00:27:10] but on average or, or for the, the broad strokes, just [00:27:15] not. And, it's interesting because we've talked about the political camping before where people, it seems like [00:27:20] they're locked into this societal opinion that makes no sense.
[00:27:23] Andrew: And it's... Like I've said before, [00:27:25] it's the social media algorithm. I'm, like, seeing it manifest in people who should know better, and they [00:27:30] don't. And it's on both sides of the political spectrum. And funny enough with AI a lot of Republicans hate [00:27:35] AI as much as Democrats, which is, one of the few issues, right?
[00:27:39] Andrew: It's one of [00:27:40] the few issues Americans in large... But at the same time if [00:27:45] you... To actually understand AI we're talking about probably tens of thousands of people in the United [00:27:50] States at most who really, really, really understand this or have built it or [00:27:55] are capable of actually building and training models, have an understanding of what that means [00:28:00] understand either at a high level or low level the maths [00:28:05] happening and what all this stuff even is.
[00:28:07] Andrew: And I, I know f- some of these people, [00:28:10] and we're all, I think, pretty aligned in yeah, the technology's pretty bonkers and it's [00:28:15] almost magic. But these CEOs out there blabbing that it's gonna kill us [00:28:20] all, and, like, all these people, "We need to slow down," and this and that. It's just like we're not seeing it [00:28:25] connecting, and there's all sorts of reasons why.
[00:28:28] Andrew: We've talked about this. [00:28:30] But on the flip side, we also have had this weird alien shit going on.
[00:28:34] Nathan: [00:28:35] Yeah
[00:28:35] Andrew: But we've t- we did have cast about this when the first tranche, whatever the fuck a tranche [00:28:40] is, came out, right? 'Cause I was like, "Holy shit, the government is releasing UFO documents." what is it, [00:28:45] war.gov/ufo or something?
[00:28:47] Andrew: I don't even remember.
[00:28:48] Andrew: And I, I thought that was [00:28:50] notable 'cause the f- first time in my lifetime the government's straight up like, "Yeah, there's weird shit [00:28:55] and we're gonna release what we know about it," allegedly. And there's a lot of people seem [00:29:00] split. They're either full on aliens are real and it's finally time, and then there's a bunch of [00:29:05] people who are like, "This is bullshit," or whatever political or whatever their rationale is.
[00:29:09] Andrew: And the r- [00:29:10] I think the reality is for both, it's a highly nuanced conversation and there's a [00:29:15] weird overlap of the two things. And like we were talking about [00:29:20] earlier, oh, what's his name even? I don't even know. This is how much, this is how like literally how much I pay [00:29:25] attention to it. I remember the guy we were talking about, right?
[00:29:28] Andrew: I think it's David [00:29:30] Grusch.
[00:29:31] Nathan: Crush,
[00:29:31] Andrew: Grusch, I think. I didn't know who you were referring to.
[00:29:33] Andrew: Oh,
[00:29:34] Andrew: yeah.
[00:29:34] Andrew: Yeah, yeah.[00:29:35]
[00:29:35] Andrew: I don't even... I don't know. I don't remember.
[00:29:37] Andrew: But, but he was on another [00:29:40] podcast recently. I'm not gonna plug it. But I'll plug my own stuff. I won't plug other people's [00:29:45] stuff. And I thought it was actually interesting. I, I gave it a kind of a background listen [00:29:50] and, I was very surprised by some of the claims being made.
[00:29:54] Andrew: [00:29:55] And I'm even more surprised that people aren't taking it seriously. And the, the... Get [00:30:00] rid of all the UFO shit aside, my understanding of this guy, he's testified in front of Congress under [00:30:05] oath, so if he's lying, he's gonna go to jail forever. He's given classified [00:30:10] briefings to people in Congress.
[00:30:11] Andrew: He's legit. Now he [00:30:15] goes on podcasts, says a bunch of crazy sh- sounding shit. I don't know. I'm not endorsing it, but I'm like, [00:30:20] this isn't some kooky dude who's discredited, and then you start looking [00:30:25] at a lot of these other people coming forward and there, there's something here. It, it [00:30:30] could be a conspiracy, it could be real, it could be some variation between, but there's just too [00:30:35] much kind of smoke per se.
[00:30:37] Andrew: But we were talking I don't know what to make of it. [00:30:40] I just think it's really strange that we're not talking about it
[00:30:43] Nathan: Yeah, yeah. I think [00:30:45] we, I was pinging on that point because I saw the same sort of thing where you have some [00:30:50] people in the AI camp or the alien camp that are claiming
[00:30:53] Nathan: these things. [00:30:55] Okay, billion. But I you, you have people... We talk- I think the, the, the phrase we used [00:31:00] last cast, the cast before, it was like substantial claims require substantial evidence.
[00:31:04] Nathan: It's,
[00:31:04] Andrew: Yeah
[00:31:04] Nathan: the whole [00:31:05] idea. And when you're talking about super intelligence or a, some
[00:31:08] Andrew: Yep.
[00:31:08] Nathan: that, that will
[00:31:09] Andrew: Or [00:31:10] murder robots.
[00:31:11] Nathan: Or murder robots. And the same thing with aliens, the existence of some [00:31:15] sort of supernatural being, you, you need the evidence to support that. But nobody's talking [00:31:20] about the alien stuff, whereas everyone's talking about the AI stuff, and it's, I just find it really, really weird that [00:31:25] it, it's the same thing, but it's just manifested differently.
[00:31:27] Andrew: No, exactly. It's exactly the same thing. [00:31:30] There is no evidence that AI is omniscient and super intelligent or anywhere near [00:31:35] it. Nobody can really define what that is. Nobody can really define [00:31:40] what I'm supposed to worry about. And I, I if I'm missing something, folks, please let me know. [00:31:45] Everything that people tell me, "It's gonna take over the internet," I'm like, how?
[00:31:48] Andrew: And if it's ca- [00:31:50] if the internet is being, is capable of being taken over, then it's capable of being taken over [00:31:55] today, and that's not necessarily an AI advancement. That's just an exploit on our own [00:32:00] infrastructure. Okay, fine. But with it, it's the same thing with the alien [00:32:05] thing, right? It's like you keep talking about it.
[00:32:06] Andrew: Let's see some aliens. Let's see some crash. Let's see some evidence [00:32:10] other than I know a guy or you gotta trust me. And, I, I... And again, I think for folks who [00:32:15] like don't pay attention, and I'm definitely not an AI connoisseur, but there's
[00:32:19] Nathan: alien come [00:32:20] tomorrow.
[00:32:22] Andrew: people, in government that have been in super high level [00:32:25] positions all over the military and intelligence sectors who keep saying "Yeah, there's, there's something here [00:32:30] we need to talk about. I can't talk about it, national security or whatever." Maybe we should [00:32:35] believe them, but it, it seems very unlikely that a bunch of people are just gonna be in some kind [00:32:40] of conspiracy to whatever. And then to what end? What are they getting out of it other than like [00:32:45] grief? So it's like that doesn't really make a lot of sense.
[00:32:48] Andrew: And this isn't just some random [00:32:50] people who "I got abducted" kind of shit.
[00:32:52] Andrew: Maybe they got abducted. I don't fucking know. [00:32:55] But I don't, I don't want to like... I don't... We've talked about it.
[00:32:58] Nathan: don't wanna discount people. Yeah,
[00:32:59] Andrew: the alien [00:33:00] stuff in, in AI super intelligence is like just as theoretically possible [00:33:05] quantum physics land and and, and like the, the...
[00:33:08] Andrew: Sure, we're also nowhere near there. [00:33:10] And like one of the thought experiments we were talking about is if aliens are real and they do have the ability to bend [00:33:15] space time, like I have no reason to believe... And none- I have no reason to not believe any of these [00:33:20] people because what they're actually talking about only sounds insane because of my own like [00:33:25] human slash like lens on history.
[00:33:28] Andrew: That's it. [00:33:30] So I'm not saying they're, they're, they're not lying. I'm not saying they are true. I'm not saying they're not being misled [00:33:35] themselves either. But I don't know. I think it's also just like more of a first principles way of [00:33:40] thinking of we understand that quantum gravity exists and does really strange shit.
[00:33:44] Andrew: [00:33:45] We understand that information exchange happens across vast distances [00:33:50] instantaneously. We understand that there's all these tiny little particles that make us up, and that [00:33:55] most of you and I in the world that we perceive is actually empty space, but we perceive it [00:34:00] as solid and, and so it's okay,
[00:34:02] Nathan: Yeah.
[00:34:03] Andrew: I don't think it changes [00:34:05] anything about my life, to be honest. And if aliens came tomorrow, I'd just be like, "Okay, got a lot of questions."[00:34:10]
[00:34:10] Nathan: Cool.
[00:34:11] Andrew: I don't, I don't think my life necessarily fundamentally [00:34:15] changes. I, I think a lot of people have scary questions at that point, but it's like, [00:34:20] you get over it.
[00:34:20] Andrew: But, I don't know.
[00:34:22] Nathan: Yeah, yeah, yeah
[00:34:23] Andrew: the, the, the topic was [00:34:25] fun and interesting, and I really didn't want to talk about AI today, but [00:34:30] here we are.
[00:34:31] Nathan: Yeah. To be fair, I I, no, I, I, uh, [00:34:35] yeah, I, I think I, I think I led you down the path because I just thought it was very [00:34:40] interesting to see that from the perspective of, of an AI doomer who's saying that [00:34:45] AI is super intelligent, right? And we're, we're pointing fingers and saying, "Hey, where, where's the evidence of that?" [00:34:50] And you have people coming from reputable [00:34:55] organizations saying that they've seen aliens or they know something that's happening. The same sort of... [00:35:00] I don't wanna hyperbolize it, but the, extremist belief [00:35:05] that they're, that whatever it is that they believe in actually does exist, and everyone else is outside of the [00:35:10] circle. The perception of the people who are outside of the circle of the AI [00:35:15] doomers are like, "This, these people are crazy. I don't see it." And then on the other side with aliens, it's it's [00:35:20] not quite the same extremism as "These people are crazy. I don't see it." It's more of "Oh. "Oh. [00:35:25] okay. You believe what you wanna believe kind of
[00:35:27] Andrew: Yeah, no, and, and the, the AI [00:35:30] st- we talked a little bit about it, but, specific harms. I-- [00:35:35] with AI, I, I just... There's a lot of negligence happening [00:35:40] with these company... They're not labs, they're companies, and I, it, it's like they've got [00:35:45] everybody con- It's almost like if there are aliens, we're talking to two of them on a daily basis at this [00:35:50] point, 'cause they have everybody all convinced of this alternate [00:35:55] reality kinda thing, and, yeah, I understand you gotta hype your book and sell your stuff, but [00:36:00] it's, o- one of the CEOs in particular has been dooming AI for [00:36:05] 10 years. You go back to GPT 2.5, and there were, like, a bunch of people at OpenAI who were [00:36:10] like, "This is gonna end the, end the world." Okay. That's what? Four or five years [00:36:15] ago?
[00:36:15] Nathan: It feels like, it feels like longer than that, but yeah.
[00:36:18] Andrew: I don't remember the specifics, but [00:36:20] and G- GPT 2.5 sucked. Everybody knew it. It was a massive advancement at the time, [00:36:25] but it wasn't useful as far as I'm concerned. And it's gotten a lot better, and [00:36:30] it's still mildly useful for some things. But it, it's... I think it's gonna [00:36:35] go down in history as one of the most bundle- bungled technologies [00:36:40] ever because it was so overly hyped up front.
[00:36:44] Andrew: And like we were talking [00:36:45] earlier and, Claude 5.5 is... [00:36:50] I think I said it earlier, I was like, it's the first model that feels like this is what [00:36:55] it probably should have been the whole time.
[00:36:57] Nathan: Person, yeah.
[00:36:57] Nathan: Yeah
[00:36:58] Andrew: It works pretty damn well. It's [00:37:00] pretty damn good at reasoning. It's pretty damn good at understanding what I'm asking.
[00:37:04] Andrew: The [00:37:05] information I get is still sometimes sus, but
[00:37:07] Andrew: it's frankly pretty good. It, it's... [00:37:10] I, I would say it's just as sus as asking a random person on the street I don't trust and [00:37:15] have, know any...
[00:37:16] Nathan: Yeah.
[00:37:17] Andrew: topics I don't know anything about, right?
[00:37:19] Nathan: Yeah. [00:37:20] Yeah
[00:37:20] Andrew: The energy costs are, are dropping, the water costs for the data center [00:37:25] for all the people who are flipping out about that.
[00:37:27] Andrew: You still need to look into golf courses if you're really [00:37:30] concerned about AI water, look into golf courses. Golf courses everywhere. And [00:37:35] also, very few people can use a golf course.
[00:37:38] Nathan: Yeah,
[00:37:38] Nathan: yeah.
[00:37:38] Andrew: people can use [00:37:40] AI Now I'm just mad. Yeah, I'm mad.
[00:37:42] Nathan: Okay, aliens.
[00:37:44] Andrew: Aliens. [00:37:45] So anyway, aliens are real or they're not. No one really fucking knows, or maybe [00:37:50] people do know and they can't talk about it, but trust me, bro.
[00:37:52] Nathan: Yeah,
[00:37:53] Nathan: yeah. I think now. You know what I [00:37:55] mean? It's that's exactly. That's kind of where, where I'm at as well. It's like I, I don't know what to make heads or [00:38:00] tails of anything anymore, and it's just o- all you have is, is whatever people are [00:38:05] saying
[00:38:05] Andrew: What, what did I, we should have just recorded like we always plan to do. I [00:38:10] it's like with theories of time travel, warping space time, all is bullshit. [00:38:15] The likely possibilities are that they are humans from the future
[00:38:18] Nathan: Hmm.
[00:38:19] Andrew: They are [00:38:20] aliens from the future.
[00:38:21] Nathan: Yeah, yeah, yeah
[00:38:22] Andrew: They are future us, [00:38:25] right?
[00:38:26] Nathan: Nonlinear time
[00:38:26] Nathan: space beings, right? Or we are some [00:38:30] sort of divergent, evolutionary line.
[00:38:33] Nathan: Yeah, yeah
[00:38:34] Andrew: There's, there's science [00:38:35] fiction tropes of this where the technology is so advanced in society, like half the [00:38:40] group, half the people decide not to do technology and become Luddites, and the other [00:38:45] half, do whatever crazy technology shit that the technology can do.
[00:38:49] Andrew: It's some [00:38:50] sort of known or unknown natural phenomenon. Just because we're seeing shit [00:38:55] doesn't mean they're aliens. It could totally be
[00:38:57] Nathan: Some weird
[00:38:58] Nathan: unexplainable... [00:39:00] or it could be unnatural phenomenon caused by aliens. Who the fuck knows, right?
[00:39:04] Nathan: [00:39:05] Natural phenomenon caused by it. yeah.
[00:39:06] Andrew: Yeah
[00:39:07] Nathan: Yeah
[00:39:08] Andrew: And I think the interesting thing is it [00:39:10] could have...
[00:39:10] Andrew: Th- th- this, they, them, this could have been here the whole time,[00:39:15]
[00:39:15] Andrew: right? It... And, and maybe they're not aliens, maybe we're aliens, right? Maybe we, we were [00:39:20] the... We didn't start here. Who knows, right?
[00:39:22] Nathan: Yeah, yeah.
[00:39:23] Andrew: I think when you get... When, when you...
[00:39:24] Nathan: [00:39:25] planet Earth
[00:39:26] Andrew: I, I think to, really try to start thinking about this [00:39:30] kind of stuff, you do- you have to, Like...
[00:39:31] Andrew: we were talking earlier, like we assume that we were here, like God put us here, [00:39:35] or we evolved from monkeys or whatever people believe. We frankly don't [00:39:40] know. We, we don't, and those are our best guesses, and there's some deviating theories. But [00:39:45] for the most part, I think those are the two camps of human existence, and that could totally be [00:39:50] wrong.
[00:39:50] Andrew: And I think you ha- just assume that's wrong and then assume what the hell, what, what what [00:39:55] else would make sense. And I, I think... I have al- I've talked about that before. I love the [00:40:00] Stargate theory of the TV show/movie. think it's fucking awesome. [00:40:05] I love it. I love the series. I don't ne- I don't think it's some truth or thing.
[00:40:08] Andrew: It would be hil- they [00:40:10] actually have a couple episodes about that, and I think it's fucking hilarious that they went that far. [00:40:15] Incredible writing. But, I, I think, like if, if you really wanna try to [00:40:20] understand, then you have to ask questions. And if your knowledge is [00:40:25] the limiting factor of you understanding, then I think you have to ask "Okay, how, how can I [00:40:30] break down the knowledge that I think I know," and really get to first [00:40:35] principles or beginner's mind kind of idea.
[00:40:37] Andrew: And that doesn't mean you need to believe the [00:40:40] conclusion, but I don't... I think you're gonna really push your understanding if you do that. I think you can also get to [00:40:45] crazy town but-
[00:40:46] Nathan: Very easily, yeah.
[00:40:47] Andrew: But yeah, I don't know. I don't, [00:40:50] I don't really have much more to say.
[00:40:52] Andrew: Other,
[00:40:52] Andrew: other than the thumbnail is killer, and I- if you're [00:40:55] with us still, sorry we kinda thumbnail baited you
[00:40:59] Nathan: We, we did wanna [00:41:00] go that, that way. And as, as it happens when we talk about random things, you kinda go on tangents.
[00:41:04] Andrew: I [00:41:05] know, right?
[00:41:05] Nathan: us for that.
[00:41:06] Andrew: I I I, I, don't, I, I think the core idea... [00:41:10] It's funny because I think a lot of podcasts start with the conclusion, and then they [00:41:15] the story. That's the new thing, right? I wanna watch the trailer, and if I don't know exactly what's [00:41:20] gonna happen, I'm not gonna watch it.
[00:41:21] Andrew: I don't really care. We tend to do the opposite, right? We [00:41:25] start with, a story or some seemingly random thing and talk about [00:41:30] stuff, and then we kinda get to the conclusion. That's how stories used to be
[00:41:34] Nathan: Hmm.
[00:41:34] Andrew: [00:41:35] Right? You open a book and,
[00:41:36] Nathan: Yeah,
[00:41:36] Andrew: you're like, you've just entered some random [00:41:40] world and you're walking around and you meet some characters and you have no idea what's going on, and then you [00:41:45] start to slowly understand over time what, what's going on and what they're doing and how it [00:41:50] works, and then sometimes at the end you get a conclusion, sometimes you don't
[00:41:54] Nathan: [00:41:55] Hmm.
[00:41:55] Andrew: But we've flipped it on the script.
[00:41:56] Andrew: It's really interesting for the attention economy. Nobody
[00:41:59] Nathan: What [00:42:00]
[00:42:00] Andrew: 20 minutes listening to a thing that they might not already agree with, which is, again, the problem [00:42:05] and where we're at with the whole aliens and AI debate
[00:42:07] Nathan: Yeah, yeah. Nice, nice way of bringing that back. [00:42:10] Yeah, yeah. Agree with that.
[00:42:11] Andrew: Right? Sometimes I know what I'm doing
[00:42:14] Nathan: As [00:42:15] if you just wandered into the right path.
[00:42:17] Andrew: I'm in the right multiverse
[00:42:19] Nathan: Yeah, [00:42:20] yeah.
[00:42:20] Andrew: All right. I think we should leave it there. You have any other closing thoughts?
[00:42:23] Nathan: No, that was good. [00:42:25] No, well done. Yeah.
[00:42:26] Andrew: We did it. Happy Thursday. See you next time.
[00:42:29] Nathan: [00:42:30] Bye. That was well done.