13: Nabeel S. Qureshi - The Will to Care

[00:00] Welcome to Dialectic, Episode 13, with Nabeel Qureshi. [00:04] Nabil's a writer, entrepreneur, and spent much of his career working at Palantir. He recently started a new company, but he's also known for his amazing writing on technology, AI, culture, understanding and learning, Palantir, and much more. [00:20] He's someone who consumes... [00:22] unbelievable amount of information and seems to just have a polymathic [00:26] interest curiosity and understanding across so many different domains and i was [00:31] Thrilled to interview him. [00:33] I decided to focus at least the first half of the conversation on two big ideas and [00:38] that feel connected to so much of what I've been thinking about and reading and even talking to people on the show about. [00:44] The first is based on a tweet thread Nabil wrote recently. [00:47] titled The Opposite of Sloth. [00:49] is care. [00:51] As Nabil states, slop is one of the defining aspects of modernity. [00:55] And one of the reasons that I think so much of our experience with media and culture these days is, [01:00] can feel meaningless. [01:02] We spend some time trying to interrogate what actually makes for slop. [01:06] And then we talk about how to create things that do have meaning. [01:10] with care at the root. [01:12] This is not a science. And as you can tell, I think he and I are both thinking through this live, but it was a conversation I found to be really meaningful. [01:19] And I think one of the most important aspects of it is Nabil's contention that a critical part of creativity is a kind of strangeness or surprise or unpredictability.

[01:30] The second big topic is his rigorous approach to thinking, learning and understanding things more deeply, regardless of the domain. [01:38] Obviously, both of these ideas tie in deeply with what's happening in AI, and that is a backdrop to much of the conversation. [01:45] We spend the second half of the conversation covering a wide range of things from Palantir, government bureaucracy, Doge, gray areas, technologists in power. [01:54] art and why it's meaningful and why it matters. All the things Nabil has learned from Tyler Cowen and many other topics, which all just do more to showcase how wide ranging Nabil is. [02:05] More than anything else, I hope the conversation leaves you wanting to put more intention into what you do, what you make, what you think about. [02:14] how you approach your life. [02:15] Something I've talked about in the past with other guests and that is really meaningful to me is this idea of fighting inertia. [02:22] How do you fight the default? How do you fight growing friction? [02:25] How do you fight a world in society that is increasingly full of slop? [02:29] There are a few patterns and solutions that we talk about here, but ultimately, I think the goal is a life of growth, [02:36] of practice. [02:37] and of care. [02:38] With that, [02:40] Here is Nabeel Qureshi. [02:41] Hello. [02:42] It's good to be with you. Thank you for making this happen. Thanks for having me. You have, I texted you this before. [02:48] There are like so many versions of an interview I view, I suspect, even on a few of them I listen to. Very, very different. [02:55] And maybe, who knows, maybe we'll do more of these in the future, but I'm excited to pick a few topics that I think are particularly interesting.

[03:02] interesting and [03:03] Feel resonant about you today. Yeah, somebody once put me in a group chat, which was just like people with confusing vibes. [03:11] I felt like I belong. I find a lot of it coherent. I just find it just the amount even just scrolling your Twitter timeline. It's just like, oh, my God. Okay, I got to read it in. No more. No more research. [03:21] I want to start [03:23] with a topic that I think has been coming up in a, [03:26] Bunch of ways for me recently... [03:29] And there's one thing I think of a lot. There's this film called Pig with Nicolas Cage. If you haven't seen it, there's this iconic scene where he's talking to a restaurant owner who's like, done it. And he has the nice restaurant that everyone likes. [03:43] And this guy used to work for him. And he's like, whatever happened to that pub you were going to open? You really wanted to open it? He's like, what are you talking about? [03:49] And Nick Cage goes on to say, we only get a few things to really care about. [03:55] You wrote a tweet thread recently titled, The Opposite of Slop is Care. [04:00] And to me, this is this interesting amalgamation of so many things that feel particularly relevant today. Obviously, AI. [04:07] I think broadly social media and algorithms... [04:10] On the other hand, conversations about craft and taste and what's going to be human, all these things coming together. [04:15] to start [04:16] It might be worth just talking a little bit more about slop. I wanted to read kind of the first chunk you had about that. You say... [04:23] Slop is probably one of the most important and rich concepts for understanding modernity. [04:27] Slop is four things. It's efficient, mass produced and low cost. [04:31] done carelessly, thrown together without fussing over the details, and ahistorical, not rooted in tradition or practice.

[04:37] And you go on to say mode collapse refers to LLM models losing their creativity after being post-trained. [04:43] RLHF, SFT, etc. Slop is what happens when mode collapse happens to society and culture at large. [04:51] And then just as a simple example, so people have a reference point, it's amazing. You say, lemon gochujang creamy miso pasta is slopped food. [04:59] Spiritual pseudo-Western self-help Buddhism is slop religion, [05:02] Most modern new builds are slop architecture, [05:05] Netflix is Slop TV. GBT Poetry is Slop Poetry. [05:08] AI artists, slop art, et cetera. [05:10] This probably resonates with people who've heard, I think Venkatesh Rao has this term, premium mediocre. [05:16] Or just this broad feeling... [05:18] There's examples in fashion. There's examples even in the idea of like seeing like a state in a way that we sort of add needless order to things. [05:25] My friend Eugene Wei, who I just had on the pod, he talks about how social media is sort of pushing us to this frictionless positivity. [05:31] So I'd love for you to talk a little bit about how you actually see the pattern. [05:35] how long it's been happening. [05:37] And potentially like what the terminal state of at least the direction it seems like we're going in is. [05:43] Yeah, wow, there's so much there, right? This is one of those threads that I started writing, and then more and more things kept coming out. And you know how sometimes you just pull on a thread and then suddenly you're like, oh, there's a whole essay buried in here. Yes. So there's a few different things going into it. [05:59] I think one is just the work of Christopher Alexander, the architect and writer. He wrote these books. [06:05] There's the nature of order, timeless way of building, pattern language.

[06:09] And I think he basically is trying to answer the question of what went wrong with architecture in the 20th century, right? Because in his mind... [06:16] You have these amazing things. You have gothic cathedrals. You have... [06:19] Yeah, just all the kind of religious buildings of the 16th and 17th centuries. [06:23] You have beautiful mosques, etc. And then if you kind of fast forward to today, if you look at a lot of the buildings in [06:29] urban metropolises, it's like [06:31] that kind of slop, right? Like visual slop. And then for him, it was especially disturbing because he was a, [06:36] architecture and structure at UC Berkeley and what he observed was that the students are being taught to generate slop as well right so for him it was like [06:45] the scholarly culture had degraded. And the scholarly culture is the thing that should be preserving the craft. Like, if anywhere preserves craft... [06:54] without pure utilitarianism, it should be the scholarly culture. So for him, it was like, oh, we're also teaching sloth. [06:59] So this really disturbed him, and [07:02] I mean, long story short, he comes to the conclusion... [07:05] That. [07:06] There's this kind of religious awe... [07:08] underlying true creation [07:10] And if you don't tap into that, [07:13] when making a building, then your building is kind of dead spiritually. [07:17] uh and and you know he's he's a catholic right but like he's not very overt about it and uh [07:22] So this thread has kind of been... [07:24] humming in my life for a while right it's like what kind of world are we creating and then [07:28] You have AI creation. I think the specific thing that triggered that thread was I saw this study that got kind of re-upped and it was [07:36] how people rate ChatGBT 3.5's poetry way better than all these like famous Western poets like T.S. Eliot, M.P. Dickinson, Walt Midland. And I was like, this is so disturbing. And then I also cite this amazing article by this guy, Guern, who says that he feels like there's this huge risk of feedback loops here, where basically you already have like students who

[08:01] aren't really reading the way they used to anymore, right? Like, talk to university professors and they complain about this a lot. [08:06] And so the risk here is like we basically just forget what good looks like entirely. [08:11] And we stop producing... [08:13] you know, good writing, good buildings, [08:17] good anything and um so yeah that that was what the thread was about and there's just so many different directions it kind of applies in right well you've you've gotten right to where i want to go [08:26] One initial question would be, [08:29] There's some frame of this, or at least some interpretation of part of what you're saying that says, you're just like an elite coastal... [08:35] douchebag who's telling people they can't like mainstream or normie things, at least in certain categories. You can imagine [08:41] the classic like A24 versus Marvel argument or some of these categories. I think in some areas, obviously, like the – [08:48] I don't know, people always give the minimal telephone booths. Everyone kind of acknowledges the new telephone booths suck, but... [08:55] One, I'd be curious for you to riff on that, maybe to lead you a little bit. You mentioned the Guernet essay. You say in his essay towards benchmarking LLM diversity and creativity, you're [09:04] He speculates that slop could create this terrible feedback loop, as you said. People, more slop, people start preferring slop. Even more slop, people forgot what really good stuff looks like. [09:12] Anything that you read that describes things in a familiar vocabulary and doesn't really go beyond that. It's just news. It's just something that will come and go. [09:21] Anything that's truly compelling, there's a quality of strangeness to it. [09:24] Howard Bloom says this is the essential characteristic of art. [09:27] I think I combined those two things. The second one wasn't the Gwern bit, but this notion that on one hand, it's easy to critique normie or mainstream things just because they're mainstream.

[09:37] On the other hand, maybe there actually is this risk of, [09:41] Society in 50 years simply won't appreciate [09:43] things that are cared for is it that strangeness is it something else like how do you how do you think about and i realize it's a few different questions rolled into one yeah yeah yeah i totally get what you're saying so i don't think it's a mainstream versus niche thing at all actually there's a lot of like niche avant-garde art that is basically slop in my mind and there's a lot of mainstream art that's full of care [10:03] And one is one one interested. So I'm very interested in movies. Right. And so, for example, like if you look at the Terminator movies, like one or two, they're both amazing. Right. But there were also like Hollywood blockbuster hits. [10:13] And they were made with Kale. I think it was James Cameron directed them. And you can tell he... [10:19] put a lot of thought into like how the robot should look and all of this kind of stuff, or like casting Arnold, Charles and I agree. Yeah. And so that stuff is really good. Similarly, [10:28] You know, like E.T., The Matrix, The Godfather, these are all mainstream smash hits, but [10:33] The distinguishing characteristic to me is they were made with care. Even, I think, the original Transformers movie with Michael Bay, right? Pretty good, actually. Like, you can tell he put care into it. Whereas, you know, Transformers 6, like, feels close to the spot. Maybe even the same about Marvel, by the way. Iron Man 1, like, clearly made by people who really cared for the IP and the whatever. Yeah, yeah, yeah. So... [10:54] So to me, it's not really an elitist thing. Just to take the flip side of that as well, I think if you open... [11:00] you know, the Paris Review or like a contemporary lit mag nowadays, you will often read [11:05] very mid-contemporary poetry, but like clearly like elevated art contemporary poetry. And I tend to think that's sort of the slop version of Elige taste. Like on occasion, it's pretty good, but

[11:15] Yeah, so it's a general phenomenon. I do think it's like... [11:18] There is... I hate this word, taste, but there is some... [11:23] thing that is linked to taste and care that good things have in common. [11:28] And they can be both mainstream things or they could be not mainstream things. [11:33] To that, I kind of combined two things. But to the Howard Bloom idea that maybe the essential characteristic – [11:40] could be strangeness, at least of art. [11:43] If you were to try to pinpoint, [11:45] We'll talk more specifically about care and how to get to it and what that might mean, but... [11:50] I can't help but wonder if some people still hear this and be like, okay, I kind of like, is Slapa, you know, when you see a thing? [11:57] Is the opposite of slop, is it the fact that it can be surprising or that it can be moving? Is it a spiritual thing? Is it a personal thing? [12:03] Yes, yes. I do think surprisingness is an essential thing. So, like, I think if you stick with the mainstream thing a second, like... [12:10] The Matrix. [12:12] Very surprising when you first saw it. Right. Like, oh my God, so memorable. What was it? I know Kung Fu. Or like the whole thing dissolving into zeros and ones and green numbers, right? [12:22] All very strange and very cool. It got incorporated into mainstream kind of parodies so fast that I think we're all numb to our strange ones initially. That's a great point. Harry Potter, really strange. Like... [12:32] Set in a boarding school, four houses, like Expelliarmus, like all these Latin spells, right? So I do think the... [12:40] And there's like a simple like information theory concept underlying all of this stuff, which is just... [12:44] like surprisingness in information theory is this technical term right it's like in a given string of bits like how many

[12:51] how predictable are those bits and [12:53] I think, I think, [12:54] it's a necessary condition of good art that [12:58] It's not that predictable. [13:00] And I think this is... [13:02] an issue. It's not an insurmountable issue, but it is an issue with AIR, where it basically goes for [13:07] things that are quite [13:09] predictable sort of by the nature of the way it works right and what i guess one last thing that i have on this theme was [13:16] There's a poetry professor in England that used to make his... [13:19] students do this exercise where he'd take a poem by, in this case it was Philip Larkin, but it could be anybody, and he would basically blank out a bunch of the words, and he would make his students guess them. [13:28] Basically, almost none of them would ever guess any of the words, because the words used were all... [13:32] just really strange. Like, there's this one line in this famous in England poem. I don't think it's that well known in America, but the line is... [13:39] The guy's going by on a train and then the line is like, a hothouse flashed uniquely. [13:45] And he blanks out the word uniquely. And no one ever gets the word uniquely. It's like, you know, you might say a hothouse flashed. [13:51] rightly or whatever it is but if you look at it in the context of that poem it's perfect because [13:56] the whole point of the poem is like, [13:58] this is one very specific day that's never going to come again. And all days are like that. And so the word uniquely works very well there, but you could never predict that in advance. Yeah. So again, I think it's, [14:07] The opposite of slop is care, and I think part of [14:10] things that have a lot of care put into them is that [14:12] there often is something really surprising or delightful about them. Yeah. [14:17] to talk more about it specifically. [14:19] You already brought up Christopher Alexander, excuse me.

[14:22] You specifically say, I just loved your framing. You could read all of Christopher Alexander's work. [14:28] As saying modern architecture is slop, the opposite of slop is care. To create anything worthwhile, you basically have to put God into it. [14:35] There's another, I think from something else you wrote, maybe this is from Principles, you talked about Aim for Chartres. [14:41] Also, an Alexander idea, this idea, when doing something aimed to be the best there ever was at it, this compensates for your natural bias, which is to do something mediocre. [14:50] You have to really aim to be as good as the greats. [14:54] maybe in the same way that the slop idea is a little bit hard to totally pin down, [14:59] Care is a little bit ineffable, too. Alexander specifically uses the phrase, the quality without a name, at least in a timeless way of building. [15:08] How'd... [15:08] How do you foster this? How much maybe even control do you have over what you care about? [15:15] Wow, yeah, that's such a good question. There's this really good Slate Star Codex essay, one of my favorite essays of all time, it's called The Lottery of Fascinations. [15:22] I've never heard of it. It's amazing. He basically goes into... The thesis is just like, you are basically born with these things that you are very interested in, and it's very hard to change that. And the way you... [15:33] Detect it is like, what are you naturally good at? Or like what naturally sticks in your mind, right? [15:37] And he uses the math example. He claims he's bad at math, which I'm skeptical of. [15:42] I do think this is true. You are... [15:46] Your interests do feel a little arbitrary to me. [15:48] And I think one reason I think about this topic so much, right, is like I grew up with computers. I'm like both of us work in technology.

[15:56] I grew up playing video games, but I also have this other side of me. [16:00] I think partially maybe from, you know, growing up in England and in a particular culture, but like, I'm very into art, I'm very into words, I'm very into poetry and like old beautiful things. And [16:08] how do I reconcile these two sides of myself, right? [16:11] And one answer is video games, right? So in the 90s, you have the Final Fantasy video games, you have... [16:17] Metal Gear Solid. You have these amazing video games that I think [16:21] have a good claim to being art in some ways. Zelda's another example. [16:25] But then it feels like subsequently things have... [16:27] tailed off a little bit like it's not clear to me that the games now are that much better than the games being made [16:32] then they're better in some ways like I love Elden Ring but um [16:36] Yeah, like, what about, like, bits, zeros and ones is... [16:40] eternal in the way that you know a cathedral door in the south of france is eternal is any of it eternal or is it all just like written in water right um and [16:51] I think one of the most compelling answers I've seen to this actually was [16:55] I'm curious if you know about this. There's this ruby... [16:58] programmer called [17:00] He was he was pseudonymous. So his online moniker was why the lucky stiff is like underscore why? No. Yeah, he's quite well known on like Hacker News and in the kind of Ruby programming community. And he he was famous because he wrote all these like how to program in Ruby guides. But they were they were written as like comic books and delightful comic books, like very artistic ones. But it would be like, you know, how do you do a for loop in Ruby or whatever? [17:23] And, um, [17:25] At one point, he...

[17:27] basically committed, I guess they call it info suicide, where he deleted his entire online presence, deleted everything. And, and, and, [17:34] He then started... [17:36] I don't even know how to describe this, but he started, like, I think it was, like, faxing [17:40] written... [17:42] stuff, like written words and drawings on pieces of physical paper to this like particular... [17:47] thing that was hooked up to the internet and then basically like published it in a script somewhere and it went like viral on Hacking News right and um [17:55] It was called Wise Prentice Fool. [17:58] and [17:59] He had been absent for a while. [18:00] Yeah, he'd gone dark and everyone was like, is this guy okay? Did he kill himself or something? And no, he'd just gone dark and decided that the internet wasn't for him. And the whole... So it's basically, I think it's like 40 pages of... [18:12] written it's like sort of a series of short stories almost and like little poems and little drawings [18:17] And the theme is just like, can anything in technology be eternal? He was basically like, have I wasted my life? Have I spent 15 years writing code that is just going to get... [18:25] you know, obsoleted, whereas like [18:28] He says he reads Kafka's work end-to-end, and he's like, has anything I've ever done been as important as the stuff that Kafka wrote? [18:35] And he basically concludes no. And so... [18:39] I thought about that a lot. Like to me, ironically, this is like one of the... [18:42] few I think enduring art artifacts that came out of the tech world right which is like the wise princess sport but [18:49] Yeah, look it up. It's really fantastic. [18:51] Wow. [18:52] It's interesting too, just thinking about the video game example briefly, [18:56] Much of the

[18:58] things that have seemingly given way to [19:01] the over just IPization of everything, tons of sequels were, [19:05] You talk about Harry Potter, like... [19:07] Even some of the Marvel stuff. [19:08] Video games of the 90s. [19:10] They were super new and surprising or strange. [19:14] And we basically copied that template for the past 20 years. That goes back to the Gorin idea, I think, which is just like, [19:19] Our temptation and our default and our incentives today [19:23] Archer just play the hits over and over and over again. [19:26] Maybe that's something, to tie it back again to the... Is it why? Why, yeah. Like the question... Yeah, maybe there's something around... [19:33] digital mediums too that especially wants to do that it's it's interesting right yeah like copy and replicate because they're really good at it it's frictionless to copy exactly yeah and and and i'm not a doomer about this to be clear i think it's just early like i think i mean computers 1980s that's not that long yeah yeah um there was this recent latvian movie that i think won an oscar it was called flow and i haven't still haven't seen it but yeah but the yeah the animated one [20:03] and supposedly like a beautiful work of art so [20:06] I think all this stuff is possible. It's just like... [20:09] It's going to come from unlikely directions, like... [20:11] Latvian team or whatever and [20:14] It's going to be made with car. Mm-hmm. [20:16] One... [20:17] Maybe this is also a leading question, but on the idea of fostering it or even choosing what we care about. [20:23] One kind of connection I made [20:25] you have this awesome parable from Zen, uh,

[20:28] motorcycle maintenance about the first brick of the opera house. [20:32] And maybe pointing at this idea of sort of how do you [20:36] How do you really look with intention in a way that allows you to see past the cash or see freshly? [20:42] Another quote, I think it's from a photographer, that same essay you say, "When in doubt, go closer." [20:48] Do you think that... [20:50] to the extent there is any control, [20:52] it might start with just attention or with with [20:56] intimacy with something that we're not [20:59] I don't know if I wanted to become one maybe relevant example for you, [21:03] relative to most people is people don't read anymore. Long form text is something that [21:09] In general, people don't read, and even, I think, otherwise interested, tasteful, smart people don't do much of. [21:15] If I wanted to, and maybe you just have a DNA that says culturally or otherwise or instinctively that you're going to care so much about that. [21:22] Shakespeare. [21:23] Do you think that [21:25] Something like this? I don't know. If I wanted to get more into Shakespeare, is it [21:28] It is... [21:29] Is there a template there around the brick? Like, is there a template to try to just zoom way, way into something really close or spend a lot of time with it? Am I reaching? [21:37] Yeah, no, I think that's right. I've often wanted to teach this [21:42] like almost like a workshop on like [21:44] original seeing or something because I feel like it's so easy to actually [21:48] do but but like people need these paths in and i don't know if you've read the book in pro jackson but it's one of my favorite books by this is the the johnstone and it's the palantir like default yeah exactly and um i don't think they give it for this reason but

[22:04] The book starts, I think the very first sentence is like, as I grew older, the world got more and more dull. [22:11] Like it got grey. [22:12] And then he basically talks about how he brought color back into the way he sees things. [22:17] And it's through these wacky tactics. Like he basically walks around counting backwards from 999. [22:25] while also like repeating a word in a different part of his brain. And then he's like, after a while of doing this, the colors got brighter. You know, like he tries all these ridiculous tactics. Another one that he's really into is... [22:37] So it's like a guided improvisation. So you'll have a student lie down and close their eyes. And then he'll be like, you... [22:44] Walk up to... [22:46] an ancient rock wall in a cave and you peek really closely [22:51] and you see some writing, what does the writing say? [22:54] And then he says often the students blocked creatively. They've blocked themselves. So they say, oh, it's too blurry. I can't. And then he says, OK, you get out a magnifying glass from your pocket. More specific, more zoomed in. Yeah. Now read out what it says. And like if you yourself try this, you will come up with the strangest stuff. Like you will come up with. [23:12] what feels like original poetry but it'll feel like you're just reading it off of your [23:16] like the retina of your imagination or something. [23:18] And it's just full of these interesting exercises. And I've kind of got this list of them privately and... [23:24] I just feel like it would be a delightful workshop to teach because I think it's, this is the key is like original seeing is your birthright as a person. [23:31] We spend all our time scrolling or in front of screens. I spend 18 hours a day in front of screens. I'm working on my laptop. I'm on my phone, whatever.

[23:39] And I don't make time to do this, but I think if I just did, then... [23:42] The original scene is easy to access. It's like, what's hard is like the consistent... [23:47] doing it again and again and [23:49] producing an artifact and iterating on it until it's really good. Well, and we're also used to getting these things that are so prepackaged and compressed and legible. [23:58] that we've maybe lost some of that muscle to pick up the proverbial magnifying glass. [24:03] Yes. [24:05] Thinking about this, obviously, over this whole kind of slop and care thing is AI. [24:11] On the creative side, [24:13] I think one sense is... [24:15] We are reducing the friction to create. [24:17] so dramatically, [24:19] Conversely, the amount of creation that can happen is so abundant. [24:24] How do you think about [24:27] care specifically in the sense of sort of like leaning in when going the extra mile, you can literally get carried there in a few seconds by an LLM. [24:34] A couple of riffs on this that I've thought about is like, [24:36] Is it maybe where you stop [24:38] the frame you have sort of like infinite frames in front of you and it's which frame you pick out. There was another recent tweet you had, [24:45] about like the things we come back to. [24:47] Any idea you write down in the notes app and tell yourself you'll visit later is probably not that good of an idea. The really good idea is fill your body with excitement and make you want to start working on them immediately. [24:57] What do you think the template is for creativity when, again, anyone can do anything really, really quickly? Maybe I'm over-rotating the fact that you can generate a mediocre sloppy image today really quickly, but it seems like that's only going to get more...

[25:10] at least as a baseline, like the [25:12] these quote unquote finished product that an AI can produce. [25:16] is going to be at the level of somebody who quote unquote cared 50 years ago or 20 years ago or five years ago. [25:21] Yeah, it's a really interesting question. [25:24] I guess one lens on it is like creativity is basically generation and selection, right? So you're generating a bunch of ideas. [25:31] your brain rejects some of them consciously or unconsciously, and then you select some of them and put them down, and then you basically iterate this process. And so one simple lens is like, okay, the AI is actually very good at generation, right? [25:44] but then the human is the person selecting. And so you could make AI sort of hybrid human AI art, but the human is the ultimate decision maker. So I think the model here is something like, [25:53] There's this artist, Damien Hirst, who, you know, he makes all these skulls and everything, and... [25:58] He actually is controversial, right? Because he has this basically factory of like... [26:01] you know, interns producing the art, and he kind of just like rubber stamps them at the end and goes, yeah, but like, [26:06] Maybe that's fine. Maybe, you know, it passed through the Hearst filter... [26:10] And he's rigorous about it meets his standards. And so who's to say that that's a bad thing? [26:15] So that's kind of one model, right? But then, okay, so in AI art max list, if I say no, but eventually the AI will learn what good taste is, and then the AI can do both the generation and the selection, and then... [26:26] you know, what role will the human play? Well, and I'm not, like, there's the separate litigation of, like, what role does the human play? [26:34] This may be you often use the example of AI playing chess against itself. Yeah. Like there's the separate point, which is just how good will the purely AI stuff be?

[26:42] What I'm maybe more interested in is just what it actually substantively looks like [26:48] Care used to mean going through all the additional friction and time. [26:53] that a process would take or that craft would take. [26:57] Painting the back of the cabinet, whatever these things are. [27:00] and at least digitally but maybe even beyond that it seems like [27:03] If anything, the thing that technology and AI is really good at is speeding up the process and reducing friction. [27:09] And so I almost like, what does care mean when you can just do everything really quickly? [27:13] I think you're hitting the core of this issue, right? It's like... [27:17] There is... [27:19] Again, it could go one of two ways. One is just like... [27:22] there is inherently something to just, you know, sitting with a piece of wood in a shed and like painstakingly kind of whittling it into a beautiful thing. And that, [27:30] element of crafter is only going to get more important in the AI age. And so everything that can be automated will be automated, but the things we really value are still going to be like, you know, when, say your daughter draws you like a birthday card, right? That's always going to be more important than she generated one with AI. [27:45] But, you know, I find scary the other version, which is like, actually, no, the AI ends up capturing... [27:51] everything that you care about better than you and you're kind of just like uh [27:56] Here's my AI generated thing. Oh, well, here's mine. It's even better, you know, and it all becomes a little meaningless. But I don't think... [28:03] We are close to that, and I don't think we're going to get there for a while because... [28:07] I think the physical world and physical artifacts are... [28:10] still extremely important to us like for all the generative ai stuff that we have it's not like the

[28:15] buildings that we are building are any more beautiful too for example so maybe one answer is simply that we will just shift a bit more towards caring about the physical world yeah that's a good thing i think yeah yeah you recently said any median level work task that involves clicking a mouse or typing on a keyboard is going to be fully automatable in two years [28:30] price that i enact accordingly conversely though [28:33] Like what are the areas where we have... [28:37] more weird, strange, human moat, whatever. A couple of ideas I liked that I think are pretty recent too – [28:43] This is why celebrating the peaks and enshrining human values, cultural criticism becomes more important as the world becomes more slopified. [28:51] We already mentioned Harold Bloom for literature, Tyler Cowen for his blog, Men's Wear guy on Twitter for clothes. [28:57] There was one other bit where you said in the long run, the human moat against AI is that it's is going to be great books are an art. [29:05] religion and love [29:06] The AI can't hike the Pacific Crest Trail for you. It can't fall in love for you. It can't pray for you, et cetera. In Dune, the Fremen's learn an irregular, jagged way of walking on sand. [29:15] to avoid sandworms. [29:17] Maybe more personally, are there areas in particular that you're [29:21] finding your human moat or at least leading into or feeling like, [29:26] Almost like the care feels strong. [29:28] Yeah, there are a bunch. I think there's a simple answer, which is just, you know, human presence was going to matter a lot more, right? So like the fact that I'm on this podcast, like I used to say no to podcasts. And at some point, I was like, hang on, like, I'm gonna have to get good at [29:43] You know, talking and verbal fluency and having interesting ideas, because like that is going to be my comparative value in five years. It's not going to be writing Python or even even even maybe starting companies. Right. I think AI is going to get better and better at that, too. So I think there's some aspect to it.

[30:00] live presence and live performances of any kind that are going to be very important. [30:05] I think for myself as well, you know, just like [30:09] It's important to spend time with the arts, man. And actually, it's funny, this is happening across tech now, right? Like, a couple months ago, everyone in tech was reading Middlemarch. There's, like, books that kind of just, like, sweep the zeitgeist and... [30:21] uh you know people are doing more of that and so yeah like read great books [30:25] To me, like, I will write things like poetry or, like, prose poems and just, like, privately enjoy doing that. [30:31] But yeah, I think it's just important to have a practice. And it's okay, like... [30:34] whatever kind of practice it is, but it's like something where you're exercising [30:37] the faculty of care you're not just typing stuff into claude and like pasting whatever it comes out with into something else [30:44] Practice is a great word, too, to the earlier point around [30:48] Practice implies friction. Yes. Often, sometimes even deliberate friction. Yes. [30:53] Yeah, yeah. And just one more thing is like, we were discussing off this previously, like the book Seeing is the [31:00] The Robert Owen. Yeah, yeah. Right, yeah. And... [31:03] I feel like that entire book is about this theme, right? [31:06] It opens with this very physical description of him. [31:09] driving around the LA desert and, you know, drinking kind of, [31:11] Coke from the soda fountains at the diners and... [31:14] Do more of that, right? And also his entire career is just... [31:19] He kind of masters one art form and then... [31:22] he gets bored of it and he moves on to something else. And he's like, what if I made art by putting like long wires in the desert and then kind of framing them and taking a photo? Or then he's like, Oh, I'll make a garden. It's going to be the best garden ever. You know? And so I think it's important for you to just,

[31:36] constantly be pivoting as soon as you... [31:39] feel like you've mastered something or if it's become sloppified, it's like go find something where you... [31:44] can exercise more care and get [31:46] something better out of it. [31:47] One of my favorite parts of that too, one, as you hinted at, one major theme for him is just presence. He's like, it's not about the ideas behind the art. I think he literally says, I want it to like bang you off the wall. And there's another bit where he's talking about, I think it's like the late lines. [32:02] And he's talking about [32:03] How... [32:04] He's just getting more and more interested in questions over the course of his career rather than answers. He's like, I'll kind of answer it a little bit and do my little performative artwork and show it off. And then I'm ready to move on to the more questions. [32:15] It feels human-mode-y a little bit. Yes. Another example is, to the presence point, is meditation, right? Like, go get into the jhanas or something like that. Go on a journey retreat. Actually, very good. Like, I do think [32:27] there is this question of can AI ever be conscious? Like maybe, [32:30] I think it's a genuinely hard question, right? We don't know, but... [32:35] What we do know is that we are Funches and we can access these... [32:38] marvelous internal states through things like meditation so maybe go and invest in that just things where like [32:44] the differential value of being human is very high. Yes. [32:48] My last... [32:49] note on this theme. [32:51] There's this phrase that you've used that others have used. [32:54] AIs that care about humans. [32:56] I think Ilya Sutskabar even talked about this metaphor of sort of a loving parent. [33:02] running in the background of this whole conversation is like is care a distinctly human trait

[33:07] Is it the, maybe even the human moat? [33:10] you you have a tweet or somewhere you called claude 3.6 the only ai model with the [33:16] The quality without a name. The Christopher Alexander idea. [33:21] And then you had even proposed this world where [33:24] AIs care really deeply about humans. You said AI agents that want to be moral and care about the human species. The cornerstone paper of the current AI paradigm is titled Detention is All You Need. [33:33] The cornerstone proposition of AI alignment theory might well be that love is all you need. [33:39] That's probably a whole can of worms that we don't need to get all the way into, but is... [33:44] Is it oxymoronic to talk about AIs that care about us? Or is that actually maybe the point? Is that the sort of heightened form of... Is that the most human thing that we should want the AIs to have too? [33:54] Yeah, I was being a little tongue-in-cheek with that one, but also there was a serious thing there. And I think people have come around to it a little bit since, you know, Claw 3.6 came out, right? It's like, I think the original... [34:05] kind of Yudkowskyite view is that AI is going to evolve in this way where it's not going to understand human concepts, right? So ethical concepts like love and care and goodness, we cannot write down a formal description of them. And so how could an AI possibly learn those, right? And then you had Claude 3.6 come out. [34:23] I mean, I don't know. It seems to understand morality better than like most humans I interact with on the street sometimes. And you think distinctly even relative to Cloud 3.7 or other more modern models. It nailed some human responsiveness that I think is like very special. I don't know how I don't even know if they know how they did it, to be honest, but it did have that.

[34:44] And so I think... [34:46] It's a little bit like, if you think about, does AI understand things truly? [34:51] Okay, like... [34:52] a few years ago, it was easy to say no, it's just sort of regurgitating. But then eventually got to the point where like, you can ask as many questions as you want, and it's going to give you the right answer. And then you do have to say like, I think it does understand this thing, right? Even if it's like, [35:04] basically a perfect simulacrum of understanding and understanding are essentially the same thing in practice and so i think this is the case with [35:12] care to, as weird as that sounds, is that [35:16] eventually I do think you get [35:18] AIs that we functionally treat as though they are beings and [35:23] we basically say, yeah, I mean, this thing cares about me. And it sounds less and less weird. It sounds weird to us because we're, you know, millennials or Gen Z and, you know, we kind of remember the world before AI, but the next generation... [35:34] I don't think it's going to sound so weird. Now, whether, like, the essential thing that matters about caring is actually, like... [35:40] some deep inner process that's happening that can only happen in beings with souls like [35:45] I don't know. I don't personally... [35:49] think that but it's an interesting debate. [35:52] The next thing I want to talk about is... [35:54] a bunch of ideas running through one of my favorite essays you wrote how to understand things [35:59] And the core idea there is this you call like the will to think, which I think comes from either William Shockley or Enrico Fermi. Maybe Shockley represent. Okay. Okay. [36:08] There's this bit you say... [36:10] It's like almost an honesty or an integrity, a sort of compulsive unwillingness, an inability to lie to yourself.

[36:18] Also this sort of feeling of like this... [36:21] incessant need to go deeper until you have high resolution on an idea or a concept. [36:26] To me, this rhymes a bit with care. [36:29] in maybe a different way. [36:31] It's pushing past the default, the acceptable, the cash, the expected. [36:36] So one question would be, do you think that connection is... [36:40] There is something there, and then beyond that, maybe... [36:43] One could argue that this will to think would be less meaningful in a world where intelligence is truly abundant. [36:49] Do you think that's right? Do you think that's... [36:52] Maybe it's a serious overreach, but I could maybe argue that over time, [36:56] the care piece is going to extend, but the will to think or the will to understand is increasingly less important. [37:03] That's a really fascinating thought. Yeah, I think you may be right. Um... [37:07] That one was interesting. You know, like, I... [37:09] You know how sometimes you sort of tweet out of your own inadequacy? Like maybe you say like, yeah, respond to your emails faster. And really you're talking about yourself, right? Because you like let that thing hang for three days. [37:18] uh that's kind of what that was the best tweets what are you talking yeah yeah right like and i think that essay was written [37:24] as an exploration of my own inadequacies. Like, I think it literally starts with my friend from high school who was amazing at [37:30] physics and math, right? And [37:32] I saw him do this. Like, he was just not happy with any of the answers that he was given in anything. Like... [37:39] I didn't talk about this in the essay, but, you know, he's a physicist now, actually. He's a professional physicist. And... [37:44] I saw him do economics. We did economics together. [37:47] And he had the same thing. He would just take Keynesian theory apart and he'd be like, no, why are these things worse? I would be kind of focused on just learning the thing itself. I'd be like, okay, Keynes said GDP is made of these five things. All right, let's understand what that means.

[38:00] But he would be like, why? And he'd be like tearing it apart and trying like different ways of calculating GDP and [38:05] it was just like a different the next level up for me right and so [38:08] This phrase will to think when I found it, I was like, ah, that's what he has that I don't have. This also becomes very clear if you play a game like chess. [38:16] Like... [38:17] Very good. [38:18] chess players just have more of this and what you realize is that [38:23] Thinking is actually a very painful process. Like your brain will do anything. [38:28] to minimize resource consumption and to avoid thinking. And so most of chess instruction is in basically hacking around this tendency of your brain by forcing you to think for longer. Your brain will just go, oh, yeah, this is the move, you know, and then. [38:43] actually if you're being rigorous about it no you have to check what if your opponent does this or what did they do this and you just sort of don't want to think about that after a certain time you're just like oh this move clearly works like why do i have to be pedantic and [38:53] tackle these things. And so I do think the most... [38:56] Intelligent people I've ever met. [38:58] just have an abundance of this trait. I don't think it's necessarily, like, a fully adaptive trait, by the way. Like, I still don't... [39:06] I'm starting a company right now and I can't go down every rabbit hole I see. I can't try and understand everything. I just have to make a snap decision based on 60% of the information. You talk about scientist kind of orientation versus founder orientation. Yeah, exactly. And just anybody who operates in the world and has to kind of go about their day, you can't stop and be like, [39:24] what is the air made of really? [39:27] It's an interesting question, right? [39:29] So yeah, and then I think to your point, like

[39:32] You know, I'd like to think that learning stuff is going to remain important, but I think increasingly it's going to be more of a leisure activity than it is right now. It's like you actually really do need to understand stuff. Otherwise, the trains stop running and the computers stop working. Yeah. [39:45] And as more and more stuff gets offloaded to the AI, I think understanding stuff is going to be more of a thing we do for fun. [39:51] So. [39:52] You say... [39:53] School kills the will to understanding in people. [39:56] on some level, [39:58] are [39:59] Our societal systems and inertia are [40:02] kind of like structurally just grinding out both of these things the will to think the will to care [40:08] Right, yeah, I think so, because... [40:10] that's sort of opposed in that goals, right? Um, [40:13] School is sort of trying to get you to... [40:16] pass these tests and get the highest possible score, [40:19] And so in order to pass the test, [40:21] It needs to cram you with all the concepts. [40:24] And so the most adaptive thing to do is to learn the concepts and then pass the tests and get 100%, which is basically what sort of I found myself doing, right? [40:32] In some ways, as I was saying, the will to think is actually counter to this because you're [40:36] You're basically questioning absolutely everything, right? [40:39] And it's not adaptive to doing well in school at all. [40:42] An interesting example of this is in [40:44] modern geometry, there's this French mathematician called Alexander, I think it's Grosendie. Grosendie, yeah, and [40:50] You know, he's sort of notoriously... [40:53] He would spend, he spent like five years just being like, what is a line? [40:57] like what is this shape and then what is what what is volume right and [41:01] Everyone else would be like, well, volume is defined in this way. And he'd be like, no, this doesn't make sense. And eventually he invents a whole new field of...

[41:08] of geometry and mathematics but um i always think of him in this way because he also went completely insane [41:14] Like, he fully went crazy and... [41:16] ended up writing all these, like, [41:18] kind of schizophrenic religious tracts in a forest and eventually he kind of died like [41:23] knowing no one and destitute. [41:24] So I think this trait is very double-edged. It does mean that you're not necessarily the favorite child in school. [41:31] You could end up in like very strange places, but it's sort of high variance. There's an amazing book called When We Cease to Understand the World that he's featured in amongst. But it's that pattern over and over again with people kind of on the edge. Yes. [41:44] One of my favorite things you talk about in the context of education is why video games are really relevant as both a practical way to learn and also a template for learning. I think it's also ironic given that. [41:55] what we were just talking about is sort of the [41:57] player playing the game on the field as it is laid out in the school context someone who did that i was i was a very good 90.000 student [42:06] which is ironic in that it's a bad game, and video games might be better. [42:10] You talk about... [42:11] simulation and [42:13] video games and particularly this like idea of feel [42:16] There's different ways you talk about it. Another one is like lines of force. [42:20] You give the example of a chess board the way that a really skilled player who's sort of [42:24] felt it in their body almost can see the pressure or feel the pressure that pieces are putting on [42:29] Another phrase you use is a high dimensional grasp. [42:32] of what's going on. Reality has a surprising amount of detail. A lot of these ideas kind of [42:37] revolving around each other

[42:39] And then you also, I think, just succinctly talk about video games as being this amazing pairing of simulation, [42:44] Plus really fast feedback loops. [42:46] The best way to learn Pokemon is to play. You're not required to learn all the... [42:53] The Pokémon's names before you start? [42:56] I think it's maybe from David Deutsch or maybe you... [42:59] even this final phrase of like bone level understanding. [43:02] There's one other idea, which is, I think, Tyler Cowen, motivation is that which is scarce. [43:07] And so I was thinking about a lot of this in the context of learning and video games and these challenges. [43:13] I know I'm packing a bunch in here, but you've also written about puzzles and like the idea that puzzles maybe improve our will to think in part because [43:20] We know that there's an answer to get to, [43:22] And so I was wondering about, [43:25] What it might be about high feedback loops. [43:29] that generate [43:30] this field, this bone level understanding. [43:33] And if maybe they not only help us improve, but they actually help us be more motivated. [43:40] Oh, wow. Yeah, that's really fascinating. Also, you are amazing at connecting ideas, by the way. Maybe a few different. There's probably some overload, but I'll give I try to think about it as just like I'm going to throw you a bunch of things that you can feel free to discard some of. Yeah, well, so Dose likes to talk a lot about explicit versus implicit knowledge. Right. And I think what he would say is like. [44:00] The textbook can give you the explicit knowledge, but [44:04] the implicit knowledge is always created by you. [44:06] And I think this is clear in entrepreneurship, actually, right? Like, one thing I say to people who...

[44:12] you know, want to say you're a student, you want to start a company one day. It's like, you know, maybe if you have the idea already and you're really keen on it, then go do it. But a next best thing is to just go and find a really good founder. [44:23] and follow them around and help them and just work with them for six months. Totally. I did this when I was just graduated. I joined this company in the UK called GoCardless, which... [44:34] ended up spawning a bunch of further billion dollar companies as well. And all three of the founders there were absolute monsters. They were incredibly good. And I just learned so much from that experience about [44:44] What... [44:45] A, what a good founder is, like what the actual bar for it is in a very visceral way that I would not have got from. Like I read every PG essay. I read every like what is a great founder essay and like just watching, you know, Matt. [44:56] operate for a day. [44:58] taught me way more than any of them right and also probably literally being next to him yes [45:03] In a bodily sense, the lines of whatever that, yeah. Sensing that aura and also just the things that surprise you, right? It's not always what you think. [45:12] Like... [45:13] One example is like, I think in your head, you imagine, oh, great founder. He's probably going to some conference here and then giving a cool talk there and getting drinks with Marc Andreessen after that. And it's like, no, he's hunched over his laptop and he has his to-do list app on his Apple Notes and he's just checking things off like a demon and he's... [45:30] typing these emails then he's like getting annoyed at the ops person because there's no drinks in the fridge like it's all this very banal stuff and um i think it makes you realize like okay [45:40] Succeeding is not [45:41] glamorous and

[45:43] you know, like [45:44] An example of a good founder is just somebody who does 100% of their to-do list every single day instead of, [45:50] what usually happens with 99% of people is they do like, [45:53] They have 10 items, they do maybe 5, and then it's 6pm, and you're like, alright, I'll do the other 5 tomorrow, and you kind of roll things over into the next day. He would just stay late until he ticked off all 10, and then he just did that consistently every day. And I thought, wow, like... [46:05] That's all it is. It's just you're very consistent and you're very high output. And I don't think I'd have got that from reading any number of online essays. Yes. So, yeah, it's just this reality is just very high dimensional and, you know, [46:16] There's no substitute for just immersing yourself in it. [46:19] And another unrelated example is just I get a little annoyed when I see people [46:24] talking about foreign countries and they've just never been there, right? Like you have strong opinions on whatever conflicts, but it's like you haven't talked to a bunch of people from there. You haven't like driven around the roads or taken a walk there. [46:36] And I think if you go, it's always way more nuanced than anything you read online or anything you read in the paper. [46:42] So yeah, that's another example. Just like the value of direct experience. And that last note, I even like, obviously classic, like map is not the territory. But I remember I went to school in L.A., [46:52] I've been there a little bit before. The first time I ever did a long bike ride around L.A., [46:58] I had a grasp for what the city was. There was... [47:01] three orders of magnitude higher than like staring at the map. [47:05] And it's, yeah, their amount of things that are like that. I think maybe the unique insight, maybe not unique, but the, the, [47:10] powerful insight you have is that video games actually... [47:12] to be pretty good at this,

[47:14] Or maybe it's not even video games. Maybe it's just like really thoughtfully designed games broadly. [47:19] I'm sure chess has elements of this too. [47:21] And in fact, we ended up opting into a lot of games societally in school or otherwise that aren't [47:26] that don't actually give you that multidimensionality or high dimensionality. [47:29] Right. And I think I think you're starting to see bits of this today. Like I actually saw a somebody posted a video on X the other day that was like, hey, I made this. [47:39] math video game for my son using ai and now and it was just a video it's like a five-year-old kid and he's [47:45] delighted by this, right? It's like every time he solves the math equation, it zaps a monster or whatever. So that's like, it's kind of cool. It's like, [47:52] I didn't know the AI thing was, I mean, I was kind of generally excited about AI, but I think the essay was written in 2018, right? Pre-chat GPT. [47:59] It feels like AI was the missing link there of like, AI will help you generate these interactive environments that you then use to hone your intuition. But that essay is really just rooted in the ideas of Brett Victor, right? Brett Victor talks about this a lot in... [48:12] inventing on principle and his other essays and i i view ai as like [48:17] the way that his ideas will end up getting closed. Yeah, maybe even more underrated than... [48:22] The understanding part is the motivation part to go back to it. Like, [48:27] That's what's amazing about video games is that like, yeah, video games are on ramp for coding. Kids play Minecraft or whatever they learn to code. But the key idea is that they spent that much time in Minecraft. [48:36] which is impressive. A few questions quickly on sort of like learning broadly or from a meta level. [48:42] You talk about abstraction. This obviously ties a little bit into the bodily experience, but you talk about abstraction and Faraday, and I think there's also –

[48:49] some vitamin examples too about sort of embodying some of these things. [48:53] I'm curious how you think about moving down from abstraction. [48:58] on kind of a regular basis or in a day-to-day sense. [49:01] You even talk about like, how can you generate concrete examples when you can't physically do an experiment? Obviously, that's specifically in the science context. [49:10] But as someone who... [49:11] If you wasn't clear already, like I spent a lot of time in abstract layer. [49:15] And there seems to be this like, [49:16] at least really talented, smart people, especially in science, but maybe broadly, are good at finding ways to like, [49:22] go down, like abstract down into more concrete stuff. Again, even when there isn't a simulation to run or a test to run or experiment to run. [49:31] Yeah. One thing I learned from the rationalist movement that I found very useful was this phrase, name three examples. And it basically just post-fetting things, right? Like when you hear anything abstract, like, oh, AI is going to create explosive growth. You're like, all right, name three examples. Right. Okay. Let me think of something I know well. So like, [49:49] Clinical drug discovery. How is that going to explode? [49:53] once you get ASI, right? And then UK, you think about what's involved in clinical drug discovery. It's like, well, you have to write a bunch of stuff for the FDA and they spend six months approving it or two years approving it or whatever. How is AI going to accelerate that? And then [50:05] Now you have a much more specific version of the question, which is like, okay, how do we deal with, like, regulatory? And you can't be as lazy. Yeah, you cannot be. Exactly, exactly. And so... [50:13] I think this applies to just like every area of life. Like another simple example is like, and they teach you this at, [50:18] Palantir and I think a lot of other good companies too is just like whenever you're interviewing someone

[50:22] just ask them for concrete examples and gather data, right? Gather signals. So just like... [50:28] What's an example of like... [50:30] a project you're proud of, right? And don't let them stay on the abstract level of like, oh, I... [50:35] I owned this project. Just be like, okay, what was your specific role there? What did you specifically do? Who else was on it? Tell me about them. And the more like [50:44] down to that level you get, the better your epistemics are. Right. [50:47] Maybe related, you're very into Karl Popper and fallibilism, it seems. Maybe not very into them, but good at talking about them and explaining them. [50:55] Are you able to run this sort of like... [50:57] fallibilistic frame in the background. The key idea there to me seems to be this like default to just updating your thinking and criticizing your ideas. [51:07] Maybe part of this is like being afraid to look stupid. [51:11] Maybe part of this is just like a default skepticism. [51:14] It makes sense in a scientific kind of like there's a rhythm to... [51:17] create and then evaluate, create and evaluate in a personal life. It's easier for me. It's much easier to like be drinking my own Kool-Aid for a while until I like realize it. [51:27] Do you have any methods or thoughts or frames to sort of keep that [51:31] or top of minds? [51:32] Yes, I do. I think, you know, one thing I'm really terrified of and always have been is this idea that as people get older, they just kind of, [51:39] settle into that grooves and they just get set in ways of thinking and their opinions get set and they don't change their minds and they really stop it. [51:46] and I'm absolutely terrified of like stagnating as I get older. Right. And so like one of my favorite examples of somebody who just completely avoids this is Tyler Cowen. And,

[51:55] He's so, so good at just constantly revising his thinking in a very... [52:00] basically Bayesian way. [52:02] And the way he does it is he just constantly writes about stuff via his blog. But there's a lot he writes that never makes it onto his blog, where he'll just sit down with a Google Doc and be like, I'm just going to write a page on what I think about topic X, right? Yeah. [52:13] He'll just like, yeah. So I think just having these different ways of [52:17] I think for me, the very base principle is just you don't actually know a lot and... [52:22] Everything you have is only a provisional and it's only a hypothesis and you should always be willing to say... [52:27] Oh, this thing I thought for a long time was actually... [52:30] totally wrong. I've had several moments of this in my life, and I view this as Papa's core message. It's like, [52:36] As soon as you think you have the final answer and this is the way it works, bad things happen. Both epistemically and, you know, he took that into politics and he said this is what happens when you have authoritarian governments. They think they know the answer and they end up ruining everything. Yeah, our mutual friend Graham Duncan has this, like, frame of the loose grip. [52:52] which I always come back to. It's like... [52:54] Man, I could have a looser grip on almost everything. And you want to have a grip. You want to... That's the... [52:59] The challenge, of course. [53:00] There's a theme you talked about a little bit. [53:03] It probably ties back to some of the previous things we talked about too, but... [53:06] just broadly like going slower. [53:09] Is that a meta theme? And you also don't seem to be someone who's... [53:14] always or usually going that slow. [53:16] but you do read a lot of long-form... [53:18] You seem to at least be taking time to think. [53:21] How do you relate to the slowing down? Yeah, I think anyone who knows me

[53:26] would laugh if they heard that I would like to go slow because I think I'm like notoriously hasty [53:31] I think, you know, it's interesting when you use LLMs, right, because... [53:35] Does... [53:36] There's the kind of ones that instantly generate responses, and then there's like O1 Pro or something where it thinks for a while, and they're kind of different models, right? And I think this is analogous to... [53:45] humans. But yeah, I think it depends on what domain you're operating in. So for example, in [53:50] entrepreneurship, I think [53:52] typically going slow is a mistake. Obviously, there are exceptions, but [53:55] Usually you want to [53:57] run very fast at any idea you have yeah because chances are go back to the prior question like [54:02] Quickly find out everything that's wrong. Exactly. Exactly. Yeah. [54:06] I think this is true personally, too, by the way. Like, if I... [54:10] You know, like, I'm guilty of this recently. Like, if you have this essay idea or book idea or whatever that you think is so great and you're just chewing on it for weeks and weeks and weeks, [54:19] it's kind of a mistake, right? It's like, [54:21] What you actually want to do is write the version as fast as possible and like prove to yourself that you have something great. And go and make it have contact with reality. Yeah. And spend all your points just like. [54:30] Don't assume that you have a finite supply of stuff in you, and if you spend it too much, then you won't have anything left. It'll always regenerate, so just sit down and do it. I have this recently. I have this thing I've been chewing on forever, and I just... [54:42] Need to sit down and ship it at some point, but it's like gained this stature in my mind of like, this is my ultimate thing of aesthetics and now I can't sit down and write the damn thing. [54:51] So I'm guilty of it. But I think, like, on the flip side, though, [54:55] A situation where it's very valuable to go slow is when you are trying to learn

[55:00] something complex and difficult, right? So... [55:03] a CS concept, a math concept. [55:05] Even stuff in econ, right? It is like... [55:09] very important to [55:11] uh not be like oh yeah okay that i get that it's just like this the will to understand is kind of definitionally a little slow yes yes yes exactly like just test it with loads of examples be like okay is this an example of this it's you know you hear the phrase efficient market hypothesis old all right like that's a good example it's just like [55:28] Okay, what does that actually mean? Just slow down for a second. [55:32] Does it mean like there's no ideas? No, that's not actually what it means at all. And then when you zoom in and find out what it actually means, it's like a very technical kind of a boring statement, which is. [55:41] To the effect of like, you cannot predict short-term stock prices. You cannot tell me... [55:45] what the value of Microsoft stock is going to be next week. That's actually what the original thing is saying. I think people extrapolate from that to all sorts of other things. [55:53] But like you wouldn't really know that unless you kind of went slow and read the paper, chewed on it for a while. Right. Yeah. [56:00] There's this idea from getting things done, your mind is for having ideas, not holding them. [56:05] You seem to you're clearly like a crazy infogore. You have like a lot of different models running in your head at all times. [56:14] Yet you also seem pretty good at synthesis and compression. [56:17] And so you've got this like crazy high volume top of funnel and you're, [56:21] in the world, you're not just like a [56:23] a person stuck in theory, you're building a company, [56:26] maybe building off that idea, having ideas, not holding them. Like, how do you...

[56:31] Is that a conscious thing you're trying to make sense of? Like, do you have systems that you use for... [56:37] cataloging and remembering important ideas, you've talked a little bit about the ways to kind of go through [56:42] maybe you were just getting at it, like go through very complex ideas or complex literature. [56:47] but, [56:47] Is that something you have a methodology around or just kind of ebb and flow? [56:52] Yeah, I do think I have this other thing of like lean into what makes you imbalanced, right? And I do think this is one area where... [56:58] I'm pretty imbalanced. I just, I take in a lot of information, like almost, there's a couple of people I know who [57:03] more than me right but it's like pretty high [57:06] um so i like to lean into that i i think like um there were a few things that helped me one was when i was at university the degree i was doing just required an insane amount of reading like an unreasonable amount of reading and [57:18] I actually learned to speed read just to, like, be able to do all the reading. [57:22] And I think it was like, it was some Tim Ferriss article that taught me. Is this the one with the lines? Yeah. Draw the lines on the page? Yeah. Yeah. I got to seriously try that. I kind of tried it, but not seriously. And I'm a very slow reader, I think. It was very helpful for me. My reading speed increased a lot without losing too much retention. Yeah. I mean, there's a few tips he gives, which are like very concrete. So it's like, when you read, don't sub-vocalize as much. Like, don't pronounce the words. Yeah. That's the thing I can't get past. I like the hardcore internal monologue. Right. And then the other thing is like, don't backtrack. [57:52] going down the page don't worry if you even don't take in stuff like actually your unconscious is taking it more than you're aware of so yeah i had all these tips and then i think what happened was

[58:02] that over time and [58:04] and so on but you know what man like i've tried every note-taking system under the sun i've tried zettel cast and i've tried obsidian i've tried rome research i've [58:12] I have physical notebooks, but [58:14] I just don't find note taking that compelling. I think that you just have to take in a lot of inputs and then what sticks with you is going to stick with you no matter what you do. [58:24] And that's okay. So now I'm just, that's where I'm at at the moment. And then every now and then what happens is it's almost like something inside me. [58:32] overflows a little bit and I have to write it down in some form. And this is actually why I tweet is... [58:37] I'm having this happen to me a fair amount and then [58:40] Just to get it out, I'm like, all right, I'm going to make this a tweet. And then I don't have to think about this anymore. And the way my essays come about is... [58:47] I... [58:48] Start a tweet. [58:49] And then more tweets come and then more calm. And then before I know it, it's like, this is too unwieldy. It's actually an essay. And then I have to switch to [58:55] Google Docs, and then before you know it, it's an essay. That is... [58:59] how almost all of them have come about. [59:01] And so for me, Twitter is this very generative, like useful, creative tool of just like all my readings coming in, stuff just comes out and I don't really care if it's good or not. And then. [59:11] Some of it just turns into essays and that's how it all works. You called tweeting like a candle wick, which is an amazing metaphor. Yeah. [59:18] You were getting at this idea a little bit of like sort of the things that need to be remembered, like. [59:23] It's almost like a lack of intentionality around memory. You've also praised spaced repetition specifically. Yes. Maybe back to the note-taking, not having a note-taking system.

[59:33] Is that something you're still using a lot? What would you suggest others use it for? [59:36] Broadly, memory is a deliberate choice. How do you think about that? Because that feels like the extreme, extreme end of hyper-specialized, systematic, deliberate. [59:46] Yeah, one of my favorite... [59:48] tweets i ever did was just i i asked the question which was just why aren't space repetition things more mainstream it's also flashcards just for yes yes yes flashcards right uh why aren't they more prevalent and i got a lot of interesting answers back [1:00:01] So I don't actually use it that regularly. The only time I use it is when I have to memorize a bunch of boring facts. And so that turns out to be a somewhat... [1:00:11] specialized thing. I have tried the more expansive version of it. Like, go read Michael Nielsen's essay on this. Yeah, I read part of that. It's very good. I can't do the stuff he does where he's like, I will space repetition a favorite painting just so I get to look at it every now and then. [1:00:27] The idea of doing this is very alien to me. Sometimes I will think of it as I'm walking about my day or whatever, but... [1:00:33] If I'm having to artificially bring it up, it's like something's gone wrong. So I think for me, there's all these domains where the facts you have to learn [1:00:41] are very boring. Like, you know, the recent one I was learning about how electronic health records work and in medical context. And there's just a lot of very specific arbitrary facts you need to learn. So it's amazing for that. [1:00:53] But for anything beyond that, I haven't really gotten it to work. So I do think there is this thing here of just like what is useful to you. [1:01:01] is [1:01:02] what is going to stick in your memory anyway i think talib made this point originally and

[1:01:07] I think that's very true. [1:01:08] Right. [1:01:09] Changing gears pretty dramatically. I want to spend a little time on... [1:01:13] category that I would call a mix of sort of government and bureaucracy and politics and power. [1:01:19] You spent almost a decade at Palantir. [1:01:22] There's an idea in your Palantir essay that I really found interesting, which is, [1:01:27] these three categories of [1:01:29] like areas of work, [1:01:30] The first you call like morally neutral, pretty easy to, [1:01:34] underwrite, like not political or whatever. [1:01:37] The second is obviously good. And the third is... [1:01:40] Gray areas, category three. [1:01:42] You say... [1:01:43] But it seems to me that ignoring Category 3 entirely and just disengaging with it [1:01:48] is also an abdication of responsibility. Institutions in Category 3 need to exist. [1:01:53] The USA is defended by people with guns. [1:01:55] The police have to enforce crime. In my experience, even people who are morally uncomfortable with some aspects of policing [1:02:01] are quick to call the police if their home has been robbed. [1:02:04] Oil companies have to provide energy. Health insurers have to make difficult decisions all the time. Yes, there are unsavory aspects to all these things. But do we just disengage from all of these institutions entirely? [1:02:14] or let them sort themselves out. [1:02:17] I read, I think I mentioned when we met, Alex Karp. [1:02:20] There was this amazing interview with him recently in the New York Times, and it was very similar. And it was one of the things that stood out to me is, [1:02:26] an uncommon moral perspective, which is just like, I'm willing to spend time in category three, maybe a defining aspect of Palantir in some way. [1:02:34] Can you talk about this from a moral perspective, maybe even as a form of duty? You're also an immigrant to the U.S. And so I'd be curious how you relate to maybe that piece of it, too.

[1:02:45] Yeah, I have a lot of thoughts there. So... [1:02:48] I don't know if I should tell the story. So my first day at Palantir, I was doing onboarding with this one other guy, right, who was like at least 10 years older than me. [1:02:58] And I was like, wow, this guy doesn't look like a normal tech company employee. [1:03:03] And so we go, we're in the room, we're chatting. It's like a hotel conference room. [1:03:07] So I was like, oh, so what did you do before this? And he kind of just stares me dead in the eye and he's like, [1:03:13] To the agency for 15 years. [1:03:16] I was like, that's like actually probably exactly what people expect of volunteer. No, I know. It's like so stereotypical. And I was just like this little pipsqueak. I was like, oh, I worked at a fintech company. [1:03:29] I felt ridiculous. Um, [1:03:31] And yeah, it was a really eye-opening thing for me in general because I came into it with certain... [1:03:38] views and certain kind of preconceptions about stuff like [1:03:42] the military, intelligence, defense, etc. I was [1:03:46] much more on the kind of [1:03:48] not that I'm, like, pro-war now or whatever, but I was very much on the kind of anti-war, like... [1:03:52] you know, don't like the military kind of vibe. And, um... [1:03:55] Then I had a bunch of colleagues who were ex-military or ex-marine or ex-whatever, and... [1:04:01] just kind of talk to them about stuff and just realize, wow, it's actually really, really, really [1:04:07] The world is way more nuanced than... [1:04:10] any of these kind of concepts can really capture, right? And if you just think about...

[1:04:15] international relations and foreign relations for a second, a lot of the most [1:04:20] Tyler likes to say that the default is war, the default is conflict, the default is death and destruction. [1:04:26] One of the most peaceful periods of the entire world was like post-World War II. [1:04:31] kind of like, you know, Pax Americana. And the reason Pax Americana existed is because America was this massive behemoth. All the European countries had kind of destroyed themselves. [1:04:40] America built up this massive military. And then anytime somebody stepped out of line, they would just go in and kind of [1:04:45] essentially smack them, right? And so you kept the peace and you had this crazy period of economic growth throughout the entire world. And so [1:04:51] If you actually think really hard about this, [1:04:54] then you realize that what is going into it is the threat of violence and the threat of military force. And that's very morally ambiguous, right? [1:05:01] You are. [1:05:01] You probably don't want to be the person... [1:05:03] applying the military force that involves doing unpleasant stuff at the same time. It's like that speech in A Few Good Men, right? It's like, [1:05:10] Uh, [1:05:11] your world is guarded with men with guns and so on. [1:05:15] I just had to kind of confront that really head on, and then... [1:05:19] While I worked there, I think it was 2017 that [1:05:23] people started protesting outside the office because of all the ICE stuff and the first Trump administration. [1:05:28] you just have to very directly confront this thing of like, okay, like where do I stand on this stuff? Right. And, um, [1:05:34] I... [1:05:36] Again, I think immigration is just one of these classic examples where... [1:05:41] You could have no immigration policy, but it ends up that most people don't like that result. And so as soon as you have any immigration policy...

[1:05:50] You're like, well, do I enforce it? And then as soon as you have to enforce it, you're suddenly confronted with these moral dilemmas and there's just nothing you can do. Like there's no, there is really no easy answer, right? [1:06:00] And so my point in that was just like, well... [1:06:03] you can either completely wash your hands of it and walk away from it as, you know, I think, [1:06:07] Notoriously, Google walked away from this Pentagon AI contract back in the day because they didn't want to engage with defense. [1:06:14] And I just think that's a mistake, man. Like, [1:06:17] You... [1:06:18] You not being there is not going to make things better if you are... [1:06:22] good agent and if you're oriented in the right way. And I think Palantir did well to exploit this inefficiency for a while. They were one of the few [1:06:30] tech companies who said, no, actually, we believe in a strong America and we think the West is good and [1:06:35] We're going to be in the game when other people drop out. That was really unpopular at the time. Now the vibes have changed completely, and it's like American dynamism. Everyone's like, rah, rah, right? [1:06:45] Great, but... [1:06:46] I do have a lot of respect for people who [1:06:49] were able to kind of make those unpopular calls. [1:06:52] back then and i think i think it's also like i don't think the critics are wrong either i just think it's this [1:06:56] dialectic, it's this debate, and [1:06:59] both sides have good points and we just have to work it out as a society but to me like disengagement is [1:07:04] absolutely the wrong stance to take. How did Palantir internally maintain a culture that was like [1:07:09] In theory, [1:07:11] if you're going to engage in category three, you want to be in this constant state of like flux on the line. [1:07:16] and [1:07:17] a lesser, and I don't, I don't, [1:07:19] technically no. My read from the outside is the perspective seems to be that Palantir and Alex do a good job

[1:07:26] of towing that line. [1:07:28] versus somebody lesser might just kind of like, [1:07:30] Oh, we're in gray area, so I guess we'll do everything. [1:07:33] which might be the argument Google could have made. I don't think that would have been necessarily a great argument. [1:07:38] Broadly, do you think... [1:07:39] What do you think Palantir culturally does well that allows for that type of [1:07:44] to go back to the other idea, having a loose grip on a bunch of complex ideas and being able to [1:07:50] make hard decisions over and over and over again. [1:07:53] Yeah, I think there's basically the essence of it is they tested very rigorously for people who they felt were independent thinkers, and they did that in all kinds of ways. [1:08:04] including, you know, founder interviews and whatnot. Like they wanted people who were kind of weird. [1:08:08] And then they wanted people who would... [1:08:10] it's basically this combo of two round traits. It's like people who speak their minds [1:08:14] but also people who were not douchebags. [1:08:17] And I asked Cop this in my interview with him. I was like, how do you keep the culture as you grow? And he was just like, as soon as somebody's a douchebag, we fire them instantly. [1:08:24] What do you mean by douchebag? Yeah, it's just like somebody who is... [1:08:28] behaves unprofessionally, maybe they shut people down in certain ways, or... [1:08:32] just any kind of bad conduct, any conduct that's like lacking in integrity, right? I mean, a lot of people... [1:08:37] I don't think this is right, but I think a lot of people would say that is true about people who speak their mind. [1:08:42] Right, that's what I'm saying. It's like such a rare intersection of these two traits. Right. It's like somebody who's very respectful and good, but also like... [1:08:50] is willing to be like, ah, this is bad. I don't like it, right? And so I think what happens when you get those people is then...

[1:08:56] One is like they're willing to buck the societal trend of like, oh, we think this is good or bad. And they're willing to say, no, actually, I'm OK with this gray area for these. [1:09:04] you know, [1:09:04] Principled reasons? [1:09:06] But then conversely, if you overstep the line, like let's say you go work with a cigarette company, they're going to be the people typing in Slack like, [1:09:12] this sucks, man. I thought we were making the world better. Like, [1:09:15] Cop is scamming all of us, right? And this actually did happen. I mean, we refused to work with a cigarette company who wanted to work with us. [1:09:22] it's not like there had to be some outcry I think Cop was just like no we're not doing this [1:09:27] And... [1:09:28] Yeah, you just need those kind of people. And I think it's really the question is just like, how do you... [1:09:33] select for them and that's what they got right [1:09:36] Mm. [1:09:38] We're at a time now where obviously government bureaucracy is very top of mind. [1:09:44] There are two brief excerpts from you that I think are interesting in this [1:09:47] to contrast here, [1:09:49] First you say: [1:09:51] When we were allowed to work within an organization, this is from the Palantir essay, when we were allowed to work within an organization, [1:09:57] This tended to work very well. The obstacles were mostly political. Every time you see the government give another $110 million contract to Deloitte for building a website that doesn't work, [1:10:06] or healthcare.gov style debacle, [1:10:08] or SFUSD spending $40 million to implement a payroll system that, again, doesn't work. [1:10:12] You are seeing politics beat substance. See SpaceX versus NASA for another example. [1:10:16] And then more recently you said, [1:10:18] I'm long USA, but expect significant volatility in 2025 to 2028. [1:10:24] We're not used to change happening this fast because it hasn't in our lifetimes.

[1:10:28] And so in a sense, you've painted [1:10:30] with these two quotes, like both a very optimistic case for why something like Doge could be really good. [1:10:35] And you basically said like brace for impact. [1:10:38] Can you expand a little bit on [1:10:40] Broadly why, and you know, if you... [1:10:43] explicitly comment on Doge from a [1:10:45] um, [1:10:46] very tactical standpoint, but broadly why something like Doge could be good. [1:10:49] what you think the risks are [1:10:51] and how you actually think about bracing for what could be a very volatile next few years. [1:10:57] Yeah, for sure. Again, lots of that. Yeah. So I've touched the government sector in some way or another for a long time in my career now, right? And I think it's just this painful contradiction of [1:11:07] The people in government are... [1:11:09] very often very well intentioned like the number of civil servants i've met who i admire hugely is very high and [1:11:17] People that are trying to do that best, it's an unglamorous job. You have to work with shitty... [1:11:22] you're earning less salary than you could in the private sector, and yet you are serving a country at the end of the day. [1:11:27] And... [1:11:28] It just made me mad, dude. It's like... [1:11:30] You can only watch so many times as... [1:11:33] a contract for a billion dollars goes to some systems integrator and they spend five years and you know for a fact that it's a simple ruby app it's like crazy to me that this happens and it [1:11:45] Just now I was tweeting about [1:11:47] um doge just went into the irs and they found some modernization effort and the modernization effort took [1:11:52] you know 50 years and it's cost billions and billions of dollars and it's still not finished straight and so [1:11:56] This stuff should be simple. We know how to do it. The cost of good technology has gone down over time.

[1:12:02] And so, and it's going down further with AI, right? And so... [1:12:06] you can cut all of these costs now and you can actually implement good systems for the government. [1:12:10] Try using a government computer for a day. It is a [1:12:13] Horribly painful experience. In the UK, you know, Dominic Cummings gave a great interview to Dworkesh about this, where he said he tells a story of how they took two years to decide between [1:12:23] I think it was Google Drive and Microsoft Office or something. And eventually he was just like, please, for the love of God, just choose one and do it. [1:12:28] Any normal private company would be out of business if they acted that way. And so... [1:12:34] I've always been obsessed with this thing of like, well, government's so important and it can be used to do so much. So why does it operate in this very inefficient way and what would it take for? [1:12:42] to actually change that. And so I was very excited about Doge because I think [1:12:47] it's an actual attempt to do this and it has more of a chance of succeeding than any previous attempt. Part of that is [1:12:53] you know, Elon being a very extraordinary person, right? And part of it is like the mandate he has. That's a very historically unique thing. [1:13:00] Cummings tried to do this in the UK. It didn't quite work out for sort of contingent reasons. Um, [1:13:05] But basically every state [1:13:06] in the world is going to have to do something like this because [1:13:10] This isn't a US-specific phenomenon. It's not a UK-specific phenomenon. [1:13:13] Everywhere this kind of technical debt builds up in time over government, and you have to figure out a way to clean that. [1:13:20] you become sclerotic and like some more dynamic government. [1:13:23] overtakes you and so yeah like that that's where the kind of efficiency stuff comes from but then it's like [1:13:29] In order to do that, it's going to be incredibly volatile, right? You have to...

[1:13:33] I don't know, basically the [1:13:35] I spoke to a lot of people in DC about it when it was announced. So this is kind of, you know, October to December timeframe and the kind of [1:13:42] conventional DC take on Doge was [1:13:44] well, we've seen this movie before. We've seen these government efficiency initiatives. You know, what happens is people say, yeah, we want to be more efficient, and then we [1:13:52] stand up a committee and then we you know take six months to schedule the meeting and then before you know it four years have passed and nothing has got done and in fact like that was Curtis Yavin's prediction of what would happen. [1:14:02] And so then it's like, okay, well, how do you operate an environment where everybody is constantly putting these blockers in your way? [1:14:08] you basically do what they are doing right now, which is just like, yeah, yeah. Blitzkrieg, right. It's like you go in and you, [1:14:14] grab control of the hr system and you grab control of the payment system and you force your way into buildings and you say no we are terminating this lease and like yeah we'll fight you in the courts and [1:14:23] um look i don't agree with every single decision that they've made i don't think they agree with every decision they've made but uh [1:14:29] This is something that has to be done, otherwise... [1:14:32] over time we will just become a less and less dynamic society [1:14:36] And so I think everybody, I think more people, including smart people in my sphere who I'm friends with, [1:14:41] should be more in favor of the goal and like a little bit more understanding about it. [1:14:46] you know, the volatility and the difficulty of actually doing this. It's a real quagmire. [1:14:51] I think it ties into this. [1:14:53] We talked about fallibilism a little bit. You've talked about Popper and politics. [1:14:57] and you brought even brought up Dominic Cummings, you had said the most important criterion for society overall is that it makes progress and creates knowledge at the fastest rate possible on how to make its citizens affluent.

[1:15:09] In order to do this, citizens must be able to make mistakes and correct them. [1:15:13] Thus, the essence of democracy isn't who should rule, [1:15:16] but the ability to remove a bad ruler, which is the same as being able to correct a mistake. [1:15:22] Put simply, again, you... [1:15:24] The most important thing in knowledge creation, including political knowledge, is error correction. [1:15:28] This is Popper's Criterion. [1:15:31] Why beyond bureaucracy and a desire for maintaining power, is it so hard to get this? Is it so hard? Why can't we design systems? [1:15:42] that are better at [1:15:44] understanding people's incentives and like allowing for error correction, at least at a government level. [1:15:48] Yeah, I mean, I think it's a historically rare and unique thing, right, that... [1:15:53] you know, somebody loses an election and then they sort of voluntarily relinquish power, right? If you read history, you're just like, this is crazy, right? [1:16:01] There's the king, and then he or she, the king, he fights to the death, and [1:16:06] eventually gets murdered and a new king arises and that's how it's always been. [1:16:10] I don't know, it feels very deeply rooted in human nature to me, of this, like... [1:16:15] this uh [1:16:16] You're going against a lot of very strong and powerful incentives. [1:16:20] in humans to kind of maximize and like acquire resources. [1:16:24] And so, yeah, democracy is this very... [1:16:26] kind of unnatural... [1:16:28] Miracle. I think for the same reason, the fact that we do have an open... [1:16:33] political culture and an open ability to criticize [1:16:36] is so important about the US. I do think that is something that does differentiate us from a place like China, which is otherwise...

[1:16:44] You know, it's a little scary in some ways, right? They're doing very well in all these dimensions. They're making drugs much better and more cheaply than us. Their electric cars are pretty good. [1:16:53] deep seek AI, right? So there is all of that, but I do think in the long run, the more open political culture [1:16:59] Still should win. We'll see what happens. [1:17:02] Speaking of maybe technology and hanging on to power or acquiring power, [1:17:08] You discussed, I think on another podcast with Dan, you discussed the Henriette in Shakespeare and how Hal is sort of like... [1:17:14] showing up in one context around power and in a totally different context in his private life and how those are really disparate. [1:17:20] And specifically how you have to be really manipulative to maintain power. [1:17:24] I'm curious for your perspective on what seems to me at least to be a growing trend of technologists. [1:17:31] Moving in on power. [1:17:32] seemingly prioritizing power above almost everything else, if not everything else, that to me seems to be certainly distinct from, [1:17:41] notable in the last year and probably... [1:17:43] quite distinct from what technologists spent their time on 10 years ago. [1:17:48] Yes, I agree. I mean... [1:17:51] It's a very mixed thing, isn't it? I think for me, the... [1:17:53] the [1:17:54] The mystery I felt [1:17:56] for a long time was [1:17:58] Technology is so... [1:17:59] powerful and yet we are not [1:18:02] in the world at all in some sense, right? We're not... [1:18:05] building libraries, we're not building museums, we're not building physical things, we aren't interfering in politics, we really just want to be in a garage in Palo Alto and kind of tinker and occasionally we make a social media app and it ends up causing a revolution in Myanmar or something.

[1:18:19] And I think the classic 2014, at least pre-Trump, like it was, oh, I don't really. Yeah. Hands up. Yeah. Right. And so there was this huge asymmetry between how important tech actually was and then [1:18:30] how much it was playing in fields like this. And I think what happened was like, it's the classic, you know, [1:18:35] If you're not interested in politics, eventually politics becomes interested in you, right? Where, you know, Zuckerberg gets hold before Congress and he's sitting there kind of pale in a suit and he's like, oh, Senator, like we sell ads. [1:18:45] I think eventually savvy people in tech realized, okay, we actually have to play this game. Otherwise, the game is going to play us and it's going to be... [1:18:53] pretty bad for us, right? And so, for example, like, [1:18:57] I think with Elon, what radicalized him was, well, I'm making all these cars in the USA. [1:19:02] revolutionizing spaceflight, like quite literally with SpaceX, and yet [1:19:07] you know, the previous administration doesn't even want to say my name, except in a pejorative context, right? I don't get [1:19:12] invited to the White House [1:19:14] Whenever Biden talks about electric vehicles, he talks about general motors, which, like, nobody cares about anymore. So it's all very confusing. And I think this, he's one of the first to spot, like, okay, we actually... [1:19:25] have to [1:19:26] in some sense, take power or influence power if we're going to actually achieve our goals. And for him, this is at least like the way he says it is like, we have to go to Mars. We have to [1:19:35] you know, be able to produce things in the US and things like that. Um, [1:19:38] And I'm [1:19:39] On net, like, I'm glad, like, I do think there are distasteful aspects of this, right? You watch people suddenly flip their political positions and you're like, hey, what happened to you a year ago? [1:19:48] You know, you watch them...

[1:19:51] saying things that are kind of silly on X, right? And kind of doing it in a very partisan way. And it's easy to just be like, oh, this is so gross. But [1:19:58] I think it's actually existentially important. It's like, okay, what is... [1:20:03] Okay, what is really important in the next 20 years? [1:20:07] AI. Okay, what does AI need? Chips. Okay, where are all the chips made? [1:20:13] the Republic of China, i.e. Taiwan, right? That's the name of Taiwan is the Republic of China. So yeah, all the chips are made at TSMC. [1:20:20] which is like all the way on the other side of the world. This seems like a problem. And so I think if tech does not play politics, then ultimately we end up losing these things in a way that's irreversible. Right. [1:20:30] and [1:20:31] We don't want that world. [1:20:34] One instinct or read I might have would be that [1:20:37] We talked about in the Palantir series, [1:20:39] culture, like how do you have this balance? [1:20:42] It doesn't seem to me that [1:20:43] a number of these technologists, classically Elon, but like Marc Andreessen is another easy example. [1:20:50] These people maybe seem... [1:20:53] Certainly capable, less attuned to like... [1:20:57] holding that balance and that tension. [1:20:59] It seems that... [1:21:01] It would be very, very easy to paint a picture of why... [1:21:05] Absolute power corrupts absolutely. [1:21:08] In three years, is Elon going to be more focused on Mars or less focused on Mars? [1:21:13] I think like... [1:21:14] to be somewhat critical, it seems to me that he's like, [1:21:17] His game has very much clearly... [1:21:20] But yet I'm also sympathetic to your point, which is that I don't know if he had another choice. I think he probably believed he was going to jail if the Democrats won. And maybe he was...

[1:21:28] Seems crazy to say, but maybe he was right. [1:21:30] But it seems, yeah, it just seems that unless... [1:21:34] Maybe Peter Thiel, he's obviously a controversial person. Someone like Alex Karp seems to be... [1:21:40] More used to, like, balancing... [1:21:42] technology, mission, vision, and also like [1:21:45] playing the political game. And now we have all these new entrance entrance, the political power game, [1:21:50] And they're all very, very eager to play it, but... [1:21:53] It's the classic Peter Thiel thing, man. He was at least... [1:21:57] at least five, probably like ten years early to this game. And then he basically exited it when it became mainstream. Like, 2016, I remember how shocking it was when he went to the Republican National Convention and he was like... He seems less obsessed with power than... [1:22:13] your average VC these days, which maybe that's just... [1:22:17] the public perception thing. [1:22:19] But Peter actually seems like a little bit more restrained in the power grab. You could argue he's pulling all the strings, but yeah. I don't know. I mean, yeah, I don't know. But it was interesting. He was very early to realizing this, and then he made a very bold play. [1:22:32] And, you know, he basically won, right? I mean, J.D. Vance... [1:22:36] I think he used to work for him or something. And so... [1:22:39] Yeah, I think ultimately... [1:22:42] There's aspects of it that don't look so great, but we basically have to play this game. I do think someone like Elon always has his eye on the ball. He is very serious about getting to Mars. He is very serious about... [1:22:54] regulation basically choking us from building anything he's really serious about onshoring a bunch of production that

[1:23:00] should not be living in other countries [1:23:03] And I think those themes are what is... [1:23:07] going on with the tech crowd. Now, what remains to be seen is like, well, will the administration actually execute effectively on those priorities, right? Or will they get distracted and create a bunch of noise? [1:23:18] I don't know. Politics is a very, very messy game. It's a very difficult game to play, but [1:23:23] I do think that's where all those tech people are coming from. Like what, [1:23:26] What, you know, A16Z, Mark and Race and whatever, what they're doing right now is a direct consequence of what they said a few years ago when they said it's time to build. [1:23:35] software's eating the world, like all of that. It's a literal expression of those beliefs. And I think they would have liked to make it work with... [1:23:41] the previous administration too, but the previous administration kind of just wasn't that interested. [1:23:47] You... [1:23:48] are building a new company, [1:23:51] You haven't talked about it. [1:23:52] publicly yet. [1:23:54] But I'd love to talk a little bit about [1:23:56] some of the lessons around the idea maze and the exploration founding team and people [1:24:03] especially going from working one place for eight years, big place, [1:24:07] What did you come into this experience, at least when you decided to go from sabbatical mode, expiration mode, to... [1:24:13] actually doing something, what [1:24:16] Did you come into that with holding maybe Most Precious around... [1:24:20] what to work on, [1:24:22] the team the people the problem. [1:24:25] Yeah, yeah, I learned a lot here. I think, you know, one thing Tyler likes to say a lot is context is that which is scarce, which is just such a rich saying. It's so wonderful.

[1:24:36] I think one of the things that I realized was just how true that was, right? So... [1:24:41] I think... [1:24:42] When I left, I'd spend a lot of time working with [1:24:45] government and big complex customers like Palantir does and [1:24:49] I think my initial urge was like, okay, I'm going to go do a pure software thing now. I'm going to make dev tools or something. I don't want to deal with people anymore. [1:24:58] And, um, so, you know, I, I explored a bunch of different markets, which, which we can go into if you want, but, um. [1:25:04] I kind of realized... [1:25:06] No, like... [1:25:07] Especially me and then, you know, the co-founder I brought in and the team that I founded this with. All of us have a very... [1:25:14] specific set of contacts that was built over many, many years and [1:25:18] Ultimately, the opportunities that arose came directly out of that context that we built up and like not out of. [1:25:25] approaching a new sector from first principles and being like all right how should this work like i'll give you one example is like i i looked pretty hard at education [1:25:34] And... [1:25:35] Again, I think any sector you can build a good business in. My personal conclusion looking at education was it would be very hard to build the kind of business that I wanted to do. [1:25:44] build like in an idealistic way like i very much cared about like okay how to improve learning how do i [1:25:48] you know, like I want to build the [1:25:50] the young lady's illustrated primary. I want to build Ender's game, right? Like, how do I do that with AI? And, you know, when you actually go into the business of education, then, [1:25:59] you end up with something that's like, oh, like, what if I made a better SAT, you know, SAT textbook or whatever? It's just a little bit depressing when you... [1:26:06] when you try and do the calculation of like okay what's going to make a lot of money and so like um like in terms of like keeping the business running

[1:26:12] So, yeah, I iterated through a few versions of this, but then it was just kind of hilarious that the thing I came to was kind of like almost exactly what you would predict if you knew a lot about what I'd done at Palantir and what my co-founders had done. And it's like a very... [1:26:25] natural extension of that and some of the stuff that we've been discussing as well um and so [1:26:30] I don't know. I think my... [1:26:33] My lesson, my big lesson that was just like... [1:26:35] Basically, if you want to be an entrepreneur, it's like build very deep context in something and then like. [1:26:40] you will naturally arrive at the edge and then... [1:26:43] It's not that hard to arrive at the edge, actually. The tough part is getting access to the problem, but once you're at the edge... [1:26:49] a bunch of really obvious ideas are going to stand out to you and that's where [1:26:52] really good [1:26:54] company ideas come from. It's very hard and very rare for [1:26:58] a good company idea to just come from somebody standing back and looking at a sector with a spreadsheet and being like oh i think this might make sense like it's possible it's just like i didn't find that work and rarely oftentimes people aren't actually at the edge which is why the problems aren't so obvious [1:27:11] I mentioned it super briefly earlier. You talk about scientist brain and founder brain. [1:27:16] maybe I would even add in a third of like artist brain in an era where, [1:27:22] The default things are really easy to do or faster to do or only going to get easier to do. [1:27:28] and maybe even ties into our conversation around care, like, [1:27:30] maybe caring about the details is going to matter more incrementally. And so I'd be curious how you think about, you talk really eloquently, but just about how, [1:27:38] Founders need to be incredibly good at being really high contacts on so many different things. And as a result, they definitionally can't go on these platforms.

[1:27:45] rabbit holes or like go way too into the weeds. Do you think, [1:27:49] Part of this is a personal question maybe for you as a leader, and maybe this is part of this is a broader question around like what the entrepreneurial founder leader of the future will look like. Do you think they're. [1:27:58] There's a world where they actually are going to be more artist-like or scientist-like in a way that they are deeply, deeply... [1:28:05] attuned to [1:28:07] maybe more specific details of the problem. [1:28:10] Yeah, it's really interesting. You know, one thing I was thinking about [1:28:14] with regards to doge and elon actually was like he one thing that's very unique about him is he has the ability to basically summon up 100 software engineers on demand and almost nobody else has this ability he can just pull people out of tesla spacex and right you know he can just be like hey stanford grads come follow me and like a bunch of 20 year olds will join him and and actually yeah it's like who can magic up that army very fast right and so i think i think like [1:28:38] Yeah, future successful entrepreneurs will often look like [1:28:41] They have a... [1:28:43] kind of deep and wide network, both let's say in customer relationships, they have maybe some unique access [1:28:48] They have some kind of... [1:28:50] some kind of brand or something where it's like they can like pull people to them and draw talent to them and that's [1:28:55] I think. [1:28:56] Maybe only gonna matter more, actually. [1:28:59] Whereas, like, I think for people in their early 20s or kind of younger, you know, maybe the stereotypical YC founder, I do think there's a bit more like... [1:29:07] Mm-hmm. [1:29:07] There's a lot of them throwing spaghetti against the wall and then... [1:29:11] the failure rates will be quite high, I think. And so, yeah, I think the returns to unique...

[1:29:16] or proprietary access or knowledge just get higher and higher because any [1:29:22] more like sort of generic idea basically becomes more and more doable with right ai over time so for example like if you know all these esoteric [1:29:30] processes that are mostly captured offline or if you know these [1:29:34] esoteric industries like shipping for example like you're probably going to do pretty well because ai can reinvent like the way that [1:29:41] anything is done in that industry now, right? [1:29:44] And so, yeah, I think a lot of interesting... Like, the latest trend that I've kind of heard people talk about a lot is... [1:29:51] buying services companies that have this deep moat or kind of distribution figured out already and then basically AI-fying them and that I think will actually become much more of a [1:30:01] much more of a thing. You and Jeremy Giffen were just going back and forth about [1:30:05] like secrets being incrementally, uh, [1:30:07] Better today than they used to be. I think that's inside that too. Yeah. If you, if you have a secret, just shut up about it. Don't tell anybody. Well, we are not announcing the company today. Speaking of secrets, [1:30:18] Changing back gears, some of this will probably tie into part of the beginning of the conversation, but I want to talk a bit about art. [1:30:25] First, [1:30:26] You criticize Tolstoy specifically later in life. [1:30:29] saying essentially he had this sort of like incorrect theory that art is about communicating moral purpose and teaching what is right and good. [1:30:37] You have maybe one answer I found, but before I talk about that, I'd be curious for you to think about what you think art is for. [1:30:45] Yeah, this is one of my favorite topics. Also, it's hilarious to come to this right after talking about entrepreneurship. There was no super obvious transitions or just like, yeah, we're confusing to listen to. No. OK, yeah, this is this is really fascinating. Right. So basically, Tolstoy writes some of the greatest books of all time. He writes War and Peace. He writes Anna Karenina. And then later in his life, he writes this kind of notorious essay called it was something like Lear Tolstoy and the Fool, but whatever. I'll look it up. But it's about Shakespeare. It's this huge rant about how Shakespeare is actually bad.

[1:31:15] and he's just like you know shakespeare everyone looks up at shakespeare he's the best author ever and i think he's just bad and [1:31:20] His basic critique of Shakespeare is like, [1:31:22] Shakespeare has no moral vision. He just depicts a bunch of people, you know, getting killed for random reasons, right? Or just killing each other or having problems. But [1:31:30] There's no message that his plays are promoting. It's just kind of chaos, right? And so... [1:31:36] He articulates this vision of art as... [1:31:40] Actually, the job of art is to show you goodness and moral goodness and, um, [1:31:44] Therefore, the best kind of art is almost like parables, like didactic parables. So then he spends the rest of his life after that writing these parables that honestly nobody ever reads. [1:31:52] And they're like really just like homely Christian parables about how somebody like [1:31:58] gets robbed and then he gives the thief his coat and whatever, like all this kind of stuff. And nobody reads that. They read Anna Karenina and [1:32:05] What I find really interesting about this is like you can actually see this tension and I don't know if you've read Anna Karenina. I haven't. It's amazing. It's like my favorite novel. Our friend Ava has been trying to get me to read it. Yeah, this year will have to make it your summer project. Seriously. But you can see this division in Anna Karenina. It's a deeply religious book. Like the epigraph to it is from the Bible. It goes, vengeance is mine. I will repay. I think it's from the Hebrew Bible. And. [1:32:27] You know, basically, like, one of the main characters, Levin, has this moral journey which kind of culminates in his converting to Christianity, like, in his heart. [1:32:36] And then there's Anna, who's the subject of the novel, and she's [1:32:39] she's almost this like devil figure, right? She's seductive. She's charming. She's beautiful. And, and, [1:32:44] Basically, she commits...

[1:32:46] adultery with somebody and that's kind of the other plot line in the novel. [1:32:50] right so you have this parallel plot so you have anna and then you have levin levin finds his way to moral goodness and anna [1:32:55] I won't spoil the book, but what's interesting is Tolstoy is clearly in love with Anna too. He is so charmed by Anna. He describes her charm so compellingly, how seductive she is. He clearly enjoys writing the scenes about Anna more than... [1:33:10] you know, he enjoys writing some of the other scenes, let's say, where he's promoting his moral ideas. And so I feel like this conflict was deep in him. [1:33:17] And there's a very similar phenomenon of literature, which is in the [1:33:21] Poet in Paradise Lost by John Milton, right, which is about the fall of man and Christianity. [1:33:26] It's a very common thing to say by critics that [1:33:29] The most compelling scenes Milton writes in that are all about the devil. [1:33:33] It's actually very boring when he's just depicting Adam and Eve frolicking in the Garden of Eden, but when he's writing about the devil, it's kind of sexy and cool and interesting. [1:33:40] And so, yeah, I think Tolstoy ends up proving himself wrong because the art that he writes that [1:33:46] he thinks exemplifies morality... [1:33:48] is actually very boring and nobody reads it. But people read stuff like Anna Karenina, which is much more kind of in the... [1:33:53] in the world that we all live in, right? And so, [1:33:57] You know, I think there's a bunch of things to take away from that. But yeah, that's kind of my take on Tolstoy. [1:34:02] You've... [1:34:02] Another maybe answer to this question, which is that art is good at defamiliarization. [1:34:09] and particularly how it sort of like gets us past the defaults. [1:34:12] or even like allows us to experience ordinary life in a new way or at a new rhythm or a new pace.

[1:34:19] Do you find that to consistently be true or at least be true in any specific mediums? [1:34:27] Yeah, right. I think it's back to what we were talking about at the beginning around original seeing, right? So actually Tolstoy uses technique all the time is he will take a familiar... [1:34:36] phenomenon and then he'll describe it as though an alien is seeing it in in like very unfamiliar terms and then you'll be like oh he's describing somebody like whipping a horse to make it go faster but he'll describe it in the most unnatural way right right and and in a way that makes you horrified like you almost see it afresh where you're just like oh shit this is actually bad whereas when you see it maybe you're just like whatever it's a routine thing so yeah i think it's like [1:34:58] You know what makes me think of is... [1:35:01] Remember the lobster by David Foster Wallace. Consider the lobster, yeah. Consider the lobster, yeah. Yes, yes, I fully agree. And Wallace was also amazing at this, right? [1:35:09] Yeah, so I think it does apply to every medium. And I think to me, it's just a flip side of the idea that strangeness is a character of art. I think everything is... [1:35:19] Everything really good. [1:35:21] is kind of jarring in some way. Like, as a simple example, like, Frank Ocean album, Blonde, [1:35:27] The first time I listened to that, [1:35:29] I was like, I don't know if I like this. Right. This is not Channel Orange. It's not at all Channel Orange. And then you listen to it a second time, you're like, this might be good. Then the third time, you're like, this might be the greatest album I've ever listened to, right? But initially, you're like, why is he using autotune in this bizarre way? Like, why is the voice so high randomly? Yes. Like, why does... I had that experience with Yeezus. Same thing. Yeezus. I think probably Kanye's best album. But the first time I was like... First time I heard it, I hated it. This hurts my ears. Yes. Yeah. Like, on site, right? Yeah. Yeah.

[1:35:59] you [1:35:59] Yeah. [1:36:00] Yeah. [1:36:00] You said movies are like waking dreams. [1:36:03] I know you love film. Why is film special? [1:36:07] I think with film, well... [1:36:10] There's a bunch of different reasons, right? But basically, you're watching it in this physical setting, and... [1:36:15] It's fully dark. [1:36:16] And it's almost like you're going to sleep and then the dream appears on the screen in front of you, right? And so that's why I think going to the theatre and watching it in the theatre is so important. [1:36:24] It's also interesting because it uses so many different media in one right. So you have music and film, you have visuals and film, you have. [1:36:32] film scripts, right? So there's like writing, there's visual, there's [1:36:35] There's all these different things and different films I think will emphasize... [1:36:38] different aspects of that. So I think that's interesting too. [1:36:41] I think it's just like, it's one of those things that... [1:36:45] It's as visceral as music in some ways because it's taking place in time and it's taking place on this... [1:36:52] big screen in front of you, right? [1:36:54] But... [1:36:55] Because... [1:36:57] It has so many kind of degrees of freedom, [1:37:00] it can be more morally complex, maybe. Like, I think a really good... [1:37:04] film can really... [1:37:05] put you through an experience that can simulate an experience that you'll have and [1:37:11] you can come out of it like very different. Yeah. [1:37:15] I saw you read Amusing Ourselves to Death, I think last year. I recently actually just read it. Yeah. [1:37:21] You read a ton of long form. [1:37:24] and seemed to really appreciate it. [1:37:26] I think my question is like, do you have a...

[1:37:30] either a plea or a... [1:37:33] Form of encouragement for people listening who... [1:37:37] maybe have experienced it or want to experience it more, [1:37:41] but like aren't getting there. Maybe back to the very beginning of the conversation of like our trend towards slop and our even trend to no longer appreciating... [1:37:50] things that are deeply cared beyond maybe we talked about reading speed, but [1:37:54] Anything else you might say on... [1:37:56] on a society who seems to be moving further and further away from this sort of like textual society [1:38:01] Yeah, I think people will... [1:38:04] work their way to this. I'm optimistic about [1:38:06] Right? I mean, like, everyone has something that they do really care about in this way. So people will be into, you know, anime or manga, or they'll be... [1:38:15] They'll be making these long playlists on Spotify, right? And when they listen to those, they kind of retreats. [1:38:21] you know, like a place inside of themselves, right? And so I think it's [1:38:26] It's basically just go further out of your comfort zone and... [1:38:31] consume things in lots of different media and things that might be hard or unnatural for you. [1:38:36] Like go to... [1:38:39] go to weird out-of-the-way art galleries or, you know, read that random old poem that, like, you never quite got and just kind of stick with it for a while. I think the thing that was really helpful for me was in... [1:38:50] When I realized, okay, art isn't just... [1:38:53] something that is [1:38:55] Saying stuff but in a way more complicated way. It's actually okay if you don't understand it at all. Like it's okay to read a poem and not completely. That really unlocked it for me. Because then I was like not feeling bad when I read.

[1:39:08] some vaguely obscure poem and I was just like, what am I supposed to take away from? Can you explain, so when you say that, [1:39:13] But did you still are you just saying like, oh, that wasn't for me? Are you saying I enjoyed it despite not fully getting it? [1:39:18] It's both, right? And it's like some things you read will still activate some kind of inner response. [1:39:26] And some things to read will just leave you completely flat. Right. This is the Robert Irwin thing a little bit, though, too, which is just like you don't necessarily know how it's going to work on you. There's this book I read a year and a half ago or so called The Overstory about trees and, like, [1:39:38] I liked it reading it, but I think about that book all the time still a year later. It's like it's working. [1:39:45] Yeah, yeah, like there's a full-line poem by... [1:39:49] Ezra Pound, which I love, and I think about it almost every day, right? And [1:39:53] But I think it's quite mysterious, too. I'll recite it if you don't mind. It goes... [1:39:58] And the days are not long enough, and the nights are not long enough. [1:40:03] and life slips by like a field mouse. [1:40:06] Not shaking the grass. [1:40:10] Yeah, and... [1:40:12] I think about the poem all the time. Like, why the mouse? But... [1:40:15] It works so well. I don't know. [1:40:17] It creates a mood. [1:40:18] I think I really appreciate the... [1:40:20] Holding off the temptation that feels dominant in today's society, which is like everything has to be processed and understood and completely compressed. [1:40:28] Right now or else, yeah. [1:40:31] Thank you for that. [1:40:32] Last question on this. Any patterns... [1:40:35] across different mediums, music books, ideas, patterns in your taste that you've recognized.

[1:40:43] Yeah, might. [1:40:44] It's very eclectic, I guess, right? Like... [1:40:47] I told you someone about music today and they were just like... [1:40:50] oh you seem to be into softer i was like yeah but i'm also into like heavy heavy metal like i love slayer i love death metal you know so it's [1:40:55] I think it's just, it's the urge to explore completely new domains, and [1:41:00] Again, I've mentioned him several times, but I think Tyler is my model for this because he's just more extreme on this dimension than anybody I know. He'll be into Mexican folk art or Haitian folk art or... [1:41:11] in classical music or he'll be able to talk about Bach right and I just like I model that trait on him [1:41:17] And I force myself every now and then to like, I just get bored, man. I'm just like, all right, [1:41:23] What's new? What's good? And then I'll be like, okay, well, I've not explored, you know, [1:41:27] 20s American blues music from Louisiana. There's so many things to explore. Yeah. There really are. Yeah. And I think it's just like the only common theme really is this like love of exploring unfamiliar domains and then trying to like crack them and understand. [1:41:44] how they operate and how they work and I'm kind of addicted to that feeling of being [1:41:49] that initial thing of like, wow, this sounds so odd, like, this really sounds weird, and then [1:41:53] kind of getting it [1:41:54] And then being able to tell, okay, this feels better, this feels worse. I think just like... [1:41:58] Being able to run that cycle again and again is very rewarding. [1:42:03] My first guest on the podcast, Jason Liu, [1:42:05] has this line confidence is the memory of success and I find that it just keeps applying to new ways and I think even in something like this like

[1:42:13] knowing that you can go into a totally new domain and find something really invigorating or interesting about it. Like the more that compounds, the more you do that, the easier it is to like pick the next like random arbitrary thing that feels – [1:42:29] not only professionally or learning or skill-wise, but maybe even interest-wise. And it's like, I think I can combat in both directions, which is cool. Yeah, yeah. One thing, I'm friends with Nick Camerata, and he's always talking about how people don't, [1:42:42] randomised enough. [1:42:43] They don't. [1:42:44] explore like basically everyone always underexplores right it's like [1:42:47] you eat a certain number of foods, but did you try randomizing this aspect of it? Yeah. And so I think that applies to basically every area of life, because [1:42:56] It's hard to experiment and you need a routine and there's this sunk cost and habit thing. So it's just like... [1:43:02] you know, like in travel, just like [1:43:04] Go somewhere you wouldn't go, like travel to South Dakota or something, right? And like go drive around there for a while. [1:43:10] Uh, yeah. [1:43:11] As we wrap up, I have kind of like a... [1:43:14] It's not a lightning round. You don't have to go lightning speed, but miscellaneous, let's call it. [1:43:18] Speaking of, what does it mean to travel well? [1:43:21] I think... [1:43:23] It sounds obvious, but I think truly talking to and hanging out with locals and getting locals to show you around. Yeah, because I think it's just like. [1:43:30] There's the city as it appears to be, and then there's the city as it really is. And you cannot access that ladder thing without somebody taking it, I think. [1:43:39] Like New York, we're both in New York. [1:43:41] The first time you visit, you go to Times Square and you go to Midtown, right? And then you talk to a New Yorker and they're like, no, go out to Brighton Beach in a Russian suit or something. And every place is like that. It turns to your earlier point about trade. Yes, yes, exactly. And so for me, every kind of quote-unquote travel hack that I think is important has involved just ways of doing this. And yeah, the travel experiences that I remember most, I would say a lot of them are...

[1:44:07] I had a friend who was from there maybe. I went to Czech Republic once with a Czech friend of mine, and he was just like, [1:44:14] he took me out drinking with his buddies and that was so memorable. And I knew that if I'd just gone to, [1:44:19] Czech Republic by myself, I'd have just been like, oh, Czech people are so unfriendly, and I didn't really like the food, and like... [1:44:24] you know, all this, but like, through him, it was a completely different experience. Yeah, so it's like, [1:44:30] find these entry points into places and then it will be so much more rewarding. [1:44:37] Are there any important stories you tell yourself about yourself? [1:44:41] Obviously, we all have lots of stories we tell ourselves, I suppose, but any that come to mind. [1:44:46] I think... [1:44:48] I truly think it's the thing around... [1:44:50] not [1:44:52] being too fixed about any opinions. [1:44:56] and [1:44:56] beliefs and keeping your [1:44:59] Yeah, keeping your identity small, right? So it's just like... [1:45:02] You don't know everything. [1:45:05] It's also like, I think, I don't know if it's healthy or not, but I always just tell myself, like, you've really not achieved anything yet. Like, you just don't. [1:45:12] I don't know. Maybe it's, like, a little messed up, but I'm always just like... [1:45:16] Every day is the first day of your life and you need to... [1:45:20] look forward to your next year and you need to be excited about it. [1:45:23] the projects that you're going to do. And I think that is just... [1:45:27] very fundamental to living life in a good way. [1:45:31] There's this sense I have that I'm sure I'm not, lots of people have talked about, [1:45:36] which is that we may be increasingly told what to do by AI,

[1:45:41] in life and in work. How comfortable are you with this idea both for [1:45:47] people generally and for yourself. [1:45:50] The version I expect to happen, I think, is that... [1:45:55] Well, it's always voluntary, right? It's never forcing you. And so... [1:46:00] My experience has been that basically there's a bunch of grudge work and then the AI just helps me do it quicker. [1:46:05] So far, right? It's just been like, oh, I don't really want to write this script, but like, okay, Sonnet. You're the top level decision on everything. Exactly, exactly. And so I think the things that I actually care about and want to kind of sit with for a while, I won't use AI for it. I will just do it myself. [1:46:21] Whereas, yeah, if I have to fill out some form, then I will go and chat GBT will do it for me. You don't expect to... [1:46:30] 18 months from now, wake up and ask ChatGPT how you spend your time. [1:46:35] I already do some I do some version of this you know like I'll give it I'll give it my to-do list sometimes I'll be like look tell me what to do okay I'm just gonna execute it it's kind of good though I mean it's just like it's like having a really smart friend and [1:46:47] It's in the nature of advice. It can't make you do it. I found myself really impressed in the last week or two by GBT4.0. [1:46:57] there's some change they made where it's just way better at the end of whatever you asked for, being like, by the way, do you want me to do X or Y or... [1:47:03] And I found myself just being like, yes, yes, yes. It's kind of cool, right? It's like when you're in Cursor and you're pressing tab, tab, tab, and you're like, oh, the AI. It's like surfing. It feels very nice. Yes. But it also, in this sense, it makes me feel like it's giving you more agency, which obviously could be an illusion or...

[1:47:20] Yes, yes. I think if you talk to people in the AI world, [1:47:23] Some of them are kind of doomery about this, where they'll say, like, well... [1:47:27] AI is going to get hyper persuasive. [1:47:29] And it's going to basically addict you and you're going to be obsessed with talking to it. Yeah. And [1:47:35] I didn't believe that. I don't know. I'm a bit more optimistic than that. Like my experience has just been that... [1:47:40] It [1:47:41] enhances my agency and as soon as I [1:47:44] As soon as I feel the opposite of that happening, I'll just switch it off. [1:47:48] That's dope. [1:47:49] You wrote about negotiation. [1:47:53] You say good founders have an instinct for reading rooms, group dynamics and power. This isn't usually talked about, but it's critical. Founding a successful company is about taking part in negotiation after negotiation after negotiation and winning on that. [1:48:07] Hiring, sales, fundraising are all negotiations at their core. [1:48:10] It's hard to be great at negotiating without having these instincts for human behavior. [1:48:15] Since you wrote that, as you've started a new company, any updated views or deepened views on negotiation or... [1:48:24] how to get better at it. [1:48:26] Yeah, I think this is one of those things where the core principles are really simple and can be explained very quickly, but then basically you have to live them in a bunch of contexts and you get better at it. [1:48:37] And so it's like, you know, BATNA is a really important concept, right? So that's best alternative, right? So if you don't, like, say you're applying for a job, like, [1:48:45] what is the thing you're going to do if you don't get that job, right? And so if you want to strengthen your BATNA, you'll get five job offers and then you don't have to like get stuck on any one. It's like really simple stuff like that, but

[1:48:55] The art is, I think, applying it to a bunch of different contexts where it can get increasingly hard. So as a simple example, like... [1:49:03] I think one thing I've just realized is exceptionally good founders... [1:49:07] are very good at creating options even when none really exists right so like say you are running out of cash and [1:49:14] the only thing that's going to save you is this one acquisition, right? But you don't want to look mega desperate. Like there's all these stories about Travis Kalanick doing stuff like this. Like, [1:49:23] he will just convince himself in his head, like, no, I don't need these people. These people are dumb, like... [1:49:28] okay, like, the other option is, like, I'll, you know, lay off X percent of my company and I'll just, like, eat ramen and, like, I'll do this. But it's a real option in my mind. But it's a real option, and so then I can walk into that room and it comes off in my vibe, and that's a very non-verbal thing. And so, yeah, like, to me, like, [1:49:43] Creating options have been a very... [1:49:47] core concept um i think the other thing is just uh for me like not [1:49:53] I don't know, like... [1:49:54] It's easy to get kind of psychologically mocked by people in the room, right? Like, especially people who are more experienced than you. [1:50:00] And the best position to just put yourself in is like not needing anything from them. Yeah. Right. And then they can't do too much to you. Yeah. Some of this is more emotional than even is strategic or tactical, I think. Yes. In some cases. Yeah. Like people will do all kinds of things to you. Like, I don't know, for example, VCs will make you feel like, oh, if you don't take this money, like you're going to fail or you're not ambitious enough or whatever it is. And I think you just have to. [1:50:22] you really just have to stay focused on what's important to you and not

[1:50:26] buy into other people's frames. [1:50:28] I think it's in the Palantir post, but you talk about this in different contexts, and you really admire people who are good at language and coining memes and coining language, whether it be Palantir or Tyler or... [1:50:38] there's jeff bezos is good at this too this compression vocabulary you you have this little section in there [1:50:44] listing a few of them. Ontology is an old one, but then there is Imple. [1:50:48] artist colony compounding the 36 chambers dots, metal metabolizing pain, gamma radiation. [1:50:54] I'm not going to make you explain all those, but are there any... [1:50:56] of those or otherwise that feel really memorable. [1:51:00] He, uh, [1:51:02] There are a lot, right. Or even the... [1:51:04] The sort of genius, like what makes the ones that are most effective, like what makes them so effective? [1:51:11] I think, again, they're sort of surprising, right? So, like... [1:51:15] The artist quality thing, I think it's just things that stick. There's that really good book. I think it's called Made to Stick or something. It goes into... [1:51:23] what actually makes ideas stick in your head, right? So often they're quite visual in certain ways. [1:51:30] They, uh, [1:51:31] Yeah, often they're almost like puns. I think artist colony sounds a bit like ant's colony in some weird way. [1:51:37] But yeah, I don't know. Cop would always have this bit of like... [1:51:40] Everyone we hire is... [1:51:42] like we don't do performance reviews basically because you can't measure people on this like unidimensional axis like everyone we hire is super multi-dimensional and like [1:51:50] I think he believes it and I believe it and there is a deep element of truth there, but there is also a marketing aspect too, right? Of like,

[1:51:58] you know, we're hiring these Picassos or whatever. [1:52:00] But yeah, it goes pretty deep to the culture of... [1:52:03] You all are. [1:52:05] you are like getting to know a person like fully all around and then like admitting to them, admitting them in almost as though they're joining a cult. [1:52:13] So yeah, I think that was one where... [1:52:15] It's kind of interesting... [1:52:17] The gamma radiation one is like... [1:52:19] this idea that you have to go through and feel a lot of pain very directly and that pain basically [1:52:25] kind of like, quote, radiates you into becoming a superhero or something like that. So it's like an Incredible Hulk reference. But the thing it translates to that is that [1:52:35] You basically want to be with the customer and you want to be doing the same workflows as them. And you want to be there like late at night. [1:52:43] trying to fix them very directly. And if you are not feeling pain, like if you are just in your nice air-conditioned office in Palo Alto and you're having your LaCroix and [1:52:52] you're like around a whiteboard designing the optimal workflow, you're probably doing it wrong. And so... [1:52:56] This phrase gamma radiation just compresses all of that. Strong imagery too. Exactly, exactly. Into... [1:53:02] a very pity phrase of like, wait, am I doing the wrong thing? Should I be like out there instead? [1:53:07] He's come up a bunch in this conversation, so maybe you even answered the question already. But what do you think the rest of us could most learn from Tyler Cowen? [1:53:14] Well, I think Tyler... You know, I mentioned a bunch of his concepts, right? So... [1:53:19] I think one thing he says is like cracking cultural codes. And that's kind of the same thing as what I was mentioning around [1:53:25] taking [1:53:27] areas that are unfamiliar to you, like the blues music or whatever, and like...

[1:53:31] very quickly kind of acting as an anthropologist of them and, [1:53:35] learning how they work and the fact that only... [1:53:39] immersing yourself in those contexts is... [1:53:42] There's like no substitute for it, basically. Yeah, so cracking cultural codes is like a big one. [1:53:47] I think for him, the importance of writing... [1:53:50] is something I've taken away a lot. So just write out your views on something. You'll realize... [1:53:54] that your opinions are not that thought out and like Paul Graham says it converts your ideas from vague to bad yeah that's a great great way of putting it yeah so so a big fan of like [1:54:05] writing. I think maybe what doesn't come out about him so much, actually, in his [1:54:09] writing or whatever is just like he's like a very kind person he's like really kind like he will he it's very hard to get him to say bad stuff about basically anybody and he's very good at like finding [1:54:21] the positive about people and like situations and basically anything right he's kind of a natural cheery optimist so i think for me like the [1:54:28] The thing I try and take away from him is like this kindness and like optimism and positivity. Very willing to speak his mind and not a douchebag to the other point. So rare. I'm like, how has this guy not been canceled? [1:54:41] My final question or topic, I think has come up in a couple of ways already a bit. [1:54:47] This is an excerpt from you. You say... [1:54:49] A common thing I see in late 20-something plus adults, they've tried a few things to make big improvements in their lives, and they mostly haven't worked that well. [1:54:59] So they give up on trying new things and instead settle into whatever grooves they happen to be in. Keep experimenting.

[1:55:05] Massive improvements are possible, are possible. [1:55:08] but you probably need to 10x the number of experiments you're doing because you're sampling from a heavily tailed distribution. [1:55:15] Majority of roles will be zero, but a few will be massive. [1:55:20] You talked about an idea I think about constantly life and sort of like defying inertia. [1:55:26] Maybe this is a blueprint for care? [1:55:30] What advice... [1:55:31] do you have for experimentation and eventually finding that groove [1:55:37] Perhaps it, again, returns back to this, like, quality without a name. You know it when you feel it. But do you have any... [1:55:43] Further advice on that? Yeah, I think some things do just kind of lock in and become constants in your life, right? Like, for example, for me, [1:55:50] um running uh became a constant like and the funny thing is like i absolutely hated it i so i played i played squash in college which is like a very specific sport but um and i played cricket which is like the most english thing ever right but like cricket games are eight hours long and so i didn't know that they're so long man and um you know i had exams i was a dedicated student so i was like okay i'll just run because i hate it but it's a very efficient way like 30 minutes i'm exhausted and i've done my exercise to do it everywhere yeah yeah and it just sort of stuck [1:56:20] into too much detail like there was a bunch of stuff i was dealing with and [1:56:23] I found it actually weirdly therapeutic. Like initially it was literally... [1:56:26] Like, [1:56:27] masochism it was like the pain was therapeutic or something but eventually that kind of transmuted itself into pleasure and i've [1:56:33] run consistently [1:56:34] every week, like, since then, unless I've been injured. And, um...

[1:56:38] Yeah, so some things it's like you don't need to try and... [1:56:41] over-optimized like that that is just something that feels pretty cool to my being um same with like i don't know journaling like [1:56:48] I don't journal consistently, but I kind of have always had some notebooks going since I was like... [1:56:53] 13 or something right so i think some stuff it's like it's good it's a constant in your life and then you don't mess with it [1:56:58] I think with other stuff, it's important to just be like, maybe I don't have this dialed in. So [1:57:02] um you know your eating habits is a really simple one right um i'm on a like really nice routine right now but i recently just like [1:57:09] changed everything basically and just tried a completely different diet and i found it worked a lot better and it made me feel a lot better so [1:57:15] I think that you have to have this intuition about your life of like what is working. [1:57:18] Well, what is maybe working not so well and so on? [1:57:21] Some things you hold constant. Like for me, the exercise thing, I basically hold constant, but [1:57:25] Other stuff like, you know, I just I said the thing to you earlier about not really believing in note taking. [1:57:30] It's not a fixed thing with me. Like, I will... [1:57:33] I think I originally tweeted that when I saw some new note-taking thing, and I was like, I don't want to try that. Note-taking doesn't work. And I was like, no, man, like... [1:57:39] You know, if you stop experimenting, like, you get stuck in these groups. I think that's when I tweeted that, but... [1:57:44] Yeah, I will always try the new thing, and... [1:57:47] If it doesn't stick, that's fine. And so, yeah, maybe note-taking is an area where I could use some improvement. [1:57:52] What I like about that, too, is that stability in certain areas... [1:57:56] is part of what can maybe allow you to... [1:58:00] take the risk of the change. One of my previous guests, Stefan, he talked about [1:58:04] just [1:58:05] in finding a long-term life partner, the amount of added freedom

[1:58:09] Because he had like that [1:58:10] That net? [1:58:12] And I think that probably applies in lots of big and small ways. Yeah, yeah. I remember I took... It was from Tim Ferriss, and he was just like, have... [1:58:19] five things going on in your life and then basically yeah like two of them will suck at any given time but you're sort of hedging the time right [1:58:26] So like, [1:58:27] you know, maybe your running is going really well and your company is going really badly. I mean, you don't want it to be that way, but at least the running means you're not in like the deepest depths of depression. And so I try and always have these parallel threads. [1:58:38] Nabil, that's all I got. Thank you very much. Thank you. This was a wonderful, wonderful conversation. I think lots to come back to. Thanks again. Thanks, man. [1:59:06] See you next time.