A (58:19)
All right, so there we go, Jesse. That was my discussion with Arthur Brooks. It's interesting how ideas seem to travel in packs. So for the years I was working on my new book about the deep life, which is coming out next year, it really felt like it was sort of out of left field and I was the only one thinking about this. And now all at once, there's lots of books about this same general idea. Arthur Brooks is one of these books. Jim Collins, they just sent me a new copy of his book that's coming out that's also about trying to find more meaning in your life. So I think it's just in the air right now that we're in a mode where people are ready to say, let's stop just talking about the problems and start also talking about the solutions. Now all of our books do something different, so they'll all complement each other. But I'm kind of happy to see that we're entering a moment in this year of questioning how do we cultivate a life of purpose as opposed to being so unique, like, focused very myopically on just individual issues that we might want to solve. I guess we call this, like, the meaning Renaissance. I don't know. Yeah. Did you end up writing your conclusion yet? No, what I'm doing, I haven't read the conclusion yet because I want to see. I'm in the first round of edits for my book, the Deep Life, and what I've been struggling on all week, honestly, is more of my personal story in the introduction my editor wants. And I'm pretty uncomfortable writing about myself. But there are some pretty deep motivations from, like, my 20s, where I developed a lot of the ideas like lifestyle centric planning that then play a big role in my theory of the deep life involving some stuff I went through back then. So I'm trying to write about it. God is going slow. I could write 3,000 word, you know, New Yorker essay easier than I could write this sort of 1500 words about, like, my own story. But it was also, like, completely unanticipated. Right. Like you didn't expect it do that when the editor came back with those comments? No, I mean, I knew she was right. I was just hoping she wouldn't notice. Oh, okay. Yeah. I was like, do I really have to talk about it? She's right. Like, when you're talking about something like cultivating a deep life, it is as personal as it is technical. So it is. Right. But, man, it's really slowing me down. But I'm almost there. I'm gonna try to finish it this afternoon, actually, after we record and then move on to, like, more normal editing that I'm comfortable with, which is about shortening stories, adding stories, clarifying things, cutting things that don't need to be there. There's my happy zone. I love cutting and simplifying, but, man, it doesn't help that I'm lying a lot. So, like, in my story, it's a lot of me in war zones. I basically just, like, took a lot from a mix of, like, Sebastian Junger's books about being embedded in Afghanistan and the Navy SEAL Richard Marcinko's autobiography of starting SEAL Team Six. So I kind of mix a lot of that into there. I guess it may be. Maybe readers will notice there's a lot of. A lot of me doing HALO jumps, high altitude, low opening parachute jumps into terrorist camps. But, you know, I Want to keep it, Keep it real. Keep it real. That's what I say. That's what we do here. All right, speaking of keeping it real, you've heard from me and Arthur and now it's time to hear what you have to say. Jesse, it's time to open our inbox. Now, just as a reminder, if you want to ask a question or share a case story or maybe just attempt to prod me into a ranch, send a note to podcast. I think we now have a three different people who read that inbox. So it's your best chance of actually getting your information in front of me and on the show. All right, Jesse, what's the first message we're going to cover today? All right, first message. Adam recommended an article for you to read. All right, so what do we got here? This is from Adam Scott. You think this is the actor Adam Scott from Severance and Parks and Recreation? Yes, let's just assume. Yes. Look, this is his words, not mine. Emmy nominated actor Adam Scott essentially sees me as like one of his biggest inspirations. I think we can just assume that's true. All right, what did Adam Scott have to say? He said, here's an article that will be, quote, right up Cal's alley. All right, so let's load up this article. I'll put up here on the screen that Adam Scott sent me. It's from Gizmodo. Here's the title. Tech employees are reportedly being evaluated by how fast they burn through LLM tokens. Is that terminology known, do you think, Jesse? Like, if you hear me say LLM tokens, you know what that means? I'm not sure if my audience does. Can you define it? So this is what you actually, how you're actually charged for using something like a language model. So when you use a large language model, you give it text as input, and the output is a single token, which is either a word or a part of a word. Right. So what the language model thinks it's doing is it thinks the input is from a real piece of text that already exists and that it is trying to correctly guess what word or part of the word comes next. So how do you get a whole long response out of it? Well, you have a computer program, like a chatbot, for example, would have a computer program that continually calls the LLM again and again. So you give it a prompt, it gives it as input to the LLM. It gets out one token, it adds that token to the end of your prompt. Now you have a slightly longer input, it feeds that into the LLM gets another prompt token, adds that to the end of your input, feeds that back in. So you're growing out a response one word or part of a word at a time. This is called autoregression, where you keep feeding back your output back into the input to try to grow the final output. And at some point, the LLM will output a special token that says, that's the end of my answer. At which point then you return that to the user if you're in like a chatbot scenario. So every time you produce a token, your input has to go through all of the layers of the LLM and all of the hundreds of billions of parameters have to be involved in multiplications. So that's the measure of how much computation is necessary for a particular response, is how many tokens had to be generated. So it's the same thing as saying how many, how many times do we have to call the LLM? All right, so when it says tech employees are being evaluated by how fast they burn through LLM tokens, it means how much they're using language models. So let me read from this article. This article is actually, it's quoting a Kevin Roos column from the New York Times. So really we should be reading Kevin's. But this is what the. This is what they sent me. So I'll read what's actually in the Gizmodo article. All right, so it says, and I'm quoting here, according to a column by the New York Times, Kevin Roose employees at companies including Meta and OpenAI compete on, quote, internal leaderboards that show how many tokens each worker consumes, end quote. At Meta in particular, and also Shopify, Ruse says volume of AI use has become a metric that goes into people's evaluations, with managers, quote, rewarding workers who make heavy use of AI tools and chastening those who do not. The resulting numbers in terms of both tokens and money, are absolutely staggering. One OpenAI engineer, according to Roos, burned through 210 billion tokens, which Roose equates to 33 Wikipedias. A Swedish software engineer claims to roost that his company spends more than his salary on Claude code tokens alone. And then, because I guess we have to Gen Z everything, he calls this token maxing. So there we go. What do I feel about this? Well, I think this is just pseudo productivity laid bare. It's a big idea we talked about in last week's episode, which is why Emmy nominated actor Adam Scott sent me this article this week. Is this idea that in knowledge work in general, we tend to use visible activity as a proxy for useful effort. The busier you are, the more stuff you're doing, the more productive we think you are. This is why we send tons of emails and jump in tons of meetings. Even if that's not actually producing more of whatever it is that makes money for our institution. Well, this is that made even more quantitative. I don't care what you're doing. I don't care if it's producing better software. I don't care if it's shipping more things than matter faster. I just want to see that you're making lots of hits on the LLM because that means you're doing lots of things. So it's pseudo productivity laid bare. But pseudo productivity is often a trap because there's lots of stuff you can do faster or more phonetically that doesn't move the bottom line. So what was my suggestion last week? My suggestion last week is have a better scoreboard. Measure the things that directly produce value. That's what you should care about. How many meaningful features were shipped to our software clients, for example? Maybe that's what we care about. Now if that requires a lot of LLM use or not. If you use a lot of tokens or maybe use few tokens because you have very, very careful, well constrained prompts and that actually makes you more effective than the guy that's just shooting left and right prompts left and right and getting clogged in all sorts of weird loops. Right? Focus on the scoreboard that matters, not whatever is more proximate and whatever is easier. So I think this is a great example of the digital productivity tool traps we fall into. Be very wary of looking at zoomed in speed of things and be much more interested in the big picture, actual production of value. Because what leads there is not always, doesn't always seem as busy or frenetic or fast paced as you might assume. So I do appreciate that article. All right, Jesse, what other message do we have? All right, next matchup, we have a note from an anonymous student who has a question about digital books. All right, let's see if we can find this. All right, here we go. All right, so here's the note. It reads, I'm a 22 year old software engineering student. I've recently been trying to apply your ideas on digital minimalism. But I have a question about reading digital books. Recently I started quote unquote, reading PDFs using a combination of visual reading and text to speech, which I listen to while following the text. This helps me stay focused and feel like I absorb more. Here's my question for developing deep focus cognitive improvements and strong critical thinking. Is this combined method as effective as traditional reading, or does it reduce the long term cognitive benefits of reading? All right, so I think in this context, yes, it is reducing the long term cognitive benefits of reading. What I think is happening here is that you are trying to reduce the cognitive strain involved in consuming the written word. So by having your audio system going, you're taking the strain off of just, my mind has to just purely decode these symbols and create meaningful representations in my brain. You're sort of short circuiting that the audio allows you to sort of take your foot off the cognitive gas pedal and just listen for a while and then read for a while, Then listen for a while. It is a lot less cognitively demanding way to consume words. But if you're interested in using books to help develop your brain, your ability to contemplate, which I define to mean your ability to actually control and aim your mind's eye at particular targets towards useful outcomes, if that is your goal, then you want the strain. What you're doing would be the physical analogy of saying, hey, good, good news. My Navy SEAL training was really hard. I hate the pull ups they make us do, but I figured out how to use a pulley system, and if I put some counterweights on the pulley system, these. These pull ups are much easier for me to do. I just feel like I can do them easier. Well, it defeats the purpose of the pull ups. You want to strain your muscles so they get stronger so that when you're in deployment, you can actually carry that rucksack for, you know, the long hike or whatever the analogy here is. So, no, you want to confront the actual symbols printed on a piece of paper. And I want you to change your mindset. That strain you feel, think about it like Arnold Schwarzenegger and pumping iron, loving the strain he feels in his bicep when he's lifting. He's like, yeah, that means I'm getting stronger. So I would much rather you do shorter reading sessions at full intensity than longer reading sessions where you're trying to reduce the intensity because you're not actually getting the cognitive benefits of increased contemplation ability, which it sounds like you're actually trying to get. So that would be my recommendation. There is a magic to decoding printed symbols with no other types of input that creates deep reading processes, strengthens those deep reading processes, builds cognitive patients with focus, and allows you to then reverse those circuits when the time comes when you're thinking or writing to produce much more original thoughts on your own. So stick with real books, read less, but keep the reading you do at a higher level of intensity. He also asked about audiobooks. I think audiobooks are a fine way to absorb information, right? I think it's, you know, hey, I listen to this book on audio, especially like a nonfiction book where either it's just entertainment or you want to get some ideas out of it. But if what you're looking to do is increase your cognitive capacity, you really want to read physical books for that purpose, right? So audiobooks are fine. But don't think of audiobooks as your primary way of building the strongest possible cognitive results. It's just less strain is involved. So I'm all for audiobooks. Like half my book sales now are audiobooks, but it shouldn't be the only thing you're doing. If you're trying to train your brain. Half your book sales are audiobooks. Not crazy. Wow. I knew that when I first got in the game. It was like none. And then when like audible became a thing and Amazon made it easier, it was like a quarter. I think I went back and looked this up for digital minimalism, which came out in 2019. If you're looking at those first year sales, it's like a quarter of the sales. Now if you look at like slow productivity is 50%. Wow, that's incredible. So partially that's a shift in book consumption habits and partially it's a reality of the fact that we have a podcast and I do a lot of podcasts and a lot of people encounter me to an audio format. So that also, that also bumps it up. Like I've noticed, like if you look for sales spikes based on particular publicity related events, if it's a podcast related event, you get a audiobook spike. And if it's a print related event, it's more hardcovers. It's like more evenly balanced. So like when I went on like Andrew Huberman's podcast on launch day for slow productivity, that was an audiobook book spike, right? In fact, my slow productivity was on the Amazon charts for multiple weeks, the top 20 most read or most bought books of the week. But it was the audio version because like, really I was on a lot of big podcasts. But if I get something like, you know, like a big article in the New York Times or something, then you're gonna get many more hardcover. So it kind of depends on the audience. But a lot of our Audience now finds me through this podcast, so they do a lot more audiobooks. The bad news is that means I have to record my own audiobooks and it's a terrible process. It's terrible. You explained it on a prior podcast. I know I'm not looking forward to it with the new book, but c' est la vie. All right. We're getting a little shorter today since we had a very long interview, but I like to end each show by briefly checking in with what I've been up to recently. Let's start with reading. I'm working on my fifth book of March. We're recording this on March 24th. So I read the first four books. My fifth book is mentioned is this Brandon Sanderson book, Misborn that my, my middle son has insisted that I read. Jesse, I'm about 400 pages into this book, which for a Brandon Sanderson title means like I'm basically finishing up the prologue. Like I'm just getting started. These are, these are long books. I'm enjoying it though. You know, the. I hadn't read him before. I'd read. I don't read a ton of fantasy. I read some Game of Thrones. I read some Patrick Rufus. I hadn't read this before. I had read a snarky profile in Wired. There's this like famous mean profile of Brandon Sanderson from Wired magazine. And I had read that and it was like really down on his writing. It was like, oh, it's like super expositional and he is explaining redundantly how characters were feeling. And I was like, oh, maybe this is going to be clunky writing. Like he just writes these things fast. But I don't think so at all. I think it's a very well paced adventure style book against a backdrop of a complicated world building magic system. I was like, this is very well executed. I mean, it's not Ursula Le Guin, but it's also not like I was expecting after that Wired article. I read a lot of thrillers, I read a lot of adventure writing. This is well crafted especially for a 6, 700 page book that keep the momentum going. Great world building. So I am impressed. I was more impressed by Brandon Sanderson than that snarky article led me to believe. Let's check it on the hq. I got a bunch of stuff Jesse, I'm about to bring over here. It's. Our foyer is full. We got a video game cabinet in there. We got. My 500 light is in there. We got a new. A giant rug. So we're going to rug the whole floor in there. So it's going to be not so live and echoey or whatever. I got a lot of stuff. I got the vintage video game maintenance manual from 1980 for the Galaxian arcade cabinet. Because I'm taking out some of the circuit diagrams to frame to put up in there. So, like, it's. I need this done by May because that's when my semester ends and my sabbatical begins. And how long does your sabbatical go for? It'll be the whole next academic year. So I need. Because I'm going to be spending a lot of time in there working, and so I need that to all get done by May. Here's the open question I have, and I'll get the advice of the listeners, is I'm putting picture ledges staggered on the wall in front of where the computers are, where I write and Jesse does video editing. I want to put books on them that are going to be like a source of inspiration. But I will take suggestions about what type of books to put there. My current thought has long been, but let me test this with the audience. I want to put first edition techno thrillers up on that wall because a. I associate that with my childhood reading in the 1990s, which was like a lot of creative energy and inspiration, the start of like my intellectual life. And two, I think of techno thrillers. I think of this idea that writing and thinking about technology can be. It can be interesting or fun or emotive or like, really be something that catches people attention. And I like that. You know, I don't want to put just like books in the mainstream of what I write about, like tech criticism up there or something like that. I want it to be a little bit more oblique. So that's my current idea, Jesse. But I'm open. I'm open to other suggestions that people have it. But I want a sort of inspiring wall of books with the spotlight for my 500 lights shining right on them when I'm sitting there writing. So that's my current thought. How many books you think, 10? Five? Yeah, I think 10, maybe. Yeah, let's think 10. I'm not doing first printings, but that's too expensive. But first editions, that's what I'm thinking. First edition hardcovers, we'll get them up there. They're red acrylic picture ledges. All this stuff's gonna be such a pain to hang, but I'm having someone come to do it. Yeah, just hang everything all at once, because it's all I hate doing that type of thing. All right, well, so send your advice to podcast cal Newport.com or if you're the Michael Crichton estate, I will lovingly put up your first editions you send me. They will be well displayed. We will. We will. They will find a good home. All right, that's all time we have for today. On Thursday, we have an AI Reality Check episode coming up. And then next Monday, we'll have another main episode. So until then, as always, stay deep.