How to Build a Cognitive Training Plan | Monday Advice

I've been thinking recently about an article from the Chronicle of Higher Education that several of you sent me. It was titled, "My Students Can't Read." Let me start here with the intro from the article. Here's what the intro of the article says. Six weeks into the term, I assigned my rhetoric and writing students a 20-page article. It was the same length I had assigned for five years and the same length I had read without complaint as an undergraduate a decade ago. Not one student finished it. When I asked why a student answered honestly, it was too long and she kept losing track of what the paper was about. Now when we hear accounts like this, our normal response is to ring our hands and worry that we're raising sort of a new generation of kids that are barely literate. But here's the follow-up question that we need to also ask, "Could you read that 20-page article?" Like honestly, if you were given that much reading, are you still able to do it? See, I think if we pull on this thread a little bit harder, what soon becomes clear is that we don't really have a great way of quantitatively testing how fried our brains have actually become due to digital distraction. But we need these tests, right? Because in physical fitness, we have measures known as key performance indicators like your BMI or how long it takes you to run a mile. They provide us a great way to get feedback on our physical health and gives us targets we can use to systematically work on improving. Well, here's my thought. We need something similar for our cognitive health. And this is what I want to talk about today. I'm going to propose three cognitive performance indicators that you can go and test today to see exactly how distracted your brain has or has not become. I'll then talk about how to systematically improve those scores if you're not happy with your results. All right, so if you've worried at all about your cognitive fitness starting to slip in recent years, then you need to listen to this episode. Let's get into it. As always, I'm Cal Newport, and this is DeepQuestGins, the show for people seeking depth in a distracted world. All right, before we get into these KPIs, I want to start with unfolding this motivation a little bit more. I want to go back to March to an article that I published in The New York Times, just so we can bring this up on the screen here. It called for us to take cognitive fitness as seriously as we take physical fitness. And here's a key quote from that article, I said, "I think we're finally ready for a similar burst of self-reinforcing action in the fence of our cognitive fitness. I'm done seeding my brain to core of all that makes me who I am to the financial interest of a small number of technology billionaires or the short-sighted conveniences of hyperactive communication styles. It's time to move past Freddie and about our slide into the cognitive shadows and decide to actually do something about it." All right, so what does that mean to do something about it? Well, this is what we're really talking about today. We're sort of answering that question that I first posed at the end of The New York Times piece from back in March, because here's what I think is going on. I think at the moment, we're in what we can call phase one of the cognitive fitness revolution that I called for in that article. So in phase one, we recognize there's a problem that we're losing our ability to think because of this. We're also trying random small things that we hope will help arrest this problem, like we're making general commitments to use our phone last or we're saying, "Hey, I should read more books," or maybe we're doing occasional detoxes or internet Sabbath is something like this. But as in physical health, just doing a few random good nature things often doesn't lead to the major improvements that you're looking for. So what we need now is phase two of the cognitive fitness revolution. And in phase two, I think we need numbers. So this brings us to the premise I talked about in the intro. Phase two is where we start to turn to actual quantifiable numbers about where your ability to think where your level of distraction actually are. So we can face clearly and boldly, this is where I am, and this is where I want to get. Now, I want to borrow a term that's used in physical training, which is, as I mentioned earlier, key performance indicators, which are sometimes shortened as KPI. These are numbers that you can track and try to improve through training. In physical fitness, there's a lot of these. I mentioned a couple before. Your body mass index, your mild time. Maybe you would have, you would have your max squat, or how many thousands of meters you can row on an earth. Of course, producer Jesse, what do you row? Like 20,000 a day. That's something kind of crazy. But we have a lot of these in physical fitness. We don't really have any of them when it comes to cognitive fitness. So that's what we are going to do today. Now, in particular, I came up with three cognitive KPI's. And I'm going to go through each of these. Each of them is going to be a specific concrete test that you could go after this episode or even pause this episode and go do, you know, take an hour or less, right? These are numbers you can get. And I'm going to tell you how to calibrate your score on these three KPI's to figure out what this says about where your brain is. So we're going to go through these one, two, three for each. Here's the concrete test. Here is what your score to this test means. Don't worry because once we've gone through all three of those KPI's, I'll talk about the best way to actually improve those scores if you're not happy with the results that you got. Okay, so let's get into it. Let's start with our first KPI. I'll bring it up on the screen here for people who are watching. I call this KPI reading endurance. Okay, so here's the concrete test for KPI number one. What is the max number of minutes? You can comfortably remain focused on conceptual nonfiction reading without context switching. Let's just unpack that a little bit by conceptual nonfiction. I mean, nonfiction reading that has concepts or ideas for you to try to understand, right? So not just a history or, you know, here's a book about fun facts, but it's a book that has some sort of theories or ideas or arguments, right? So conceptual nonfiction. Key caveat here, it should be understandable to you, meaning if you grab a book that's going to use terminology or reference frameworks that you don't understand, that's going to be basically unreadable. So don't just go grab Heidegger, you know, on bean and be like, I don't really understand what's going on here. It should not be anything that requires understanding you don't already have. So difficult, but within your current level of understanding, and you're going to read as many minutes as you can without context switching. So zero looks at a phone, zero looks at your computer, zero getting up to do something else. And when you find it, you can no longer really keep your concentration, stop the timer, then you have a score, a score that we could measure in minutes. All right. So how do we calibrate this score once you have it? I'm going to put a table up here that has five different categories of score. I'm rating these from most least cognitively fit, let's say to most cognitively fit. So the first level will be brain-wrotted, then we'll move up the distractable, then we'll move up to what I'm calling pre-2012 average. So in other words, this is going to be roughly speaking where the average person would be before the advent of smartphones and the distraction crisis that followed above that will be the level impressive. And then finally, I'm going to throw in cognitive weapon. And this is where we'll look at someone historically who has pushed his KPI to a far extreme, just so we can see what the outer limit is for the human brain's capability. All right. So how do we calibrate your score on this test to the scale? Well, let's go through these one by one. If your max minutes of reading the nonfiction is five minutes or less, that would put you in the brain-wrotted category. So you have a pretty severe impact of technology on your ability to, on your cognitive fitness. If you can make it to 15 minutes, that puts you in the distractable category, right? So you're more distractable than to 2012 average. I'm putting 30 minutes as my, that puts you at the pre-2012 average. That's like what I would expect, you know, someone 15, 20 years ago to be able to sit down and read for 30 minutes. No problem, right? What is impressive? I'm going to say 75 minutes. So if you can make it to like an hour, 15 minutes or more, you have a really good cognitive fitness beyond where we were before the cognitive fitness crisis really began 15 years ago. All right. What about cognitive weapon level? I'm going to say one book per day. That might sound crazy, but that has been done before. Who did that? Theodore Roosevelt. Our man, Teddy Roosevelt, could read an entire book per day. How long did that take? Well, let me read you this quote here from the Theodore Roosevelt Center. He read in many different languages, including German, French, Italian, and Latin on one day during the 1900 campaign. Roosevelt's schedule included six different mentions of reading, totaling 4 and a half hours of reading for a single day that was in in the middle of a presidential campaign. So I guess there is an extreme here. So all right, that's KPI number one. Jesse, where do you think you would fall? You can easily read very long for magazine articles. So I think, but the pre-2012 average would be no problem for you. I bet you'd have to run the plane for probably 45 minutes. - Yeah. - I don't have my phone or anything. - Don't forget to show us your face. There we go. So you're already at impressive. - I guess that's where I am, too. I mean, I can easily do impressive, but I cannot book a day. I mean, I could recite, maybe. But then Roosevelt, I wanted to read a book a day for. Supposedly, during his time in the White House, like most days, not crazy, in the White House. I don't know how, I mean, I write about Roosevelt, Teddy and Deepwork, and I talked about how he, like part of his secret weapon is intense focus. And he writes about it, especially during his time as a student at Harvard. He writes about his ability to what the kids now would call locking in. He could like, lock in because he realized that was the most, he wanted to get things done. And like, the most efficient way to get like his studies done or whatever was if he could give it intense, crazy concentration, it would take less time. Which is a formula that actually shows up in my much older book, had become a straight A student. Total work produced equals time spent, time's intensity of focus. So you could, that's the thing to hack, is intensity of focus goes up, time required to do the same amount of work goes down. So Roosevelt, he kind of pushed out to an extreme. Plus it's so impressive, he's doing it in different languages. Yeah, look, what did this quote said? German, French, Italian and Latin. We must be wasting so many cycles on media, right? I mean, he was a busy guy. And he spoke four languages and would read a book a day and did all sorts of other sorts of things. I don't know, man. You like to look around on the library? I know. Oh, that's really sweet. I would like, I have not yet, I want to visit Oyster Bay. So he has their house up in wherever New England. I don't know if it's up in New York State or Maine. I think it's upstate in New York. They're the house they would go to. And it has this like famous office in it that has a lot of animal heads on the wall. Which we will soon be doing here at the HQ. Only endangered animals though, because I kind of want to make a point. A lot of koala bears. All right, let's take a quick break from talking about these performance indicators to hear from some of the sponsors that makes this show possible. But stay tuned because we have some more cognitive fitness to get to right after we're done with this. All right, so I want to talk about our friends out Caldira Lab. You've heard me talk about them before on this show. They make high-performance skincare designs specifically for men's skin, which is 25% thicker, oilier, and ages differently than women's. While they have one product in particular that I want to emphasize today, which is their eye serum. Guys, let me tell you, as we get older, you know, just in our young men anymore, we're in our early 30s now. So we're now as young as we used to be. The first thing that goes are your eyes. They get puffy, they get dark, which makes us look more tired than we already are. This is where Caldira Labs eye serum enters the scene. It's simple. You just put it on once in the morning, and again, before bed, and it helps improve the appearance of dark circles, puffiness, fine lines, and crow's feet. I've just tried it, and I've discovered that it's a small habit that delivers big results. So if you'd like to look a little more arrested and refreshed, visit caldiralab.com/deep and use code deep. To get 20% off your first order, that's caldiralab.com/deep and use code deep. I also wanna talk about our friend's Advanta. What's the one thing in business that's spreading as fast as AI? AI risk. Every new tool your team signs up for. Every vendor that turns on AI features, every new integration, each one is an opportunity for something to go wrong. And most security programs weren't built for AI's pace of growth. This is where Vanta enters the scene. Vanta is the number one agentic trust platform used by over 16,000 fast-moving companies like Ramp, Cursor, and Harvey, to ensure they're always audit ready. And now Vanta is helping companies like yours watch for the risks that show up between audits, across your vendors, your AI tools, and your whole environment. How does it do this? The Vanta agent works like a 24/7 GRC engineer in the background, finding issues, drafting fixes for you, and cutting vendor assessment time by up to 50%. So whether you're a fast growing startup or a global enterprise, Vanta is here to help you automate your security and compliance and earn and prove trust. Get started today at vanta.com/deepquestions. That's v-a-n-t-a.com/deepquestions. All right, let's get back to our discussion. Let's move on to KPI, let's see here. KPI number two, we'll bring this up on the screen. I'm calling this contemplation control. Remember, these are performance indicators for cognitive fitness. What is the contemplation control KPI? Let me read you the concrete test you do to measure this. Attempt to make progress on a personal or professional problem, working only in your head while walking. Keep redirecting your attention back to the problem if your mind wanders, keep going until you can no longer make reasonable progress that you just can't get your attention back. The fans of my book, Debork, will recognize this as the activity of productive meditation, which I long talk about is one of the most effective ways to increase your ability to contemplate your ability to actually direct your mind's eye intentionally. Now, the key thing about this test, your mind will wander. That doesn't stop your clock, right? You will think about an email you want to send or a conversation you just had. But as long as you can keep bringing your attention back to the problem and make more progress, that's okay. It's when you find yourself unable to really make more progress on the problem. Your mind is all over the place. It's exhausted. That's when you stop the timer and you got a number there in minutes about how long were you able to control your contemplation in this experiment? All right, so let's calibrate this. We got our performance levels here from Brain Rotted to Cognitive Weapon. How do we calibrate? How long you can do productive meditation on this scale? Here's the way I'm going to say it. Less than a minute. So basically if you can't do it at all, that's Brain Rotted. Now, if you're not in a younger generation, that might sound crazy. You might be saying, well, of course I can think for a while. It's actually very hard to hold your mind's eye on something, push away the distractions and make progress on it. And I think there's a lot of people, especially if you grew up completely distracted by these technologies. So I can't do this at all. Obviously, the comp here is a regular meditation, like mindfulness meditation. A lot of highly distracted people have a hard time doing it at all. It's like unbearable. They can't, the thoughts are so wild that it's sort of unbearable. So I'm going to say if you can't really do it at all or more than a handful of seconds, that's Brain Rotted. All right, if you can do one to three minutes, I'll say you're distractedable. Like, okay, have it, kind of go. Yeah, yeah, and then I lose it after a few minutes. What's the pre-20, 12 average, 10 minutes? This is hard. Right, productive meditation is hard. I talk about this in deep work, how, especially as a post-doc at MIT, I had a very intentional productive meditation habit. And it made a big difference over time. Like, I really got better at this, but it took practice. So really, if you can sit 10 minutes of really making progress in your head, that is our sort of average from the predestractable age. 30 minutes is impressive. I could do 30 minutes to an hour now. I could do like pretty substantial outlining or riding in my head or make progress on a specific issue of a proof. I think over an hour would be difficult for me. All right, cognitive weapon territory. So how do we push this to an extreme? The cognitive weapon territory is six to eight hours. All right, so who did that? Is that really possible? Yes, our friend Friedrich Nietzsche. Let me read you a quote here that I got from a Margelian article. And then this is actually quoting Nietzsche himself saying, I would walk for six or eight hours a day, composing thoughts that I would then later jot down on paper. All right, so we got a pretty wide range. 10 minutes is average, but Nietzsche could do this for six to eight hours. All right, so there we go. That's your second KPI. That's probably the quickest one to actually measure because it's the hardest, it's pretty hard. So you go out, you try it, you get your measurement on it. Have you ever tried, I don't think you ever, you know I talk about all the time. Do you have you tried productive meditation? I bring it up a lot. I try it when I bike to work. Oh, I'm not necessarily walking, but I bike to work. But aren't you thinking about not getting killed by cars? Like isn't on the path? Oh, okay. I don't have to worry about cars really. Can you make progress on things? I'm probably pre-2012. You think about pre-2012? I love productive meditation. I always say that's pull-ups for your mind because it's a similar level of hardness. Like pull-ups kind of suck. But if you do 'em, as you're able to do 'em, it works out all the muscles or whatever. You know, I'm back to those now. After my injury, my surgery, I couldn't do 'em. And now Zach has me slowly back doing like five at a time. The problem with the way he does demands that I do pull-ups, though, full extension. All the way, arms lock completely full extension before you start the next one. So you have a bar in your garage where you can feed or off the ground. - Yeah, yeah, yeah, I have to bend my legs, but I can get to my, yeah, but it's a seven foot. I guess it's a six foot bar, seven foot bar, I don't know. So there you go. All right, one more KPI that I want you to consider for testing your cognitive fitness. This one's a bit of a pain. I think it's critical. Honestly, I think it's the one that AI is probably gonna have the biggest negative impact. So those first two, I think smartphones have probably pushed a lot of people below the pre-2012 average. It's also gonna affect this one. I think AI is gonna have a particular big impact on this. All right, so let's bring it up. KPI number three, I call written persuasiveness. Here's the test. Make a written argument for a contested stance that you believe it. Contested mean there's multiple points of view on this. Different people believe different things. There's a point of view that you believe in strongly. Argu it in writing. Type it out, write it out. That's the test. All right, so how are we going to measure this? I'm actually going to assess your response. So the way you're going to assess your response is by the format and style of what you wrote. So let me go through this, right? We'll go from brain-wrotted to cognitive weapon and get a better sense of what this means. All right, so how do we describe a written response that we would call brain-wrotted? I would say if what you wrote down was disorganized, appeal to emotions and relying on ad hominid attacks. So just sort of, let's think a tweet. Let's be honest, we're talking about a tweet here. It's like, I don't know this or that, or I don't know, the other people are just evil and bad and terrible and they're communist or they're racist or whatever like side of it's on. And just like everyone knows, right? Like basically Twitter. Is there anything you're going to see on Twitter? All right, then what would be the properties of a written response that would get up to the distractable category? Reasonable list of arguments and favor of your stance. So if now you can organize your thoughts until like, let me just put a few bullet points here because of this, because of that, because of this, these are like three or four things that I think are, speak to my particular stance or sort of support that stance, right? Now as you can list and enumerate things, actual arguments. Now you're at distractable. What's the pre-2012 average? I would say a pros argument for your stance and against the opposing view. So now you're at a bullet point, so you can put an answer in pros and paragraphs like you would in school and you're not only able to make an argument for why your stance is correct, but an argument for what is wrong about the other stance. And I think for this to really count as the pre-2012 average, it has to be a good faith portrayal of that other stance, right? So in other words, like, if it just characters it, like, and the problems with the other stance is that everyone who believes it is a loser, right? Like that doesn't count. But if you're able to actually be like, yeah, this is where the other stance comes from and I kind of can give you a pretty good characterization of it and then say why that is worse than your stance. That's how we would get to that particular response. All right, so what would get us to the impressive level? I'm going to say a powerful argument from first principles. So what makes a really strong argument from in my opinion is you're able to lay out, when I say first principles, we can think of these as axioms that everyone agrees on. These like, okay, we kind of agreed this is a common starting place and you're able to build a logical scaffolding sort of inductively from those first principles that leads inevitably towards your stance being correct and the other stance having flaws. So now you're actually have an impressive level of written persuasiveness. What is cognitive weapon here? I'm going to say for this an argument that forever changes the world. So not only is what you write something as persuasive, it changes the way like the world understands or thinks about things. You change, it's so persuasive that it changes the body knowledge. My example here would be Rachel Carson. Let me read you from the National Park Service websites. In the US, Carson's work led to the establishment of the Environmental Protection Agency and the Firth Earth Day celebration in 1970. Silent Spring influenced several federal conservation policies, including the band of DDT in 1972. It completely created an environmental movement and a whole government environmental apparatus and regulatory apparatus that we take for granted today did not exist. Her argument in this book was so powerful. It completely changed the way that we and the government thought about the environment. That is a very persuasive piece of writing. All right, so we've seen three KPIs that you can measure for your cognitive fitness. Do it. Do at least one of these. Get a score to see where you fall. The reason why you might be hesitating is because you're worried about what you will find. What if you're distractible or even brain-rotting? Maybe you'll think this is like a neuronal scarlet letter that will hold you back from ever succeeding with the life of the mind, but this is not true. Just like with physical KPIs, the reason why we measure them is so that we know what to improve. It's a number that we can look at. Do effort and see what makes it larger. We can focus our efforts. Cognitive fitness can be improved. Therefore, your KPIs scores, if you're unhappy, can be boosted upwards. All right, so how do we do that? Well, again, let's look back to the physical analogy. If we have a physical KPI like your mile time or your squat max squat weight or max number of pull ups or whatever it is, what do you do? You train to improve that score. And what does training mean? Well, it doesn't mean just in general trying to be more healthy. I'm gonna try to like take the stairs more often and not eat as much junk food. That's not gonna have an impact on your mile time. That's not gonna have an impact on your max squat weights. What is gonna matter? Specifically training for those things. I'm actually running certain distances, trying to increase those distances, increase what distances I'm comfortable with, increase my speed, I'm starting with lower weights, getting my form just right, increasing the weight slowly. That's how you get those numbers to go up. And of course, the idea is getting those type of key indicators up is gonna make you more healthy in general. So if you're putting a lot of effort to specifically train for your mile time, your cardiovascular health in general is gonna get better. It's gonna have all sorts of positive, secondary side effects. You're gonna be frustrated when you get stuck and make other changes in your life. Anything that's gonna get that number down. So that's what we have to do with your cognitive KPIs. You have to train specifically those KPIs, all right? So you have to say, let me take the activity I'm measuring and try to push myself towards the score that would get me to the next level. That'll be very uncomfortable at first, but really push myself until I can more consistently get up to that next level. Maybe I'm gonna throw in some tactics or form the cognitive equivalent of form here to help. Like if I'm struggling with reading, I'm gonna maybe have some note-taking styles. Be careful about what books I choose at first. Like I'm gonna bring in some technical form or techniques or whatever's gonna help me slowly start inching that score up towards the next level. So you train those cognitive KPIs just like the physical ones. And just like in the physical world, again, obsessing over one of those KPIs and moving it up is gonna make your brain in general more healthy. There's just no way you're gonna get to a place where you can do 30 minutes of productive meditation or read for an hour or some sort of conception of nonfiction, where your ability to focus and control your mind's eye in general is not going to improve in all sorts of different aspects of your life. So we're gonna use these very precise numbers. We're gonna train those precise activities as a very specific way into the general goal of becoming more cognitively fit, all right? So that's really where I wanted to bring you in this deep dive is to this idea of training. You measure a couple of these, you see where you are and then you get to work training that activity, doing what it takes to move it to the next level, that is how you get your brain more healthy. I mean, the idea here, Jesse, is that in phase one of the cognitive fitness revolution, we're really doing the equivalent in the physical world of just saying like, hey, try to eat better or try to move more. That's what it is with what people are doing now. Like I made my screen grayscale on my phone or I'm trying to like not have my phone when I watch TV though I often do, but like I'm kinda trying not to. We know trying things that are easy in the moment, et cetera. Get you nowhere in physical health. It's the same thing with cognitive fitness. So this is what unlocks phase two is now having numbers that you can specifically train for. There's a lot more that's involved in cognitive fitness. Obviously there's all sorts of different things to train. There's general habits, there's these specific APIs. This is just one element I'm adding to this bigger canvas I hope to build of thinking through how to become more cognitively fit. All right, so that's a enough hearing about what I find interesting. I also wanna hear what you find interesting. This is where we like to do on these Monday episodes where we open up our inbox to see what you have to share or ask me about. Remember if you have a question or a comment or something that you think I might find interesting, you can send the two podcasts at calnewport.com and maybe we'll get into it on the air. All right, Jesse, who's our first message from? Our first message is from Gerald who wants to know your opinion on a particular type of syndrome. Ooh, I like syndrome. Sorry, let's see what Gerald has to say here. He says, "I think the concept of information fatigue syndrome needs to be. more mainstream. I never heard you mention that term before and it seems right up your alley because the solution is essentially going on a data diet. I guess low quality data is the new ultra processed food. I interesting just information fatigue syndrome. I haven't heard that before. Let's find out more. He sent us a link. Let's bring it up on the screen. We'll take a closer look here. So he sent us an article from Workplace Psychology called "Information Overload." Too much information becomes noise. This is a summary of an academic paper. We'll look at this briefly, but here's the intro here. It says, "An information overload causes symptoms and solutions. An article for the Harvard Graduate School of Education's Lonely Innovations Laboratory." Joseph Ruff says that we are bombarded with so much data that we are on information overload. Simply put, information overload is when our ability to process information has passed its limit and further attempts to process information or make accurate decisions from the surplus information leads to information overload. Here's an actual quote from that article. "Once capacity is surpassed, additional information becomes noise and results in decrease in information processing and decision quality." The summary goes on to say there's even a name for it, "Information fatigue syndrome." Here are the symptoms, poor concentration, polyphysic behavior, or multitasking, hurry sickness, pervasive hostility, habituation, or overstimulation, plugged-in compulsion, traditional stress, and fluid-lowering immune response. All right, I believe it. I mean, let's think this through here for a second, Jesse. It's saying if you burn out your information processing circuits, essentially, any more information that comes in, you can't process and it begins to lead to negative fatigue style side effects. Didn't we talk about something like this? Didn't we have on? Remember, we did an episode, maybe early summer late spring, where we were looking at a claim on Reddit, and we had a psychologist from Georgetown way in, and it was about not information, but stimuli, and to claim, if I remember this correctly, the claim that we were contesting or analyzing was that once you pass a threshold for stimuli, you get overloaded with stimuli and then are super stressed out and can't really function. And the psychologist came back and said, "That's not really true. You can be exposed to stimuli all day long without reaching some capacity that then puts you into a negative state." And his example is, if that was not the case, people who live in Singapore or Las Vegas would constantly be just like overloaded to unable to function, because there's a huge amount of stimuli all the time. But he did say, if I'm remembering this properly, that really concentrated activity is something that you can sort of run out of steam for. So, I mean, I believe in thinking fatigue syndrome, that there's only so much like hard thinking you can do in a given day, and that's also determined by how exhausted you are physically or mentally with other things. And once you get there, it's very hard to do more thinking. But information processing, maybe, that's interesting, if it's just reading, or I mean, I honestly, I think it's a concentration that matters, not the information. Like, if you're, the information you're taking in is hard to consume, you're reading something complicated, you'll run out of steam. I think that's a real thing. If the information is, you know, Netflix, I don't know, you can do it all day. You're processing information. People do do it all day. People do it all day, right? And they're not like more, in fact, they're less exhausted than someone who is like doing something else. So, I think there's a kernel in here that's true, but there's also maybe a temptation of something just makes sense, and then the name looks good, but it's not quite getting at what's true there. So, we'll put that in the category of, like, interesting, but probably not quite accurate as stated. And let's take a quick break from our inbox segment here to hear from some other sponsors that make this show possible, but make sure you stay tuned because coming up after the sponsor break, I do have a rant, so be ready for that. All right, so let's do the sponsor break real quick. Look, I've experienced firsthand that the thing about running a small business is you don't, you don't just do your own job, you end up as the hiring manager, the payroll department, and the benefits team, all at once. This is where Gusto enters the scene because it can help take these responsibilities off of your plate quickly and seamlessly. Gusto is online payroll and benefits software that's built for small businesses. It's all-in-one, remote-friendly, and incredibly easy to use, so you can pay higher onboard and support your team from anywhere. It does everything from automatic payroll, tax filing, simple direct deposits, health benefits, commuter benefits, workers comp 401k. You name it, Gusto makes it simple and has options for nearly every budget. It's quick and simple to switch to Gusto, just transfer your existing data to get up and running fast, plus you don't pay a cent until after you run your first payroll. Our ad agency here at the Deep Questions podcast uses Gusto, so I can tell you from personal experience as someone who receives these payments that it has made my life quite easy. So try Gusto today at gusto.com/deep and get three months free when you run your first payroll. That's three months of free payroll at gusto.com/deep one more time, gusto.com/deep. I also want to talk about our friends at Babel. Jesse, I just signed up to do a give a speech next year out in Scandinavia, and I want to brush up on my language skills, so at the very least I can talk and say a few things to show a little bit of respect to the local culture. Look, I don't want to be the guy that only speaks English everywhere he goes, but luckily I have Babel to help me out. So if you're traveling to summer, here's the travel hack. Don't wait until you land to start learning the language, instead use Babel. Even just 10 minutes a day with Babel can help you start having real conversations, and as little as just three weeks. Instead of memorizing random vocab, you'll learn phrases that you will use doing things like ordering dinner or asking for directions, or just chatting with the locals. The Babel lessons are quick, practical, and built by more than 200 language experts. Babel's award winnings app has sold over 25 million subscriptions and is backed by a 14-day money back guarantee. So if you have travel coming up, now is the time to start, so you can actually use what you learn on the trip. Right now, it's important, Babel is offering listeners up to 60% off. Go to Babel.com/deep. That's Babel, B-A-B-B-E-L.com/deep for up to 60% off rules and restrictions may apply. All right, Jesse, let's get back to our inbox segment. All right, who do we got for message number two? Our next message is actually from a publicist about a new book. Oh, yeah, so I was okay, I remember this. I don't want other publics, I get a lot of public press releases to our account. We delete them all, but I came across this one and I read it. Please do not read this as a excuse publicist to send me even more press releases. But this was a press release that a publicist sent me for a new book, and the book sounded interesting, and there was a quote about the book, a summary of one of the points from the book that I thought was really cool. So let me read you the intro to the book. Here's what the public says set. I want to give this book credit. In America, there seems to be a right way to do everything, even relax, we track our sleep, optimize our mornings, listen to podcasts on our walks, and feel vaguely guilty when an evening produces nothing at all. For best friends and lifelong parisians, Alex Durand and Tom Carls, that instinct to make every minute count was one of the biggest culture shocks they encountered after moving to the US. Their new book, Pardon Our French, the parisian way to a happier life, which comes out on September 8th, offers a new way to think about ambition, leisure, and pleasure without asking Americans to give up their drive. All right. So what was the point from the book in this press release to caught my attention? I'll read it to you now. Here was the point was summarized as treat time like wine, not water. Rather than managing time as a scarce resource to conserve and maximize, Alex and Tom suggest treating it more like a glass of wine, slow down, savour it, and stop rushing towards what comes next. I think that's kind of nice as opposed to chugging water. Yeah. Yeah. Yeah. Actually, that's a, well, I'm thinking about that for a second. I get the first half of this. All right. Explain this to me. I get the first half, treating time like wine, enjoy it, savour it. That's positive. This is very much like an Oliver Birkman type of thing that like, wow, I have like an hour of just being here and enjoying it. I don't get the not water part. So you think it's because-- Well, if you're like really thirsty and you chug water, maybe. Yeah. Yeah. You don't think about it. It's just functional. There's like chugging it. All right. It's pretty good. Anyways, I like that treat time like wine. So that's good. Maybe that book comes out. We'll have to take a look at it. Have those guys on. Arrays for the full episode. And then you can do your accent. Baguettes under our arms. Full episode. Wonderful accent. Pipe. Pipe. Full episode. Stripe Sailor shirt and full episode. Interrupting the guest every five to ten seconds. Excuse me. My shirt. My shirt. I have a little bit of a French word. Pediature, jocke en prance. See. Who's probably trade a lot more? So they can speak a little French. - We need Teddy Roosevelt here to be on that interview. - Yeah, first of all, I'd be like, hold on a second. (man grunting) Book done. And then he would be like, I'm gonna conquer an annex France. - Yeah, 'cause that's the way he worked. - All right, what do we got? What's message three? Our next message is from REMA, who is highlighting an interesting way to think about AI in thinking. - All right, let's see here. I'm always nervous, but interested in these AI ones. REMA says, "I'd like to share this interesting blog post series from the scholarly kitchen blog on cognitive responsibility framework and cognitive coherence and cognitive debt. As someone working at scholarly publishing at a university, I found this framework to be incredibly useful for thinking about the use of AI and academic publishing." All right, let's look at this. Jesse, let's bring it up on the screen. I have the article here. Reading between the lines part one, a cognitive framework for AI scholarly publishing. Let's see if there's any interesting thoughts here. Here's the intro. As an editor, an editor asked AI to summarize a stack review reports before reading them herself. Days later reviewing her own decision, she finds herself wondering whether the judgment she reached was fully her own. That small moment points to a much larger question now facing scholarly publishing. A generative AI has transformed publishing, scholarly publishing in three years, dot, dot, dot. It is now a permanent feature of the scholarly landscape. The question's no longer whether it should be used but how it should be understood and governed. All right, so I'm going to skim down here a little bit. So there's a framework that they're going to suggest. The framework is not intended to classify people or rank scholarly activities. Rather, it offers a practical vocabulary for deciding where AI can appropriately extend human cognition where human responsibility must remain primary. So what are some of these categories? She says there's a distinction between mechanical cognitive work which AI can increasingly perform and judgment-based cognitive work where human responsibility must remain central. All right, I like this general approach. You know, as I do more digital ethics and think more about AI and my academic roles, I think there's an approach here which is useful, which is again, I think, right, the way this is said is right. Clarity and vocabulary about cognitive activity, so we can better discern where something like AI makes sense and where it does not. In other words, we can put structure to our otherwise vague or incoherent intuitions about something feeling icky or not. So this particular distinction, which I just summarized at the very highest level, but of cognitive versus mechanical rather versus judgment-based cognitive activities, that's useful if there's some sort of human judgment or originality applied, that should come from a human. If it's mechanical, we got to take these numbers and we got to put them into a chart and we got to format that chart. And there's no judgment here. It's just mechanical and annoying. AI could help speed that up. I think that's useful. I would add another category when I'm thinking about AI, I do not think there should ever be communication, written text communication, that was like in any large part generated by AI, that's not clearly labeled as such. I think it's cognitive and emotional fraud to have text that presumably came from you, that was actually from AI, because the way that text is actually ingested and processed on the receiver's end is that you're simulating the mind of the sender. You're simulating another mind. And if you believe you're simulating the mind of this particular human who's transferring interstates to you almost telepathically through these artifacts on the written page, and it turns out it's not their mind, it's actually a machine. I feel like there's some sort of psychological or emotional fraud there. Riding is sacred and it's something that we should be really careful about. So anyways, I like this general, that's like my opinion. But that's that this approach is useful. You're starting to break up different activities and be clear about where humanness is important and where mechanizing might make more sense. So I don't know if you've heard about this, Jesse, we should cover this in another episode, but Anthropic came out and said, we're gonna introduce water marking into that text output of Claude, meaning it will be very reliable for people to determine if computer code or text was generated by their LLM versus a human. And they're doing this because of compliance with some European AI legislation. And it's freaking people out. Because people are really freaked out about this idea of like, wait, so everyone can always tell if I wrote this or had an LLM do it. And I'm saying, if that freaks you out, that is a sign that you weren't, you knew this was like, there's something icky here. There's something like, if you're not just proud of like, sure, this is AI. If you're kind of hiding it, it tells you there's some sort of ethical boundary that's being crossed, even if you can't put your finger on it. So that's why I think these type of efforts are useful. - But those people are just gonna use another AI. - Yeah, I would have to be universal across the AI. So you know how it works? It's actually kind of interesting. It has to do with, there's a lot of places when you're generating tokens with an LLM where there's multiple different words that are synonyms that you could use, right? It basically, not to get too technical, but if you use a large language model, it creates output one token at a time where a token is a word or a part of a word. And then you just keep calling it again and again to generate more and more tokens to get a full response. And what it actually gives you is not, this is the specific next token to expand your input by. It gives you what's known as a probability distribution over possible next tokens, right? It's trying to guess, because again, an LLM thing, so I don't mean to anthropomorphize it, but the way it's trained is that the input is given, it thinks that's a real text that exists in the world that's been cut off and is trying to guess what the next word is. And it gives you a probability distribution over guesses. I think there's a 20% chance it's this, there's a 21% chance it's this. A lot of words have 0% chance, like that just makes no sense. But often there's a bunch of words that could work and they might have kind of equal probabilities or kind of synonyms for each other. So what they're doing, I don't completely understand this yet, but essentially they have a pseudo random number generator. So like if you have the same, a kind of a sequence of random numbers, but if you have the same seed, you can create the same sequence. And they use that to resolve choices where there's multiple tokens that are basically equally likely. They use this random number sequence that choose which of those tokens they, which of those synonyms they choose. So if I'm on the other end, if you give me that same pseudo random sequence, I can sort of go through in some sense and say, hey, if there's a bunch of words that match this sequence, then for sure this was an LLM that was running this sort of watermarking software, right? It doesn't change the meaning or correctness of the text because these are like synonyms words. So it's something like that. I've that, there's details a little bit wrong. But anyways, the point is it's, the watermark is detects itself, not some hidden digital, whatever. So you can't just like retype what you see on the screen in like a fresh window like in Microsoft Word and get around, it's the words themselves. So you really would have to substantially edit the text to get rid of the watermark at which point, you're kind of riding on your own anyway. So I think that's cool. I think because of my emotional psychological fraud argument, all LLM tech should have that. The other thing is it's unclear whether you can copyright text or code proofs by LLM's, I think that's great. I think copyright should protect the human brain's output, right, not the AI company. So there's a lot there that's interesting. All right, let's do one more question. What do we got? Our last message is from Nathan, who had an article to share that's right up your alley. Oh man, what Nathan says, Cal. This seems to have everything for you. Disneyland, comments on high quality leisure and references to the need for slow productivity and deep living. Let's pull this up on the screen here, Jesse. What he was showing is that there's actually like a five part, and he's the first of the five parts, a five part article on Cory Doctrel's blog, pluralistic. Let's see, and that's not it. Here it is. Board of minutes, discontent. All right, that's part two. So let me go to part one. Part one was called, are we having fun yet? Anyways, it's a five part discussion about essentially the experience in Disneyland. And it's Cory Doctrel, he has all sorts of theories. Well, a lot of it's about the line experience and all the things they've tried to try to make the lines work. I'm gonna be honest, yeah, right, all five parts. You can imagine. - He tells me about waiting in line, right? - Waiting in line, yeah. Here was the kind of, I'll summarize the coolest point of this piece. He goes back and reminds people of the original technology, this is probably in part two, so I'll go there. The original technology for dealing with lines in Disneyland was this. So when you, it cost a certain amount of money to get into the park, but you couldn't go on any rides without a ticket. And so you would also buy for separate money, a ticket book like the one I have up on the screen, and you had a certain number of A tickets, B tickets, C tickets, D tickets, and E tickets. Like two E tickets, and as you go down, you have more and more of these tickets. So like A tickets were for A rides, which are like small things. We like taking the horse-drawn carriage down the main street. D would be for like pretty extensive rides, like Mr. Toad's Wild Ride, and E tickets would be for like the really big rides. It would be, I don't know if they would have been back then, but like by the '60s, the Haunted Mansion, or Pirates of the Caribbean, or whatever. And what he argues is like this, having these ticket books, it changed your relationship to Disney, and we could take this off, probably. It changes your relationship to Disney because you could only go on two big rides, right? And you would think what those were, and there would be like two big rides that would be it. Which means there's never long lines for those big rides, because everyone in that park could only go on two, and you would choose to two. you really want to go on and you kind of ration out your experience. I'll do some of these A's and let me get two E's and three D's. You have a lot more downtime and because of that the experience of the park itself was a bigger part of it. You would just walk around. There'd be shows. You would see things. You would enjoy just being there. Maybe I'll take a horse-drawn carriage. I'm really building up to this afternoon where I'm going to go on one e-ticket ride. And basically, once they got rid of that system and went to all you can eat, you pay it again in the park and then can go on any ride you want. It's like the whole thing fell apart because everyone just went to the e-ticket rides. And the line's got really crazy and it's just been ever since then trying to deal with that and increasingly complicated ways, none of which ever really solved the problem. And the original park experience with these ticket books was in his argument the best one, right? And it also, yeah, you'd come back. Let me come back again. I'll go on some other ride some other times. It's just fun to be there. It's beautiful. It's interesting. And so I liked that. I liked this because in some sense, it's riding that sweet spot of things I like. I mean, Disney, of course, but technology and technology's intersection with our human experience. This is a key point where the technology of tickets in line access to rides completely changes your experience of your life of the day that you're in this park. And we see this type of thing all the time with all sorts of different types of technologies in our life at home and in our work life where you introduce a technology and on paper you have some perfectly good reason and you don't realize a degree to which that technology's presence completely shapes people's experience. When you switch from these to all you can eat lines and you had fast passes a way to try to get around that, it just created change the whole experience of the park to be this sort of ride maximization. We're going to win the day and get as many e-tickets as possible, rushing from thing to thing, always waiting in line, you know, it completely changed the experience just like technologically the presence of email or the presence of AI in the college environment where you introduce this tool and it completely changes the experience. So I thought that was a really good example. As a bonus he then, the same Nathan, didn't told me that Cory Doctreau who I really like his riding, obviously, had included a link to one of my things recently, I'll bring this on the screen here in his newsletter from August 12th called Model Collapses. The name of this newsletter, he had, you know, he puts links, Cory Doctreau is a very interesting thinker. We talked about him on the show, we should have on the show, Jesse. But I think somewhere in there he linked to one of mine, let's see here, it's in his things I hate and people I hate section. People who are making matters worse, I don't know how this works, all right, here we go. Yeah, there we go. On AI coding and it's discontent. So that was my newsletter article that actually made the rounds, Jesse. I did a newsletter article, look I'll click on it for those, bring it back up here on AI coding and it's discontent. If you don't subscribe to my newsletter, you should calnupert.com. Most weeks I send a dispatch from the fight for depth and destructive world. This was from August 10th, I shared a story of a listener who wrote in to say, hey, back in January, you know, back in January, this listener had bothered me. He said, I think you're being too skeptical about agents. Cloud code changed my life as a programmer, it's everything people said it's going to be. He wrote me again a month ago and said, I've had to completely change my relationship. To Cloud code, I crashed our major consumer website, I won't say what it is, but Jesse, you would have heard of it. Everyone would have heard of this. It's a major.com. I've crashed our website twice with code that was produced with Cloud code. I'm going to get fired if I do it again. I'm back to coding largely by hand. And I put that out and a lot of people came out of the woodwork and are like, and they're so to voice, yeah, we're kind of doing the same. If you're not just vibe coding things or make a minimal viable products, if you're trying to actually write production code, they're like, he's can't trust it and it's too hard to review. It's hard to review code you didn't write. And it takes forever. And now there's a lot of quiet retreat towards you use the cloud or whatever AI coding agent more selectively. Can you do this piece for me? I'm going to do this piece because you really have to understand how your code works. And I think that's like a really important indication. Look, I've been writing about the intersection of digital technology and knowledge work for a decade now. And this is what happens. It's hard. Like some things can come through and change everything and break things like email did. And we still haven't figured out how to use it right properly. It's just hard. It's hard to have a silver bullet tool that just works and everyone uses it. And it's really, you know, it's going to be hard just AI like anything else. It's a normal technology. It's a cool technology. But it's not magic. It's not how, 9,000. It's not the, you know, small wonder from 1980s sitcom television. It's wonky and they're trying to figure it out. And some stuff is awesome. There's a lot of hard edges and it's expensive and it's difficult and a lot of these companies will fail and some will figure it out and you'll have a mature industry. But exponentials don't last forever. And anyways, this goes back to my general AI stance, which is like cool, exciting technology. Let's cover it like cool and exciting technology, which means most people don't really need to be paying attention. Stop trying to terrify everyone and hype out our butt because we're online and bored or not realizing that these people we're talking to from these companies or these think tanks are dyed in the wool, effective, altruist, rationalists whose entire reason for meaning is this belief that superintelligence will deliver either utopia or they will be John Conner and stop the apocalypse and their whole meaning system comes out of this. And they leave their group houses in the Bay Area to go talk to New York reporters and are like, this technology is very scary and we don't realize they're from these weird sort of cultists like thinking and we're like, oh my god, is this must be like COVID or something and not realizing that like they've got spock ears on and they're in polyamorous group houses. We got to cut out the hype and just treat this like a cool but normal technology. We've had other incredibly exciting technologies and it didn't make everyone stressed out all the time. So I don't want to preach about that, but John Conner, that's the terminator, right? Terminator. Right. John Conner and Neo, that's what all these rationalists, like basically they recruited a bunch of high school students and I mean college students who are like, I want to be Neo from the Matrix. So we're the ones, I mean, look at you, Kowski's writing from this period, right? It's a lot of like, you'll call himself a hero, you'll write things like, I think I'm the only one who can save the world, right? This is the way they were thinking and talking before there are LLM's. They just love this idea of it's it's interest transhumanist futurists, you know, meeting dystopia type ideas and it's spilling over into our discourse. The issue is when I push back against that hype, people that interpret it as, you're saying AI doesn't work and therefore like, hey, hey, here's something AI, a cool new product. So you're wrong, right? This is not about this AI work or not. It's about do we have to have this layer from the AI is a force against me, meaning crowd of anxiety, hype, exaggeration amplified by online people who are just bored and live in a world of made up facts anyway, live in a world of engagement, like the substrate in which this online world exists where all these people are put up these YouTube videos and tweeting about AI like, this is it guys, this is the real deal with like alarm bells or whatever. They live in a world in which the normative substrate for the determination of what is valuable or not is engagement and that has been completely wired into their heads as a completely different normative value than truth or rationality and they're picking up on the AI is a force to give us meaning crowds type of nonsense and amplifying it everywhere. AI is not your god, AI is not going to solve all your problems, AI is not going to cause all your problems, it is a tool, it is technology, it'll be powerful, some places, other place will be frustrating, it's going to create some people that were annoyed at, we need to regulate it, we need to care about it, we don't need to obsess about it. There we go. There's my rant, Jesse. Good rant. There we go. All right, let's end like we like to do on Mondays by catching in with what I've been up to. I was just on another vacation, we always spend a week at the end of summer at the beach. I finished three books while there, Jesse, so I thought I'd go over them now, so these all are August books. I finished Tom Griffith's book, The Laws of Thought. This came out earlier this year, maybe in the fall, I think it was earlier this in the winter and it's a popular science book about the various mathematical models we've developed over history for trying to understand thought, going back from like the sort of logic, trying to use propositional logic up through neural thinking, et cetera, I thought it was very good. I read Tim Berners-Lee's memoir. This is for everyone. Tim Berners-Lee invents the World Wide Web. I used to run into him all the time, Jesse, because when I was at MIT, I was on the sixth floor of the SATA center, which is where the theory group was. And Tim worked on the fifth floor, which is where the World Wide Web Consortium was and we would write the elevator sometimes together. That was a good book. History of the Web is interesting. I also read into the deep memoir, Bob Ballard, who he co-authored with Christopher Drew, who's a sort of nautical science writer, New York Times writer. He wrote Blind Man's Bough, which is a good book about submarine espionage. My brother's was a submarine or so. Yeah, you've mentioned that before, because there was like some references to another book that weren't one of them made sense. That nuclear book. There may no sense. There may no sense. Alarm bells. They don't have alarm bells on submarines, a nuclear submarines, the feats of purpose. So Bob Ballard is, you know, Bob Ballard, a guy who discovered the Titanic. Yeah. It was good. He's based out of Woods Hole, I think, for a while. For a long time. Yeah, Falmouth. Yeah. And then he went to, he's in Mystic and other places. He bounces around. Anyway, interesting guy, his memoir, serious scientist, like his whole thing, he's upset about, upset about. But he's like, he will forever be known as the guy who discovered the Titanic. But he also had like the first direct up, very quick summary. This big claim to fame in like the 70s, I guess, was like, hey, we should use these new deep sea submersibles. [BLANK_AUDIO] actually do observational science on the ocean floor. We should go down there and look at things, there's a lot to discover versus just like dragging sonar over and stuff like that. And they went down there and first of all went down there and got like the first visual confirmation of plate tectonics, which was a kind of new theory then. He's like, "Oh, we can actually see at the bottom of the ocean where this plate ends and this plate begins and there's all this lava in between." Then they found the first hydrothermal vents, hot vents at the bottom of the ocean, which changed the way that we understood chemistry because there was, or geology, I guess, and then he found these black smoke or hydro, hydro vents that had a new type of organism never existed before that was like they would live off of the chemicals and chemosynthesis, a photosynthesis. So he ended up changing geology textbooks, chemistry textbooks and biology textbooks in like a 15-year period. So he did this like really big science and he found the Titanic in part because they needed a cover. He hasn't been able to talk about this to the last five years. He was working for the Navy. He was in the Navy and the reserves. He was working for the Navy on a mission to use the submersibles to check out the thresher, which is a nuclear submarine that sunk out off of the Azores to try to see, have the Soviets been there, like it's the containment vehicle, reactor vehicle intact, and they needed a cover story for going out there with all this equipment. And he said, "Well, what if on the way back we stop and look for the Titanic?" And then we can just tell people, "Oh, I'm out on a expedition to see the Titanic." And they were like, "Okay, that makes sense." And they did. They went to the thresher and they blow the blood on the way back to look for a Titanic and found it. But he couldn't talk about that until like more recently. So there you go. Interesting book. All right, I think that's enough for today. That's all we got for today. Thank you for listening. We'll be back next Monday with another one of these advice episodes. I know we've been kind of a little bit sparse on the AI reality checks. I'm sort of thinking through a more general format of that technology type of episode. So like there's a little bit delays now. But I have some pretty grand visions for a much more robust second technology themed episode for this feed. So bear with me. That's coming back soon. And until next time, as always, stay deep. Hey, if you've made it this far, you must be ready to join my fight for depth in a distracted world. Now, the best way to do this is to join over 125,000 people who receive my email newsletter each Monday. You can sign up at calnewport.com/ideas. And when you do, I will send you a free guide to my seven best ideas about cultivating a deep life. Sign up today at calnewport.com/ideas.