Your Brain Is Being Hijacked - How To Quit Social Media (For Real This Time) | Cal Newport

0:00 Most phone users, especially of a certain age, aren't particularly happy about how much time they are spending on their phone using social apps like Tik Tok or Instagram. They've heard people like me recommending for years that they should consider quitting those apps. But most people can't quite bring themselves to do it. And the question is why this is I have an explanation I want to give here. And then I want to deconstruct this explanation and give you a new way of thinking about improving your relationship with your device. All right. So, here's what I want to argue is going on. Part of the resistance or reluctance that people have to, let's say, taking popular apps off their phone is that they worry that it is an overreaction that perhaps they are

0:46 falling into a moral panic. Now, I want to play some audio here to better emphasize what I mean. Um, this comes from the spring of 2019. It was from an interview appearance I did on Brian Coppelman's podcast as part of the book tour for my book, Digital Minimalism. So, let's hear this tape for a second, Jesse, because I could have been a car minimalist and I could have said um, you know, that there are all these costs of having a car. You're not going to see the scenery. You know, we need nature and we need to see the nature. Um you're gonna you're you're risking in you know if you're have a slight inattention you could crash. Yeah. Um and so to me it is this this argument is also um the cars

1:33 are taking over. There's nothing you can do about it. Yeah. And we better instead learn how to use this stuff the car how to drive. Well well but so for the All right. So what Copelman was arguing there is a a classic moral panic argument about we hear about uh technology these days which is like look we often get worried about new things but these things are often not so bad and in the end we might as that we end up being okay with them so we might as well just get along with the process of learning how to use them. So he was saying you just as easily he was talking about my book digital minimalism. He's saying like look you say you're a digital minimalist. He was saying in that clip you could use that same argument to have been a car minimalist in the early 20th century and be like I don't want new cars have these problems.

2:18 We shouldn't have cars but cars were inevitable. Cars were coming and they ended up you know we need cars. Life would be much harder without them. We wouldn't have the you know American growth like we had it without them. And so that was like a waste of time. It was alarmist. This is a common push back that technocritics are familiar with. Now I think this type of reasoning, this push back fits particularly well when we talk about the trajectory of mass media that tools like Tik Tok and Instagram are at the current endpoint of. So if we think about mass media in particular, we see some examples of perhaps people being really worked up into a lather when new medias came along and then uh eventually relaxing about them. So consider comic books for example. This once terrified the fedora wearing pearl

3:03 clutching adults of the era who were convinced that they were corrupting youth. At a 1954 Senate subcommittee meeting, a leading anti-comic book advocate named Frederick Worthham testified as follows. It is my opinion without any reasonable doubt and without any reservation that comic books are an important contributing factor in many cases of juvenile delinquency. He went on in that same testimony to say, "Batman is promoting homoeroticism and Wonder Woman is promoting satto masochism." Jesse, I probably should have read more comic books. Yeah, I know what's going on in those things, but man, I missed out. Uh, television engendered similar concerns. Here's a quote from Wendel Barry. He had a a famous essay called Family Life, and it was included in his 1981 essay collection, The Gift of the

3:50 Good Land. And in it, Barry said the following. As soon as we see that the TV cord is a vacuum line piping life and meaning out of the household, we can unplug it. Right? We're okay with comic books today, it blew over. Uh we're okay with TVs today. We all we all still have them. We all still watch Netflix or Max. We're not nearly as worried as Barry once was. So the question is maybe we can just think about or why can't we just think about social media as just the next stop on this trajectory and in the future when we're all using VR to hang out with alien versions of Mark Zuckerberg u we'll look back at the social media age like that come on that was just technology that came and went and we're okay with it but is this right? So, I want to take

4:36 a finer look at this argument of moral panic applied specifically to mass media, applied specifically to whether or not social media is just another thing on this trajectory. To help us make sense of this, here's what I'm going to do. I'm going to return to an analogy that I introduced in an essay last spring that I was thinking about recently and realize, wait, this gives us some useful insight into this very specific question. So, the essay I want to draw from here came out last June. And I'll put on the screen here for people who are watching instead of just listening. Um I wrote an essay June 19th 2024 that was called on ultrarocessed content. And what I did in this essay is I drew a connection between the nutritional notion of ultrarocessed food and social media content which I was

5:21 going to call ultrarocessed content. I want to read you a few pieces from this because I think it's going to really help us figure out what's actually going on here with this type of information. All right. All right. So, for my essay, I said, "Ultrarocessed foods at their most damaging extreme are made by breaking down core stock ingredients such as corn or soy into their basic organic building blocks, then recombining these elements into hyper palatable combinations rich in salt, sugar, and fat soaked with unpronouncable chemical emulsifiers and preservatives. The problem with ultrarocessed foods is that they are engineered to hijack our desire mechanisms, making them literally irresistible. The result is that we consume way more calories than we need in arguably the least healthy form possible. Right? That's ultrarocessed food. Let's connect that to digital

6:08 information. Reading from the article again here. Something similar happens with social media content. Whereas the stock ingredients for ultrarocessed food are found in vast fields of cheap corn and soy, social media content draws on vast databases of user generated information, post, reactions, videos, quips, and memes. Recommendation algorithms then sift through this monumental collection of protocontent to find new hard to resist combinations that will appeal to users. A feedback loop soon develops in which the producers of the stock content adapts to what seems to better please the platform, simplifying and purifying their outputs to more efficiently feed the algorithm's global goal of hijacking the human desire mechanisms. In this way, the users of social media platforms simulate something like the food

6:53 scientist's ability to break down corn and reconstitute it into a hyper palatable edible food-like substance. What is a Tik Tok dance mashup if not a digital Dorito? So I think this analogy is useful because we treat in the world of nutrition ultrarocessed food differently than other types of food including some food that's not that healthy for us. And I want to argue if we follow this analogy to its conclusion we'll see that maybe we should treat ultrarocessed content similar than we do other types of content even content that's not necessarily super good for us. So in particular looking back in the world of nutrition there has been a a a poorest to be sure but important dividing line that's been made between processed food more broadly and ultrarocessed food more

7:43 specifically. So processed food is you know any type of food that has gone through some sort of notable processing from its natural state. It includes some stuff that's like barely processed like roast nuts and at the other extreme is like ultrarocessed. A lot of classic processed foods are things like, you know, um, bread or fresh baked uh, cookies or something like this, right? Uh, we learned in the 20th century, hey, it's kind of a problem if all you eat is processed food all the time, like you need to eat whole foods and natural foods. But we didn't argue to people, cut every bit of processed food all together out of your diet. That is extreme and basically unaccomplishable for almost everyone. on and the people who do pull it off is an incredibly restrictive uh commitment to only having sort of like

8:30 minimally processed fresh foods. That should be a lot of your diet, but we didn't tell people try to get rid of all processed food. It would also cut out things that you know people actually enjoyed in life that were part of cultural traditions. If you really cut out all possible processed foods, you would never have your, you know, bubbies cougall or Italian grandmother's pasta. Um so it feels like this would be too far. But with ultrarocessed food, nutrition experts like that. Come on, you don't just it's easier just to cut it out alltogether. Like this is a very common dichotomy. Don't eat too much of the processed foods, you know, like enjoy it when you do, but don't make it the core of your diet. But when it comes to Doritos and Oreos, this isn't some family tradition. It's not somehow like necessary. It's not hard to get rid of them. They're super unhealthy, much more so compared than like having pasta because of all the weird chemicals. And they're way

9:15 more hyper palatable than, you know, the the the kougal that you're having or whatever. Why not just avoid those altogether? And this is a very common dividing line we make. There's unhealthy food that we eat with moderation and that's okay and it has deep history. And then there's like this new franken foods and it's easier to say no nutritionist is like, well, you know, eat some Oreos but not too much. Like it's better just not to eat that if you can avoid it. And it's not that hard to avoid it. You cannot have super junk food in your house. And it's not really going to that's not hard. Hey, it's Cal. I wanted to interrupt briefly to say that if you're enjoying this video, then you need to check out my new book, Slow Productivity: The Lost Art of Accomplishment Without Burnout. This is like the Bible for most of the ideas we talk about here in these videos. You can

10:03 get a free excerpt at calupport.com/slow. I know you're going to like it. Check it out. Now, let's get back to the video. It's not going to be hard to find calories necessarily. Um, it's not going to be cutting you off from from things that are very important in your life. So, this is one way to think about what's going on in the world of media. The social media content is the ultrarocessed food. We look at things like, I don't know, watching streaming services on TV or digital diversions like those New York Times games apps like Spelling Bee or Wordle. That's like the pasta or the Google. Don't spend all your time doing that. If you're in front of the TV binging Netflix eight hours a day, that's bad. Also, don't eat canned soup all day. But like, you can have some of that in your life. Not a big deal. But the

10:49 ultrarocessed stuff where you've broken down content like Tik Tok does into its most constituent, smallest atomic elements and then algorithmically reconstituted into something that resembles content, but is like no other content that anyone has ever constructed from scratch. And it is hyper palatable and you can't put it away. And the actual information being conveyed in the content makes you feel existentially empty or scared or outraged or upset. This is the psychological equivalent of the physical unwwellness you get from eating ultrarocessed actual nutritious foods as well. Right? Why don't we just avoid those altogether? This now becomes, I believe, an intellectually coherent argument. It doesn't mean that it's necessarily the answer. And we know people in the world

11:36 of nutrition who like can have a couple Oreos and be like, I'm done. Most people, you open up a bag of Oreos, they're going to crush four leaves. Couple some people don't, but most people, it's easier to say, just don't have Oreos in the house. So, I bring this up just to make the argument that this analogy of ultrarocessed food and how it compares to processed food. It gives us at the very least an intellectually coherent and internally consistent argument for why you might say to someone, "Yeah, it's true. we might have overreacted about television and comic books and yet you probably still shouldn't use Tik Tok because sometimes in a in a particular space as technology advances its ability to do harm gets worse and as we advance along there it's the more

12:21 advanced versions we're like okay that's too far so I do think there's an intellectually consistent or coherent in internally consistent argument for social media being an exceptional danger in the world of mass media and just something I was thinking about is this ultrarocessed food analogy helps make that clear. So there you go. I don't know why I was thinking about that, Jesse, but I it came up in a conversation. I was like, that's helpful. It's pretty incredible when most grocery stores you go into and all the processed foods like in the middle. It's only on the outskirts where you can get like the healthy foods and stuff. I feel like that's that's the internet these days. Yeah. like on the outskirts is oh here's an interesting email newsletter and podcast and in the middle is you know like a white supremacist Tik Tok dance and it's like put it in my

13:07 veins put in my veins so anyways I think that's that's useful to me at least so hopefully other people find that useful as well I like that term digital dorito that's really what these things are by the way I wrote this in more length in my newsletter so again if you're not subscribed calort.com and you can find it on there and subscribe to get stuff like this in the future. All right, we got some good questions coming up, but first want to mention one of our sponsors. Let's talk about our friends at Cozy Earth, one of my favorite companies. So, Cozy Earth, they have uh the core products you might know them for is their sheets, which their bamboo sheets, which I've really been pushing. We have multiple pairs in our house, the most comfortable sheets that we own. We bring them with us when we travel for more than a week. But they have other

13:52 things you can buy as well. I just got a new and we I thought I owned everything Cozy Earth sells, Jesse, but I just got a new Cozy Earth product, which is the all day tea. So, it's a t-shirt made out of their really comfortable style fabric. An everyday essential with natural breathable fabric. This tea ensures that you stay cool and comfortable all day long. This fabric has a way of being cooling. I wore it. I worked out in it. I wore it around the house over the weekend. I can tell you, super comfortable. I also have the studio pant collection. I don't think I've ever had such a soft pair of pants. It's like soft and stretchy. Cozy Earth knows how to do comfort. So, here's what you need. You need the sheets. I've just I'm telling you this. Buy the sheets. Father's Day is coming up. Why don't you think about the all day tea or the

14:39 studio pant collection as a potential Father's Day gift? I can tell you, I love the cozier stuff and the father in your life will as well. Keep in mind that there is a lifetime warranty on all apparel items. So, hey, if you don't like it, you send it back, but you will like it. Uh, you will love it. I've been loving my Cozy Earth stuff. Uh, it's a great way to create a sanctuary in your own home. Cozy Earth makes comfort at last, and this Father's Day, he deserves it. Go to cozyearth.com and use code deep for up to 40% off all men's apparel. That's cozyearth.com. Use code deep for the dads who work hard during their 9 to5 and deserve the best during their five to nine. There we go. I like reading things about how dads are

15:25 awesome. I'm just like I'm awesome. Send me gifts. Also want to talk about our friends at Harry's. Staying well groomed requires the finest and most advanced tools. Upgrade your routine with Harry's. I love my Harry's razor. uh good weighted thick handle. It looks really nice. The blades are really good. Uh they shave quite smoothly. It's the most advanced shaving product and you can get it at an unbeatable price. That's where Harry's comes in. $10 trial set. You can now get that for $6. Special deal at harrys.com/deep. That's an exclusive offer just for our listeners. These blades are German engineered. The Germans know how to make razor blades. They have customizable delivery options

16:11 for scheduled refills as low as $2. This is cheap. You're not spending all the money you need to to get the other brands at the drugstore. You also don't have to go through the 17 layers of security to get to the razors like you do these days at the drugstore. Just set up a refill plan. Have them just show up as you need them. We talk about autopilot scheduling. This is an example. Don't think another minute about getting razors. Just set up your subscription with Harry's. You will not regret it. um use their their uh their shaving their shaving cream, their body wash. They have all sorts of other stuff as well, deodorant. Look, it's all there. Get it shipped to you. Get automatic. Save some time. Get products you like. So, take your grooming out of the stone age. Get Harry's. Normally, the trial set is $10, but right now you

16:57 can get it for just $6 at harrys.com/deep. That's our exclusive link, harrys.com/deep for a $6 trial set. All right, Jesse, let's do some questions. First question is from Allan. I'm in a new role that demands deeper theoretical understanding. How can I improve my ability to process and critically critically engage with technical literature so I can become a better, more analytical thinker? I've been thinking a lot recently about thinking and so this is a timely question probably because I'm on a chapter of my my new book, The Deep Life, that's titled Reclaiming Your Mind or Reclaim Your Brain. So, I've been thinking about this stuff a lot. All right. There's an immediate solution and a longer term solution. And I want you to think about both. The immediate solution involves what you do right now when you're in a

17:44 session where you're trying to learn something hard or you've learned something hard and you want you want to apply it in a complicated way. Don't context shift. This right away is like taking a reduced dose of the Limitless pill from that Bradley Cooper movie. Is that Bradley Cooper? Limitless. Let's just assume it is. I'm gonna assume it is because he's a Georgetown grad, so I like to say that. U when you're not context switching, you feel much smarter. When you're context switching, meaning that you're trying to mainly focus on this one thing, but you're like quick checking your inbox and quick checking your phone and well, maybe I need to make this same game parlay for the upcoming, you know, Giants Nationals game or whatever it is you're doing. It reduces your IQ points non-trivially. So, if you actually when you do something that requires hard thinking, lock in without context shifting at all,

18:30 you are going to feel much smarter. So that's that's an immediate win. Do that right away. Longer term solution is to get yourself better at doing analytical thinking. A few things here matter. One, you need more comfort with boredom. There's not a lot of novel stimuli when you're sitting down and trying to work on within your mind's eye just a single problem. That's boring in the technical sense of lacking novel stimuli. Get used to that. Do at least a few things every day without a distraction, without a phone in your hand or an earbud in your ear, just to get used to being alone with your own thoughts and the world around you, you will acclimatize to it. If you acclimatize to it in these low stakes situation, when you get to the higher stake situation of this thing is really hard, your mind isn't going to rebel about we're bored. It's used to that. It

19:16 knows it's not really a threat. You can also train your brain to better process and make sense of uh information. to do. So, start with what I call the 2010 rule. 20 pages of reading a day, 10 minutes of being alone with your own thoughts a day, every day. Reading is fantastic for deepening analytical circuits. It's something I'm I'm really looking into now for the for the book I'm writing. But to give you the super short version of it, reading is unique. It's different than speech. It's different than vision because it's something for which humans do not come pre-equipped with a neural circuit for. We have to train ourselves from scratch how to read. We basically hijack other circuits that are in our brain for other uh utilities that existed deep into our evolutionary past. We hijack those and we rewire them to do this really novel

20:03 thing where we we have to build these reading circuits. And when we build these new circuits, they make us better able to process information. The best thinkers are great readers. Reading is as close as we have to cognitive calisthenics. So start reading 20 pages a day. At first, what matters is that you're doing it. So, I don't care what it is. Make it something that you're most likely to come back to. So, the most fun possible, interesting, easy possible thing you can. And then have 10 minutes of being alone with your own thoughts. This gets you not just used to boredom, but it gets you used to grabbing a current out of the turbulent streams within your mind, sticking with it, and trying to actually follow it to its source. That is something you have to practice as well. When you're reading,

20:48 you're kind of like actively focused on what's happening here and processing it. But when it comes time to do analytical thinking, you have to put those circuits to work on original information. So you have to be comfortable applying those circuits and being within your own thoughts and working filtering noise, focusing on what matters. So 20 m 20 pages, 10 minutes of time alone with your own thoughts, preferably walking. There's a reason why Aristotle called this philosophical school the parapotetic school. They moved. It helps you think. All right. Once you're comfortable with that, add 10 more pages and five more minutes. Now it becomes the 30-15 rule. I'm reading 30 pages a day, 15 minutes a day of walking and just thinking and trying to hold and make sense of thoughts. Get comfortable with that. Another 10 pages, another five minutes. If you can get to

21:34 something like 50 or so pages a day, 50 to 60 pages a day, and 20 to 30 minutes of reflective thinking, you're walking and thinking, holding your mind's eye, now it's like you're doing CrossFit. This is CrossFit for your mind. you're going to be in better shape than most people. Jesse, I'm looking at you because I know you do CrossFit. Jesse explains his CrossFit workouts to me. And here's like my simulation of Jesse explaining his his CrossFit workout. He's like, "Well, they came in and they said, "Okay, what we're going to do today is 97 sets. 97 sets of clean and jerks followed by 4,000 meters on the rower. And then we're gonna do 19,000 box jumps. Is that about right? Yeah,

22:21 more or less. That's about right. Yeah. Uh so what can I say? This is CrossFit for your brain. If you can get to roughly that like 60 pages a day, 20 to 30 minutes reflection on most days. Now you're ready to do some damage with your brain. So you can actually train that brain to be smarter. And I don't that's what I'd recommend. You know who doesn't need to do CrossFit? Cal Network. You you know he does it. When Cal Network walks in the CrossFit gyms, the they shut down the boxes. They shut him down. They're like, "It's not worth it. It's not worth it." And they all go to Crispy Cream because they're so intimidated by his muscles. All right, who do we got next? Next up is Arjun. I'm about to start my first internship doing software engineering work. I'm

23:07 still in college and this job is remote. How can I stay motivated, productive, and balanced during my solo work weeks? Arjen, you have to time block plan. You have to. If there's no actual physical separation between your work and no work, if there's no actual in-person visible supervision, that's going to play against more classic motivational circuits. If the boss is here, I want to do my work. I want to show them what I'm doing. It is up to you to add structure. And you have to have structure. Time block plan. When am I starting work? What blocks am I doing work? When is my breaks? When am I doing email? When am I finishing work? Shutdown routine. Then I move on to the rest of the day. Do not allow your time outside of work to blend together with your work and just say like, "Look, I'll get stuff done at some point today." Time block plan. I would suggest actually trying to keep those actual working hours tight and super

23:53 intense and then when you're done, be done, but you got to structure when you're remote work. Structuring your time is more important than ever. So, uh, get some time block planning going. I once, by the way, Jesse took a look at Cal Network's time block planner. It's a single index card. says crush it on it. That's his time block plan. All right, who do we got next? Next up is Daniel. I'm curious about your thoughts on how much time to spend developing secondary skills like improving typing speed or customizing tools that support our main priorities but can easily become distracting pseudo productivity and how to set boundaries so that these activities don't take over real skill development. I mean, I would downplay those secondary skills. Like

24:38 when you learn them, you learn them because it's clear if I got better at this, it's going to help exactly that. But really make your main focus in a demonstrable way, unamiguously demonstrable way, doing things that are rare and valuable. You got to get wired for that because what a lot of people do with this secondary skill pursuit is they're trying to actually avoid being evaluated. They want to avoid having to at some point just actually put everything on the line, say, "Here's my thing. How good is it?" And they they give you your ranking. And if you're like, "No, I'm working. I have a plan. I'm working on this skill. I built a custom productivity tool over here. Um, I'm going to try to learn how to do this like secondary thing." It allows you to seem very like busy and innovative. That allows you to keep the hope of doing well in the future while avoiding the stress of actually just putting out

25:24 there and seeing how you do. So, I want you to get obsessed first about I'm competing to do good stuff. How am I doing? How much money am I getting from clients? Am I getting the promotion? how many dollars am I bringing into the company? If you have a a a good laser-like focus on producing unambiguous value and not being afraid of falling short and celebrating when you actually get ahead of where you think, then yeah, throw in secondary skills as you need and they're not going to be a diversion. They're going to be very utilitarian. Like, man, I'm really doing well here. Ah, if I could type faster, I guess that would really help this. Now, I have a reason to learn it. So I just what I really want you to do here is not allow the secondary skills to become a diversion from actually just becoming so good you can't be ignored. So make that your obsession first. Secondary skills should remain

26:11 second. All right. What do we got next? All right. Next up is from Brian. How does Tik Tok seem to know so much about me and tailor contact so accurately to my interest and behavior? Uh, that's a good question because I think it connects to our deep dive where we talked about content like Tik Tok being ultrarocessed that unlike other types of content like television, streaming or games on your phone, it has a way of being hyper palatable and it has a way of pushing your buttons in ways that's really psychologically unhealthy. So, I I think it's worth looking into because I actually think I know something about this. this. I mean, I did some research on this back when there was a internal document that got leaked and this was sort of making its rounds and I read a bunch of accounts of this document. I had like a a generic sense of the techniques used by a tech talk

26:57 algorithm. It's probably worth briefly going over because I think people have this wrong and when you have it wrong, it changes the way you think about this world and what to do about the harms of this world. I looked up a quote here. I think this sort of [Music] establishes the way that we kind of get this wrong. So here's a here's a quote about Tik Tok that comes from the founder of an organization called Algo Transparency which I think kind of gives the wrong idea. But let me read this and then I'll give you the right idea. So this quote says, "Each video a kid watches, Tik Tok gains a piece of information on him. In a few hours, the algorithm can detect his musical taste, his physical attraction, if he's depressed, if he might be into drugs, and many many other sensitive information. there's a high risk that some of this information will be used against him. It could potentially be

27:43 used to microtarget him or make him more addicted to the platform. Technically, all of that is true, but I think the the way that we talk about algorithms here gives us a an incomplete understanding of how these things actually work. I think when we see those type of description of algorithms, what we really imagine is sort of like someone making a list of descriptive information about a user. Oh, we've learned that, we learned that, he's into this, not into that. We imagine an algorithm sort of like looking at these semantically rich descriptions of things you're into or not into and and deciding according to some sort of complicated rules that someone programmed in. All right. Well, why don't we then take this and mix it with that and we have some good stuff of that over here and and we're trying to emphasize this type of content. Oh, and I I've analyzed I think this type of content is is it feels like people like

28:29 it and so it's like a network scheduler. Let's show him more of that. Like so I think that that image of an algorithm, it's like a network scheduler looking at a board full of pieces of information on index cards and making a decision like well what are we going to schedule next? I think this is really going to catch his attention. We have that announce that that idea which is sort of personified. We imagine like there's a human involved and it gives us the sense that we can turn these various knobs and say things like, "Hey, adjust your algorithm to not recommend stuff that is going to be so addictive or adjust your algorithm to like turn down the uh the the outrage meter from like a 10 to a nine." When we think about the algorithm as like a person with all this information making scheduling decisions, we think we have all this control over it. That's not really how this works.

29:15 I'm going to try here to draw a picture for those who are watching instead of just listening. And God help us all for me doing this. But I want to give you a better sense of how the general class of algorithms that we think powers Tik Tok actually works. So you can imagine what we do is that every time uh you watch a video, there's a piece of information that's gathered, but what's gathered is a description of the video. This description though is going to be a bunch of numbers. So, there's going to be, you could think of it like a a lot of different categories you could use to apply to a video. You know, uh, cats could be a category. If has a lot of cats, that number is high or low or something like this, right? And so, you have different numbers in each of these categories sort of gives you a list of numbers that describes the video. So if we had for

30:01 example just two numbers, we could imagine having a flat surface like a two-dimensional plane where you could put a each spot in here like if we put a spot right here is associated with uh two numbers like with that number and with that number, right? So you can imagine what we do is every time you watch a video, we can put a little dot on here that represents its description. And we can make that dot bigger or smaller depending, and this seems to be what's most crucial for Tik Tok, how long you watch it, like what percentage of the video you actually watched. So like maybe you watched a video over here real briefly, small dot over here though, there's a video you watched all the way through. where there's like a big dot over there and maybe there's a

30:48 medium dot and you kind of sample some videos over here, right? And over time, we're kind of just giving you videos to watch and we see how long are you watching each of them. So now we have a bunch of these numbers. We have this plot. Technically, it's going to be a sort of weighted multi-dimensional scatter plot. And now when we want to show you a new video, we being Tik Tok, the algorithm is super mathematical. It says, "Okay, I'm just going to randomly choose a video in the spot of possible videos to show you, but I'm not going to select from every area equally likely. The more sort of green real estate there is in a certain place, like the more I'm going to be pulled in that direction." So,

31:34 like I might select something over here, but I'm way more likely to select something over here because there's so many, not only just so many other dots, but so many big dots. There's a lot of watch time over in this space of the possible videos. And what we're implicitly doing here is kind of creating a region. These are all going to be probably roughly similarly types of videos. Now, again, my algorithm is doing something very simple. It's just randomly choose a thing, but wait your choice. you know, the more stuff there is nearby it, more mass you want to put on it. And there's like a lot of probabistic ways you could do that. So, I'm going to start selecting more videos near stuff you already like. But like occasionally, you know, you roll a dice enough time. Sometimes you're going to roll six three times in a row. Occasionally, I'll pull some stuff over here as well. And maybe what I'll do,

32:20 and we think that the Tik Tok algorithm does this on purpose, it will on a semi-regular basis say forget uh forget the weights and let me just like purely draw randomly just to make sure there's not something we haven't tried yet that you really like. And then like maybe over here there is like a whole bunch of Cal Network memes and immediately these start getting watched. Well, and now it's like you're getting a lot of Cal Network memes and then like a lot of the time the stuff over here and occasionally we'll sample other types of things. That's what's happening underneath the covers. It's not just two numbers though. It'll be like a hundred numbers or a thousand numbers. So this is each of these spots is in a thousand dimensional space. Um so the math is different but it's the same

33:05 idea. What's key about this is the algorithm is stupid. the algorithm. There is no complicated code in there that humans wrote that's like how do I balance this versus that. Uh it doesn't label things necessarily in the way you think like is this outrageous or not or I mean it it does categorization to try to get rid of stuff for safety purposes but like it really just mainly treats stuff like these numbers and it's just blindly doing this simple repetitive loop. Pick another video probabilistically from the space of videos. Weight your choice towards stuff that's already been watched. So you're more likely to choose something near a bunch of other stuff that's been watched a while and other stuff on a regular basis. They'll randomly just choose something completely different to make sure that we're exploring other types of videos they might like. This simple

33:52 loop run tight enough where the only input it's getting from you is how long did you watch each video and how do we automatically categorize the video with some sort of neural net. that is enough to like within 30 or 40 minutes be eerily accurate in showing you stuff that you really like without there ever having to be anywhere like a like a an English language list of like topics and here's what this person likes and complicated algorithms. So it's a very simple weighted random sampling algorithm with a little bit of exploration for exploitation types of uh logic. This is called multi-arm bandit type stochcastic optimization. You don't have to worry about that but like occasionally sampling other stuff. It's a very simple algorithm run again and again and again that has an eerily effective complex effect. So anyways I

34:39 think that's interesting to know because it it tells us that you can't control these things as much as you think. You can filter like you can say hey before we even recommend something we can like filter is just like really inappropriate content and just like let's not recommend that at all. Uh you could like manually put shadow bands on things like okay anything with this category I'm going to type into my system like minimize its weight so we're much less likely to show other stuff like that you can do that but there is for the most part you don't have knobs to turn right you just say sample stuff near stuff they like Tik Tok works really well you could keep that on you can turn it off and that's it like we do not have the control over these algorithms that people think so anyways I think that's an important insight that I want to give there is a lot of these sort of

35:25 recommendation algorithms, the math might be complicated, but the algorithmic logic is not nearly as adjustable as you might think. A simple routine done again and again with more and more data can work eerily well, and it's pretty hard to have any sort of nuance control over it. So, there you go. I still have Tik Tok on my phone, Jesse, from that New Yorker article I wrote. Have you been on it? No, I have no in It's so weird. Here, look. I'm loading it up right now. It's such a weird app. if you're not native to this, like if you come to it just for like an article like I did. Let's see if this works. All right, I'm loading up TikTok on my phone. I haven't been on here since January. Um, don't allow. There's a woman singing. They're like taking a shirtless guy is taking

36:13 blankets off of her and now she's in like a fancy dress and there's confetti and the guy is blowing air on her. Jesse, what are we doing here? When you see your wife and three daughters see you OTW home and the kids are giving the middle finger. What are we do? Is this civilization? Turn. Turn long videos. Oh my god. Is this video appropriate for Tik Tok? Oh, they're getting data. They're getting data. They're forcing me to. All right, one more. When your teacher used to call home, someone walks into a room. Oh, it's a skit about like a guy putting on lots of pants

37:00 because his dad is going to hit him with a belt for being in trouble at school. It's a comedy skit. All right, come on. Anyways, that's Tik Tok. All right. I don't understand that, Jesse. But if I watch that long enough, see, that's random because I don't use it. So, that's just randomly sampling from like all videos. But, I'm telling you, if I watch this for like I don't know 20 minutes. It's all going to I don't know what it would be. It be like a lot of like Washington Nationals content, I guess. I don't know what else would be on there. It' be like Washington Nationals content and Cal Network memes, I guess. I don't know. But that's how that works. These algorithms just like it's like a virus. They're simple, but you let that process run long enough and fast enough, it can cause like complicated, terrible effects. Mhm. So, there we

37:46 go. Tik Tok. All right. What do we got next? All right. Jean's next. What are the best strategies for workflow when using AI agents? And we have an article that you can react to and we can show the audience. Oh, okay. You sent an article. Yep. All right. All right, I'll pull this on the screen for those who are watching. All right, the article um waiting for agents shouldn't kill your momentum. This post breaks down why no agents and too many agents both stall progress and shares a sweet spot workflow to keep you locked in flow. All right, so the first heading here says waiting kills flow. Now looking closer at this animation, I'll make this bigger here for those who are watching. What they are showing here is an AI agent working in what is at the moment the

38:33 only context that I know of where something like agentic AI code is actually not just useful but being somewhat largely applied which is in computer programming. So the AI agent that's being shown on the screen here is an AI agent that is helping you write software. So this is I think right now the perhaps the only major application of um agentic AI that I know of and again agentic AI is where you have control software surrounding the use of something like a language model. So it can query the language model for an idea or a plan and then the control software can look at the answer and then take actions based on the answer. So when it comes to computer programming, here's another example here for those who are watching. When it comes to building like

39:20 a semicomplicated vibe coding style computer program, you can have these agentic code like uh cursor windsurf or rue where what you're doing is it'll like first have the language model give you like a bunch of steps you need to do to build a software product and then it'll be like okay do you want us to do step one you say yes and it's like great and then the the software will then query the language model to get the code for step one and then it'll run it and be like oh there's errors. Do you want us to like try to fix it? You're like, "Yeah." And you kind of work on step one and the agent sort of makes queries on the language model on your behalf until that step has working code. And then it's like, do you want to go to step two? And you go to step two and it allows you to build more complicated software. Then if you just ask that the language model, give me code for this. Like if you say, "Give me code, it has

40:06 to be a relatively small thing, it it solves, but if you do this multi-step thing, Agentic AI lets you do more things." So the the more complicated project. So the the uh argument here is that and I've seen this I' I've sat next to people they've showed me them using these systems. There's a lot of waiting, right? Because these language models aren't fast to query because you know you might have 500 billion parameters being executed by multiple GPUs. There's cues. It's slow. So all these animations I showed you, a lot of it is like thinking line by line. You see things happening. So this whole article is like how do you not get too distracted while you're waiting for your agentic software when you're building software to like finish thinking and I think they are fans of my work because

40:53 they talk about if you're jumping over to social media like your phone man you're going to get so context shifted and overloaded that you're going to your brain's going to burn out. So then they argue here like what you should do and here's their solution is you should have something like an obsidianpowered to-do list next to your coding so that when this starts thinking you can go over here and work on something productive and then when it's done go back over here and then you're never really leaving the context of the project but you're also not just sitting there bored. So I don't know there's something I like and something I dislike about this. I like their general application of hey be wary of context shifts. We have this idea of deep breaks where like if you want to take a mini

41:39 break from what you're doing there's ways to do it so that you don't fry your mental circuit. So working on the to-do list for the larger project keeps you within like the general cognitive context. But the sort of less exciting side of this example is like this is kind of it for agentic AI right now. So I'm I'm I'm still we talked about this last week at the end of the show. I talked about this in my recent email newsletter post AI and work, which you can find at calupport.com. We're not sure how many applications we're going to have for agentic AI. Open AI says everything is going to be done by agents and we're never going to have to have jobs again and it's going to somehow generate like $20 billion like next year. But there's a lot of technical obstacles to applying this agentic approach to other types of work that we do that's not just like

42:25 building somewhat subtly constructed smallsiz software applications. So there's a bunch of stuff going on here. I think it's an interesting application. So yeah, if you're vi if you're coding with aentic AI, this is actually like a new problem we haven't had before. What do you do when you're waiting for the actually we have had this before. It's compile time. I take that back. This is a common problem for computer programmers is computers didn't used to be super powerful. So every time you change your code to rerun it, you had to compile and rebuild the project and it takes a while to test to see if it works. And so I actually, you know what, I take it back. Programmers have been talking about this forever. What do you do while you're waiting to compile? I think we've had questions like this before on the show, so there's nothing new here, but I think it's a good insight, at least from an AI perspective. This is like one of the places, if not the only place, where

43:10 Agentic AI is kind of doing interesting things. So, it'll be interesting to see if we get other places going forward. We don't yet have our podcast producer, Agentic AI. I think it would take this podcast in weird places. It'd be a lot of like Cal Network and Jesse Skeleton weird memes like cutting back and forth going to CrossFit gyms and uh it would be a massive success. Not there yet, though. All right, looks like we have a case study this week. It's where people send in their examples of using the type of advice I talk about this show in their own life so we can see what it looks like in the real world. If you have a case study to share, send it to jesseelport.com. All right, today's case study comes from Adam. Adam says, "While

43:57 walking our manufacturing floor the other day, I struck up a conversation with one of our maintenance technicians. I noticed he had a small, well-worn notepad he kept referring to. Out of curiosity, I asked him what he was writing down. He flipped through the pages and casually said, "I plan out my year hour by hour." I asked what he meant by that and he showed me how he divided each page into days with lines for each hour of his shift. Each hour had a specific task or category of responsibility tied to it. Things like preventative maintenance, troubleshooting, parts, ordering, and time for training or documenting issues. He explained that without this system, he'd never be able to keep up with all the side projects he and his managers were working on on top of his normal day-to-day tasks. What struck me was his clarity. If I don't write it all out like this, I'll get pulled into too many

44:43 directions, never get ahead. He said he's even has a similar version for his personal life. I told him that what he was doing had a name, time blocking, and that people like you talk about this method in books and podcast. He proceeded to give me a severe concussion by beating me around the head with a socket wrench. All right, I added that last part, but I was just imagining lecturing a maintenance technician about uh excuse me, but there's a podcast in which this technique has been thoroughly discussed. You're going to get a steeltoed work boot to the backside, right? Let me go back to what he actually said. He said he'd never heard the term before. It was just the only

45:28 thing that kept him sane. This conversation was a grounding reminder for me as someone who's constantly tempted to overoptimize bra productivity system. It was refreshing to see someone use a simple analog approach to stay focused and intentional. No apps, no elaborate dashboards, just pen and paper. An approach rooted in necessity and clarity. I'll tell you what article this guy probably did not read. The article on how to make progress on your Obsidian open source to-do list while waiting for your agentic AI to make its inference times complete. the maintenance technician was probably not reading that article and he's the better for it. Um, I like that example. Look, it's simple. When you're at work, you have a certain amount of time. That's what you have to allocate to doing things. The whole game is allocating that to the right things that

46:14 solves the dual problem of you making progress on things that are sufficiently value to keep you with a lot of career capital while also not burning you out or making you upset. That's the whole game. That's it. How you do that, right? It's like whatever works for you. It depends on the job at whatever. But if you're not having some sort of system, this is how I win at that game. You're not winning at it. If you just rock and roll, here's my email. Here's my Slack. There's stuff coming in and out. There's all these meeting invites coming at me. I'm trying to optimize my Obsidian setup while I wait for the inference time and my agentic AI. Just doing all this stuff, but you've never you don't have a clear answer to the question. How do you choose what to do with the time you have so as to achieve these two goals, capital and sustainability? If

47:00 you do not have a specific answer to that, you are being pushed around by forces outside of your control. You're operating at a fraction of your capacity. You're probably exhausting your brain. So, that's a great example. I love the simplicity of it. In the end, that's what it comes down to. And it doesn't necessarily require all that much technology to actually get it done. All right. Do we have a call? We do. All right, let's hear it. Hi, Cal. My name is Im Felino and I've been a lecturer within computer science at the Swedish Higher Education Institute for the past eight plus years. I've used some of your approaches to deep work and time management to become great at teaching and was promoted to expert lecturer a couple of years ago. I've now

47:46 invested some of my career capital and will start doctoral studies in January, but with my lecturer Sally. To be able to obsess over quality, I will need to develop a new set of skills related to research. As an experienced researcher, what general and specific skills would you recommend that I hone in on? All right, it's a great question, but first, congratulations. You worked a career capital system. Well, now you're going for your doctorate, but at a salary and not at like a research assistant salary or or you know, you you've worked it well. You became an expert lecturer. You have a connection to this university. You can do a salaried program. It's probably going to open up slots for you when you're done. So, well played uh the way you've done this so far. All right.

48:31 When it comes to how to succeed in this doctoral program, you need expert guidance. So, you fortunately have this built in your adviser. Choose an adviser who's doing good work in their field and it's a field that you think is interesting and has value and then say I want to learn how to do this. Let me right off the bat let me write it. Let me be useful to you. What paper are you working on now? I want to learn about it and I want to be useful and I want to write papers with you right off the bat to learn how it happens. To learn how it happens. So this is what I did. So a little bit of a brief divergence here when I first arrived. So I arrived in the theory group. I was working in the theory of distributed systems group uh in the theory of computation group uh at MIT and technically I was sort of hired

49:20 to do systems work for this theory group but I wanted to learn how to do theory and I remember this the first thing I did when I got there is there was a a paper they were working on for a good conference a conference I ended up down the line publishing a ton of papers at and being in the steering committee for um but I said let how can I help on this paper here's what I know how to I just want to be involved in this paper because I want to learn how people write papers for this good conference. I don't want to write papers for less conferences. I don't want to go on my own and start writing stuff for like secondary conferences and trying to figure this out. I want to publish in the best places. I said, "Here's what I can do. I'm a kind of a systemsy guy. I can build like simulations and stuff like this. I could add a simulation sort of section here. I'll figure it out. It' be like a nice, you know, half page to your paper." Like, "All right, fine. You

50:05 can do that for And now I was in the door and I was involved in this paper and I was there and I saw every draft and I saw how it worked. And when the next paper came up which was like a followup I was like well here's a corner of the theory that I think I can do. It's not too hard. And so like I found a way to be useful on the highquality stuff a small handhold on a big swing project. And I made those handholds bigger and bigger until after a year or so I was like now I know what goes into writing these papers and I could really take off and now I wasn't uh dependent on waiting for other people to be writing a paper to work on and I could start my own and I published a lot of papers. So work with your adviser on the best possible papers. That is the skill you want to learn and the only way to learn it is to work on really good papers with people who know how to publish them. Don't do anything

50:51 else. All right. All right, so we got a final segment coming up where I will go over the books I read in May as well as looking at this truly brilliant piece of fan art that we're going to have to go through years of therapy, Jesse. I think after discussing this, it's just that strange and just that wonderful. But first, let's briefly hear from our sponsors. So guys, why don't you want to look a little bit younger? Maybe you want that muscular glow of say a cal network type. Maybe get a few more compliments on your skin or just feel more confident when you work in the mirror. This is what Calira Labs is here for. Their high performance skin care is designed specifically for men. Simple, effective, and backed by science. They

51:36 have products like the good, which is an award-winning serum packed with 27 active botanicals. I counted them. And 3.4 million antioxidant units per drop. I counted that as well. It took me a while. Probably should just looked at the literature. Their eye serum. I need this. It helps reduce the appearance of tired eyes, dark circles, and puffiness by using a breakthrough peptide complex and obviously adaptogens that support skin resilience and base layer, which is a nutrient-rich moisturizer influ infused with plant stem cells and snow mushroom as um extract. Here's the thing. Guys don't know that you're supposed to put stuff on your skin, but if you don't, and we talked about this last time, if you don't, by the time you hit your 40s, you're going to look like George Hamilton. That's the name, Jesse. I couldn't remember it last week, and roughly everybody wrote us to say George

52:24 Hamilton is the overt guy. It is true. You're like 35 years old, your skin looks great, you're like, I'm going to live forever. You turn 40, and if you haven't been putting moisturizer on your face, you suddenly look like you just got back from a 19th century around the world whailing voyage. old cracked skin and you're carving scrimshaw and you probably have a peg leg. So men, you have to put stuff on your face. Calera is made for men. This is good stuff. I particularly like the base layer. Keeps my skin moist. They have cutting edge formulas, rigorous R&D cycles. They are noncomodienic formulas. I almost get one day I'm going to get that word right. But that means it doesn't clog your pores, guys. You don't want that. Plastic free, crueltyfree. So here's the thing. Skincare doesn't have to be complicated, but it should be good.

53:10 Upgrade your routine with Calera Lab and see the difference for yourself. Go to calerab.com/deep and if you use the code deep at checkout, you'll get 20% off your first order. I also want to talk about our friends at Shopify. If you run a small business, you know, there's nothing small about it. As a business owner, I get it. My business is and has always been all-consuming since I started. Every day there is a new decision to make. And even the smallest decisions feel massive. The thing that will help you most of all is knowing that you have the right platforms that you need to be successful. And if you're selling things that is Shopify, true story. True story. Cal Network began uh selling online. Began selling online old socks

53:56 that he had worn. He's like, I will sell this for you online. That business today, Jesse, $97 million a year. That's the type of success story that you can have, but the logistics will get complicated. But not if you use Shopify. Shopify is the commerce platform behind millions of businesses around the world and 10% of all e-commerce in the US. You don't need a big team. Shopify will help you handle anything from website design to managing inventory to customer service to global shipping. Look, all the stuff you need to do to successfully sell things, Shopify is there for you. They can help you market. They have email list tools. You want to grow your list, your business in person, point of sales solutions right there that hook up to your online database that can hook up to your your web interface. Shopify is what you need if you are going to sell

54:44 thing. So turn those whatifs into uh I was going to say turn those whatifs into pause for sound effect and keep giving those big dreams the best shot with Shopify. Sign up for your $1 per month trial and start selling today at shopify.com/deep. Go to shopify.com/deep. Shopify.com/deep. All right, Jesse, let's move on to our final segment. All right, so before we get into the books that I read last month, let's load up for the people who are watching instead of just listening this artwork that was sent into us. Let's see. I have some text on this somewhere. Oh, here we go. All right, here's some artwork that was sent into us. I'm going to put it on the screen. Oh my, Jesse. This was commissioned

55:30 This was commissioned artwork. I don't know if I should be excited or a little bit afraid right now. Um, I'm trying to find It's in the beginning. Yeah, I know. I moved it. Uh, Bryant sent it. Here we go. So, here's this artwork. It's a for people who are watching. God, I don't know how to explain this, Jesse. I'm just going to have to go I'm just going to have to get right into it. It is a uh pencled like charcoal drawing of a bunch of men that includes naturally uh myself, Anthony Edwards of the Minnesota Timberwolves, Columbbo, Kirk Fernettes, the head coach of the Iowa Hawkeyes football team, my buddy Brent McKay from the art of manliness, and Keith Morrison of Dline. And there's a grill in front of us. And

56:18 he's in the middle. Brent's in the middle. Yeah, but I'm I'm I'm near the middle. Oh, yeah. You're right next to him. I feel good about this. So, anyways, I think it's a cool picture. Like, these So, Brent just said like, "Hey, these are all people." It's cool what he's doing. He's like, "Oh, these are all people for various reasons I think are interesting or cool and it'd be fun to have an artwork." And imagine what if they all had a barbecue. But the thing that kind of freaks me out is having the dine guy there, right? Because this is not exactly what you would hope to see. You're like at a barbecue with all of your uh older men friends and then the dine guy shows up. Like it's trouble. Like there's probably a corpse in that grill if I had to guess. So it sort of looks like me and Anthony Edwards are about to be arrested for like statutory rape or something like that. That that part makes me a little bit nervous having the deline guy

57:04 standing next to me. But I do here's what I do appreciate being next to like a an excellent athlete like Anthony Edwards. Okay. I think it's kind of cool. Uh took me a second but I get it. I get what he's going for there. Um, and I wish we had that for the HQ. I would definitely hang that up. It would be awesome. So, cool work. Good art. All right, let's talk about books. I try to read five uh I'm all over the place with my papers, by the way, today. This is why I need Agentic AI or something. I I try to I try to read five books a month, and I did so last month in May 2025. Here are the five books I read in order that I read them. The first was a book called Building: A Carpenters's Notes on Life and the Art of Good Work. This was

…