Today, I want to give you an argument about why you feel uneasy about your smartphone that you've likely never heard before. It is, however, I think an important argument to hear. Why? Because it draws from a source that far predates these modern digital technologies. I'm going to make you an argument about why you're uneasy about your smartphone that goes back to the foundational moral philosophy of Immanuel Kant.

We're talking about a philosopher from the 1700s who is arguably the most influential source of moral ideas since the Bible. And it turns out, if you read Kant correctly, he has a lot to say about decidedly modern inventions such as Twitter and TikTok. So I'm going to ask you to stick with me here.

We're going to get a little bit technical, but I'm going to walk you through. We're not going to get too technical. All of these ideas will be accessible, and we're going to come out on the other end of this exploration with a better understanding of the role technology plays in your life right now, why that makes you uneasy, and changes you can make.

So I think it's a cool thing to add to the argument. All right, so I'm going to be drawing today entirely from a single academic paper from 2021. I'll pull it up on the screen here. I have a preprint version here that I could access, which has a link to the official final version.

I'll put it on the screen here for people who are watching instead of just listening. The paper we're going to be drawing from is titled, Is There a Duty to Be a Digital Minimalist? This was published in the Journal of Applied Philosophy back in the summer of 2021. For those of you who are tracking at home, that's volume 38, number four.

The authors are Timothy Aylsworth and Clinton Castro. These are philosophers from Florida International University. All right, so I'm going to jump through this paper somewhat selectively to pull out what I think the important parts are. So I'm actually going to start by jumping ahead here a little bit. I'll sort of keep up with this on the screen, I suppose, for those who are watching.

But I'm going to read everything, so don't worry if you're just listening. Okay, so I'm going to jump ahead here. They're talking here about digital minimalism. Let me read from the paper. Authors like Cal Newport, who coined the term digital minimalism, argue that we would be better off if we restructured our relationships with technology on our own terms.

He understands digital minimalism as, quote, a philosophy of technology use in which you focus your online time on a small number of carefully selected and optimized activities that strongly support things you value, and then happily miss out on everything else. A philosophy of technology use is a personal philosophy that covers which digital tools we allow into our lives, for what reasons, and under what constraints.

Newport's definition outlines a noble ideal, but we are happy to adopt a less demanding understanding of this notion, and kind of jumping ahead here a little bit. We understand a digital minimalist as one whose interactions with digital technology are intentional, such that they do not conflict with their ends.

For most, being a minimalist will involve a serious reduction, in some cases to the point of elimination, of interactions with smartphones, smartphone apps, and social media sites. For some, it may even require living up to Newport's ideal. I'm going to jump ahead one more time here. Newport may very well be right that we have prudential reasons to reduce our smartphone usage.

Perhaps most people would be better off if they became digital minimalists. But if the Kantian argument that follows is sound, then we might have even more compelling reasons to adopt the end of digital minimalism. We may have moral reasons. All right, so let's make sense of what just happened there.

They introduced my idea of digital minimalism. They simplify it a little bit to make it a little bit more general. But they say, basically, yes, this idea that Newport introduced is one of being very intentional about how you use your technology so that it supports instead of impeding what you value.

That is the core idea of my book, Digital Minimalism. It's at the core of my personal technology philosophy. The key thing they say, this is setting up the argument that we're going to explore, is they say, look, Cal in his book has what they call prudential reasons for why you should be a digital minimalist.

What they mean is like practical reasons. I go through, like, hey, when you let technology get in the way of your values, there's like all this stuff that you don't do that you would otherwise like to do, and I think you're going to like your life better. I'm a sort of pragmatic, practical, direct argument to your experience and intuition.

They're saying, yes, that might all be true, but we are going to make an argument that draws from Kant that says there is also a moral reason, that we can draw from moral philosophy that says whether you want to or not, you are obligated to be a digital minimalist.

That now is the argument that is made in this paper that we are going to draw out, a moral argument for being a digital minimalist. All right. So I'm going to jump back to the beginning here. They set up a quick example, which they use to explore some of the issues with modern technology.

So they begin by drawing from a quote from a comedian. So let me read this to you. I wish I could read. I really do, says comedian Esther Poviditsky. I try to read. I buy books. I open books, and then I black out, and I'm on Instagram, and I don't know what happened.

To many of us, this is a familiar occurrence. All too often, we set out to complete a task, but we are interrupted and subsequently derailed by our wireless mobile devices. Incidents of this kind might involve a moral failure, for insofar as we are morally required to cultivate and protect our autonomy, we fail to meet this requirement by falling prey to mobile phone addiction.

Okay. So the first link in the chain they're going to make in this moral argument is that our issues with smartphone usage, as captured by this anecdote of this comedian saying, "I try to read, but I can't. I'm going to read this book, but I end up on Instagram," they say this could be seen as impacting our autonomy, and if it impacts our autonomy, we might be able to find a moral reason why this is bad.

First, however, they have to establish, is the way we use things like smartphones actually affecting our autonomy, and what they do here is they go through three different ways that other thinkers have thought about autonomy, and for each of these says it's basically self-evident that the behavior we're thinking about, the behavior we observed in that comedian Povedisky, is violating these definitions of autonomy.

All right. So let's go through these real quick. They say, "In order to substantiate the claim that smartphone addiction undermines autonomy, we must say more about the concept at issue. Personal autonomy has been defined in a variety of ways, but we believe that a minimal definition of self-governance is sufficient for our purposes here." So they're going to go through some examples here of definitions of autonomy.

They say, "Let's return to Povedisky's case from the beginning. According to what they call the Frankfurt Dworkin model, Povedisky's first order desire to check Instagram while reading is inconsistent with her higher order desire, i.e., to want to read." All right, so the Frankfurt Dworkin model talks about first order and higher order desires, the things that actually is directing your activities right now versus what you want to be the case, and if these are out of sync, you have an autonomy problem.

So like by that model, looking at Instagram when you want to read is an autonomy issue because your higher order desire is, "I want to read," but your first order desire that's actually directing your activities is looking at Instagram. All right, here's another model due to Watson. On Watson's characterization, what is distinctive about compulsive behavior is that the desires and emotions and questions are more or less radically independent of the evaluational systems of these agents.

Povedisky's smartphone use is inconsistent with her evaluative judgments about what she ought to be doing, and the behavior is compulsive. All right, so Watson says, "If you're doing something that you would evaluate to be not good or less good than something else, then the behavior must be compulsive." By the way, that connects to the way psychologists think about behavioral addiction, the persistence in an activity even though it's, you know, it's not valuable or it's in the way of things you know to be more valuable.

Finally, they have a model of autonomy due to Bratman. Bratman defends a model of autonomy that requires harmony between what the agent does and her more or less long-term plans. Surely Povedisky's behavior fails on this count as well. We can suppose that Povedisky, like many of us, would like to read many books over the course of her life and to develop a disposition of being able to sit and enjoy reading for long stretches.

The action of looking at her phone compulsively is not consistent with her long-term plans. All right, so to summarize, no matter which model one adopts, the result is likely to be the same. Povedisky is not autonomous with respect to her smartphone usage. All right, so we've established this first link in our argumentative chain.

The way we use smartphones today seems to be hurting our autonomy. We can look at several official definitions of autonomy and see that smartphone usage of the type that we think of, the type in that example, is breaking those models. 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 get a free excerpt at calnewport.com/slow. I know you're going to like it. Check it out. Now let's get back to the video. Okay, so why is that bad? This is the next link in their moral argument chain.

This is where we turn our attention to Immanuel Kant. Although some ethicists reject the very notion of duties to oneself, Kant makes them a central component of his moral theory. In fact, I'll turn to this in the article here. For those who are watching along at home, I'm just scrolling.

This is section three. He says that they take first place and are the most important of all. He goes so far as to suggest that duties to oneself are the foundation of duties to others, making them the precondition of all moral duties. But he worries that they have not been properly understood and claims that no part of morals has been more defectively treated than this of the duties to oneself.

He thinks that they have been misunderstood as a mere elevation of self-interest, a duty to promote one's own happiness, which he dismisses as an absurdity. Rather than grounding such duties in egoism, Kant argues that humanity, i.e. rational nature, has an absolute inherent value, and this generates self-regarding obligations insofar as the agent is morally required to respect humanity in her own person.

Thus duties to oneself are derived from the humanity formulation of the categorical imperative, which tells us that we must always treat humanity, even in our own person, as an end, never merely as a means. All right, so Kant is arguing we have a duty to ourselves as much as we have a duty to other people, and we have a moral duty in particular for self-governing what is most important about our humanity.

Okay, so what is that? Jump ahead briefly. Kant famously claims that human beings, in virtue of their rational agency, have a uniquely elevated status which he calls "dignity." So the idea that we are rational beings and no other creatures or objects are gives us this sort of special value, and preserving what he calls the dignity of ourselves as rational beings is sort of the highest good.

We have an obligation to help protect this in ourselves and others. Moving on here, on this view, our actions can either express or fail to express the kind of respect that is becoming of human dignity. So he's saying our actions need to be focused on respecting our human dignity, and our human dignity is based on the idea that we are rational beings.

All right, so we're really in the weeds here, Jesse. We're deep in moral reasoning, but out of this and a bunch of other words that I'm kind of skipping, we get to an actual argument form here, all right? So they end up with three propositions that lead to a conclusion.

Humanity is proposition one. Humanity, i.e. rational agency, has an objective, unconditional, non-fungible value, which is dignity. Proposition two, anything that has dignity ought to be respected as an end and never treated as a mere means. Proposition three, if humanity ought to be respected as an end and never treated as a mere means, then we have an imperfect duty to cultivate and protect our rational agency.

Therefore, the conclusion of these three propositions, we have an imperfect duty to cultivate and protect our rational agency. All right, we're getting in the weeds here, logical philosophy, morality all pulled together. They then put these together, they get to their core argument. We have an imperfect duty to cultivate and protect our rational agency.

If we have an imperfect duty to cultivate and protect our rational agency, then we ought to adopt the end of digital minimalism, therefore we ought to adopt the end of digital minimalism. If you take my discrete mathematics course at Georgetown, where we study propositional logic, you will actually recognize this argument and could probably turn it into the corresponding argument form.

So basically we have just done a lot of reasoning based on Kant's ideas that end up with the conclusion we ought to adopt the end of digital minimalism. If we simplify all of this, we're basically saying it is important to respect our own dignity as rational beings. The way that we use smartphones when we're unintentional robs us of our ability to do this because it robs us of autonomy, and autonomy is at the core of respecting the dignity of being a rational being because at the core of being a rational being is the ability to make decisions about what you do rationally.

The Kantian framework is seeing this core tension between we need to respect the fact that we are rational beings and smartphones take away our ability to make rational decisions about what we want to do with our life and our time, therefore an approach to life that reduces smartphone's ability of taking away our autonomy is justified.

Digital minimalism is sort of the definition of such a life. It's an approach to digital technology. It says we want to be intentional, not be robbed of our autonomy, therefore there's a sort of fundamental Kantian moral argument that we should be very intentional about our technology use using something like digital minimalism.

Let me read the conclusion of this paper because I think they sum this up very nicely. They're all lit up here on the screen for those who are watching at home. All right, so here's the conclusion of the authors in this paper. We have argued that there is a moral obligation to be intentional about our use of smartphones and other addictive devices.

We have this duty because we are required to protect the most valuable commodity we possess, autonomy. Kant believes that the proper exercise of our autonomy is the only thing that is good without qualification, something that shines like a jewel having its full worth in itself. To wantonly forfeit some of our agency by falling prey, the technological heteronomy is to demonstrate a failure to respect this precious capacity as the treasure that it is.

All right, so why are we geeking out on, so it's like a technical academic argument for the type of things we talk about here on the show, and actually not just the type of things, specifically what we talk about on the show because, of course, they're talking about my specific digital minimalism philosophy.

It's because I think it is easy in thinking about technology and human flourishing to fall back on arguments such as, look, kids these days, for example, they're always using different technology, we get worried about it, but that's just the wheels of progress. Or we fall back on an argument that says every new technology creates moral panics, right?

And then we get over it. We worried about the car, but now we just drive cars. We worried about TV, now we don't worry about TV as much. It's easy to fall back on these arguments of status quo thinking. What's critical about this particular argument is it says, no, there's justifications for our concerns about these technologies that are much more fundamental than thinking about specific technologies.

Our uneasiness about these technologies is not just a naive reaction to the latest techno disruption and a long line of techno disruptions that ultimately end up being not so bad. We are actually reflecting a specific harm, denial of autonomy, and that we can go back to Kanter before to see that this is at the core of the human experience, is at the core of what we value as humans.

And so yes, we're uneasy, not because we're naive, we're uneasy because something basic to our humanity is at stake. So this Kantian argument is pointing towards the exceptional nature of the issue we face with things like smartphones. We cannot just ontologically speaking put in the same category as like any other type of techno fear.

It is a specific technical fear which requires analysis on its own terms and when we do that, we see there are specific harms here that cannot be ignored. So if someone's giving you a lot of trouble about your digital minimalism, if they're making fun of you or if they're trying to self-justify their own heavy phone use, you can now throw a lot of sort of annoying technical philosophical terms at them.

You could say things like heteronomy and ontological. You can mention the categorical imperative, keep dropping the word Kantian and they will just have to be quiet. All right. So there we go. A nerd argument for a very real issue. So I actually found that article, Jesse, because there's a follow-up article that said, all right, if that's true, there's also an obligation others have to protect your autonomy through digital technology as well.

That there's like a moral imperative not to distract other people. It's a whole interesting argument. All right, but enough of that nerdiness. We got some good questions. But first, let's hear from a sponsor. All right, let's talk about our longtime friends at Element. Element helps anyone stay hydrated without the sugar and other dodgy ingredients found in popular electrolyte and sports drinks.

It's important to have the right electrolytes, but you don't want sugar. You don't want junk. That's why I love Element. It is a zero sugar electrolyte drink mix and sparkling electrolyte water. So there's two products here. The mix you add to your own water or the already canned sparkling water with the electrolyte mix in it, which you can just grab out of the fridge.

They're born out of a growing body of research revealing that optimal health outcomes occurs at sodium levels two to three times government recommendations. So each stick of powder or can of sparkling water delivers a meaningful dose of electrolytes free of sugar, artificial colors, or other dodgy ingredients. It's incredibly popular among many different communities.

I use Element almost every day. Anytime that I feel like I'm dehydrated a little bit, so from like after exercising, after podcasting, you lose a lot of moisture when you talk, after a long day of lecturing, I add the drink mix to a big Nalgene full of water. I drink it every day.

We've got a big box of it because I love that it replaces electrolytes, but it doesn't have all that stuff I don't want. It's not the sugar. It's not the weird dodgy stuff. There's something else they're offering now to think about this winter. That is enjoying your Element hot.

So the limited time Element chocolate medley, including the flavors chocolate mint, chocolate chai, and chocolate raspberry, you could serve those in hot liquid. So get your electrolyte mix in a sort of refreshing hot liquid after you're running outside or shoveling snow. I think that's a fun addition to the Element family for the winter.

Now I have good news. Members of my community can receive a free Element sample pack with any order when they purchase through drinkelement.com/deep. That's drinkelementlmnt.com/deep. Keep in mind you can try Element totally risk-free. If you don't like it, give it away to a salty friend and we'll give you your money back.

No questions asked. They have a very low return rate and very high reorder rate because it is fantastic. So go to drinkelement.com/deep and you will get a free sample pack with any order. I also want to talk about our friends at Mint Mobile. Here's the thing. I love a great deal as much as the next guy, but I'm not going to crawl through a bed of hot coals just to save a few bucks.

It has to be easy. No hoops, no BS. So when Mint Mobile said it was easy to get wireless for $15 a month with the purchase of a three-month plan, I called them on it, looked into it. Turns out it really is that easy to get wireless for $15 a month.

The hardest part of the process is just the time you're going to spend breaking up with your old cell phone provider. I have been recommending Mint Mobile to a lot of people recently because my son is in middle school. We know a lot of other middle school parents. They're going through this thought process of my kid needs a phone for X, Y, or Z, taking the public transportation, getting picked up from practice.

I don't want to give him a smartphone. What do we do? I've been telling them just, hey, buy a flip phone on Amazon and give them the $15 a month plan for Mint Mobile. It's cheap, it's easy, and they'll have access. So I've been telling more and more people use Mint Mobile to quickly set up a dumb phone for their kids.

So if you want to get started, go to mintmobile.com/deep. There you'll see that right now all three-month plans are only $15 a month, including the unlimited plan. All plans come with high-speed data and unlimited talk and text delivered on the nation's largest 5G network. You can use your own phone, or a phone you just buy on Amazon if you want a dumb phone, whatever you want to do, with any Mint Mobile plan and bring your phone number along with all your existing contacts.

Find out how easy it is to switch to Mint Mobile and get three months of premium wireless for just $15 a month. To get this new customer offer and your new three-month premium wireless plan for just $15 a month, go to mintmobile.com/deep. That's mintmobile.com/deep. Check out your wireless bill to $15 a month at mintmobile.com/deep.

$45 upfront payment required, which is equivalent to $15 a month. New customers on first three-month plan only. Speed slower by 40 gigabytes in unlimited plan. Additional taxes, fees, or restrictions apply. See Mint Mobile for detail. All right, Jesse, let's do some questions. First question's from Kirsten from Missouri. "I just ordered digital minimalism to help me stop wasting time.

I tried, without success, to use some distraction-free apps. Can these apps actually work?" Well, let me back up a little bit, because it looks like you're thinking about digital minimalism as a collection of tips for trying to stop wasting time or to be less distracted. That's the standard cultural paradigm we often have for thinking about advice, especially for things like digital distraction.

Give me some tips. I want to read the five ways to save your attention in some sort of magazine article, and let me put a couple of those into play. It looks like you heard about distraction-free apps. I do talk about them in the book. You're like, "Hey, I tried them.

It didn't work well." The first thing I want you to understand is that digital minimalism is not a collection of tips, but a philosophy. It's a philosophy of technology use, a consistent way of thinking about the role of technology in your life and how you curate and engage with technologies in your life.

It's a bit of a binary proposition. You either need to adopt the philosophy or not. You can't just pick and choose specific things to show up in the philosophy itself. The metaphor I like to use is cleaning out a closet that's overstuffed with junk. So if you had a closet that's overstuffed with junk, what you are doing—so the equivalent in the closet metaphor of like, "Hey, I'm going to try to help my online distraction because I heard apps might help"—that's the equivalent in our closet metaphor of going to the container store and being like, "Hey, I bought some organizers." And then you return to your overstuffed closet and you're like, "Okay, I have a few organizer bins here that I put some of the stuff in." Your closet's still a nightmare.

What works better for the closet? Well the Marie Kondo approach of, "I'm going to empty the whole thing out, empty it to zero. I'm going to put everything that was in that closet in the piles. I'm going to go through and say, 'Which of this stuff do I really, really need?' And that stuff I will put back into the closet very carefully.

If I don't really need it, it doesn't go back in." That's how you organize your closet. It's a net-zero budget. You start from zero and add back in the stuff you need carefully, not just trying to throw organizational bins at the junk that's already in there. So that's what digital minimalism is, right?

Instead of just throwing a particular piece of advice at your digital distraction, you're going to reinvent your digital life from scratch, taking everything out, reflecting for a month, and then only adding back in what matters with rules about how you're going to use it. Okay. To get to your specific point about distraction-free apps, in this process they can have a use—typically they're useful training tools.

If there's a particular technology that you need to use but only use in a limited way and you have a very strong urge to keep going back to it, distraction-free apps can help you train to resist that urge because it makes it very difficult to access those technologies in times you don't want to.

Typically what happens with people is after a few months of using distraction-free apps, they lose the urge to go use that technology compulsively because they've gotten that groove out of their mind, their reward circuit weakens, and then they don't use them anymore. So you might use a distraction-free app as part of your efforts to recreate your digital life, but typically their uses are more temporary, they're a training tool, not a permanent feature of your digital landscape.

I love the condo advice, how she—and you get rid of something and you express gratitude to it, and then you're like, "Bye." TikTok, I express gratitude to the role you play in my life. But now our journeys have crossed. Our journeys have parted. Yeah, she's very calming. All right, who do we got next?

Next question is from JP, "I get stressed with my goals due to fear of failure, I keep everything in my head and can't distinguish from urgent and non-urgent, I pretty much never finish what I start." Well, look, you can't, in the modern world, you can't organize your life just in your head.

Just trying to remember what you want to do, all the things you have to do, your priorities, somehow use this all to make a decision about what to do next. The human brain can't do that. It's like trying to teach a bear to drive a car. It might be funny or terrifying, but it's probably not going to work very well.

The human brain can't, on its own, organize a modern life. So what do you need to do instead? Well, you're going to need something like multi-scale planning that is a system that can get all of the stuff you need to do and your plan for how you're going to tackle it out of your head and into a sort of trusted permanent system that you can frequently access.

You basically have to extend your mind like a cyborg with other tools to make this complicated task of organizing your life much more tractable. So multi-player scanning, which I talk about a lot on this show, has you planning things on multiple scales, each of which have their own systems to go with it.

At the highest scale, you have a plan for the current season or quarter where you're making sense of your bigger priorities. You reference that quarterly or seasonal plan every week. When you make a weekly plan, you physically write out your weekly plan. Also when you do your weekly plan, you confront your calendar.

You make adjustments. You add to your calendar appointments with yourself to work on particularly important priorities from your quarterly or seasonal plans. So now you have a written weekly plan and a calendar that's been updated and corrected for the week. And then every day, you look at your calendar and your weekly plan when you make a time block plan for that day where you give every minute of your workday a job.

So you're not just trying to decide on the fly, "What do I want to work on next?" You've made a plan for the time that remains in your day between meetings and other appointments, what you want to do at that time, so you can balance your energy with your needs.

You can batch. You can be efficient. You can avoid excessive context switching, etc. So each of these levels, you have different tools. All of this is going to be supported by a task system where you're going to keep track of all the obligations you have to do. You'll look at that task system when you're doing your weekly plans.

You'll reference it during admin blocks in your daily plans. All of these are external systems with structure around them that you use so that your brain doesn't have to be responsible on its own for keeping track of what you have to do and making decisions in an ad hoc way.

So you need something like multi-scale planning if you're going to keep track of your life. So you shouldn't be worried, I would say, that you're struggling to do this in your head. That's a bear driving a car. That is pretty impossible. Actually, a true point about bears driving cars, Jesse, really hard to get insured.

It's a really high insurance rate. Especially if you live in the west coast of Florida. If you live in the west coast of Florida, yeah, get a bear to drive a car, high insurance rate. All right. What do we got? Next question is from Joseph. I'm a teacher looking to improve my efficiency with admin tasks.

When I have a quiz in two different subjects to grade and record, is the task context switching effect less if I were to grade and record one class than the other or if I were to grade both and then record both as blocks? Well, it's a good question because I'm glad you're thinking about context shifting as more or less the number one productivity poison you want to be wary about.

Long-time listeners know this. It takes time to switch your target of your attention from one target to another. So if you're moving your attention back and forth rapidly, you're going to put your mind into the state of continuous partial attention, which is a self-imposed cognitive deficit. You make yourself quite literally dumber.

In this case, recording grades into a gradebook is mechanical and largely non-cognitive. In other words, you don't have to do difficult thinking. You're just taking numbers, matching it to a name, writing that name in there. You don't have to do difficult thinking. You don't have to load up complicated cognitive context.

You don't have to make decisions or pull from complicated memories. So I'm not too super worried about the context shift price when you go to just entering grades. Because again, mechanical thing, it's almost like you're working hard on something like writing a hard book chapter. If you get up and go make a cup of tea and then come back, that's not actually going to be a big hit.

That context shift is not going to be a big hit on your primary test because it's mechanical and not cognitive. So this is all to say, it doesn't really matter where you put the grade recording. It's whatever you have preference for. So you could grade one thing, then grade the other thing, and then have a long block of just mechanically entering grades.

Or you could grade one thing, enter the grades, grade another thing, enter the grades. It's not going to make a difference cognitively. It's just going to be a matter of what's going to feel better for you. I would suspect the difference would come down to how demanding the grading is.

So if the grading is really hard, and if the subjects between the two things you're grading the two quizzes is separate, like it's not the same cognitive context, I would enter the grades right after grading to give your mind a breather. I would also consider, let's get advanced here, cognitively advanced.

After grading the first thing, before you enter those grades, go look at the second thing, and maybe read one of the quizzes to try to start loading that cognitive context. Grade a single quiz of the next thing. This is going to go slow at first, right? Because you now have colliding cognitive context as you look at the second quiz for the first time, that's a separate context from the quiz you just graded, and so there's going to be a collision as your brain is trying to shut down the context of the first grading block and load the context of the second.

So grading that first quiz of the second, the first assignment of the first quiz, I don't know how to say this, Jesse. It's a quiz, and he has multiple quizzes to grade, and there's two different quizzes. That make sense? Yeah. All right. So you're in the second type of quiz, and you grade the first quiz from the second type of quiz, and that's very hard because it's a new context.

Grade just one from the second type of quiz. Then here's my advanced advice. Go back and enter the grades from your first quiz, because here's what's happening. While you are entering the grades from the first type of quiz, your brain is continuing in the background the process of switching its context over to the second type of quiz.

You initiated that by looking at a single quiz of that second type, and now you go back and just mechanically enter grades, your brain is going to continue making this switch. So now when you're done entering those grades and you return to the second type of quiz, your context has more thoroughly shifted, and your grading is going to get up to speed much quicker.

All right. This is kind of an advanced way of thinking about it, but this is what you probably would see this effect. If instead you grade the first type of quiz, you enter the grades, then you turn to the second type of quiz, it might take—you might have to grade five or six students' quizzes before you get that momentum going of your brain completely shifting.

Whereas with my tactic, you grade the first type of quiz, grade one of the second type, enter the grades, then return to the rest of the second type, you'll probably get up to speed much quicker. This is just, at this point, like attention hacking. The differences might be minor.

But I do like the type of thinking this induces, which is to think about cognitive context. It's like one of the most important properties of modern work, and it's the property that we think almost nothing about in modern office productivity. We completely disregard it. We put low friction as a priority.

We put information velocity as a priority. We put access to tools and data as a priority, and we completely disregard the cost of context shifting. So I love any discussion like this that gets us in the weeds on it. This sort of psychologically aware productivity is really where we should be.

So I appreciate the chance to sort of nerd out on that. We'll have to have Joseph respond and see how it goes. Yeah, he should. So Joseph, if you hear this, let us know if that technique works. All right, where are we? Next question is from Francois. I'm a professor and also have a French podcast.

My university has agreed to include it in my official duties. I haven't accepted their offer yet. However, if I do accept it, how should I think about organizing my podcast within the traditional academic framework of research, teaching, and service? Should this be included in my academic tasks? It's a good question, Francois.

I had a French podcast for a while. I don't know if you know this. Yeah, and you had a pipe and a hat. I had a pipe and a hat, and I just did my French accent. Long story short, I am no longer welcome in the Republic of France.

I have been banned from setting foot in France after they heard my awesome accent. Francois, it's a good question. I don't know the French system super well, so I'm going to answer this from the perspective of the American academic system, which I think is roughly congruent. All right, so in the American academic system, by far the most important thing for promotion and recognition is research.

You have to do service. You need to be a good teacher. But those alone can't get you promoted or recognized. It has to be the quality of your research. So if your university is going to allow you to count your podcast as an official academic task, I would recommend that you are very clear about which of the three major tasks, research, service, and teaching, that it counts as, and I would try to make it count as, service.

When you count it as service, what this means is you can reduce the amount of other service you do, let the podcast take the place of other service obligations so you're not increasing your time obligations, and critically, you're not reducing the time you spend on research. Because when it comes to service and promotion, it's a little bit more binary.

Was this person a good citizen of the institution and his community? Not how good of a service person were they, right? So if you can use your podcast as a way to reduce other types of service so your overall time footprint's the same, that's great. Do not let it, however, impinge on the time you spend doing research.

That's ultimately what matters most. Trust me, I've gone through two promotions. I'm done with promotions now, but I went through both my promotions from assistant to associate with tenure and from associate with tenure to full. Both of those promotions, I had large portfolios of more public-facing work, and I had to deal with them carefully.

When I was promoted to associate, I didn't mention my books. It was all computer science research. When I went to full, I did mention them, because as we just saw in the deep dive, some of the work, public-facing work I did also has a very big academic footprint, right?

We just did a whole paper from the Journal of Applied Philosophy in the deep dive that was responding to my digital minimalism book. My book, Deep Work, has been cited in academic articles close to 800 times now. I was just looking at that the other day. So I did sort of count that more, but I didn't lean into my podcast, even though I was past a 10 million download point at that point, because it didn't quite fit clearly into it.

It's not research. I didn't have an agreement like yours that this counted as service. I sort of had it as sort of on the side. So I know this world well. Ultimately promotions matter. Are you doing work that's influencing the academic culture? So that's what you've got to be careful about.

So yeah, if you can use your podcast to reduce other service loads, then I think that's great because your podcast will probably have a higher impact than the other service you're replacing. Just don't let it get in the way of research. So now that you're a full professor, do you still have to write as many papers?

It's more flexible. Yeah, it's more flexible. Right now I'm focusing more on technology and digital ethics than I am computer science. And that's the type of thing you can explore. It's sort of the advantage of full professoredom in 10 years. You can make those explorations. And if you don't like the way it's going, you can switch back to something else.

One of the things that made a big difference for me is that-- so Google Scholar is a quick way you can keep up on people's publications. What have they published? How much have they been cited? What are their statistics, like their H index, their I-10 index? What are their total citation counts?

What are their total citation counts by years? Once Google Scholar figured out-- because I write under two different names. My academic computer science papers are typically written under Calvin Newport. And of course, my public-facing writing is under Cal Newport. When it figured out, oh, Calvin Newport and Cal Newport are the same person, it really changed my statistics.

So where you saw a bit of a fall off in citations in recent years now shows a steady high level of citations because the public-facing work on technology I was doing as Cal Newport gets cited a lot academically. So it shows a sort of smooth transition from less computer science, more digital ethics.

And the impact is measured by citations as sort of stayed steady. I thought you were saying the other name was going to be your French name. Yeah. Well, yeah. There's Calvin Newport. There's Cal Newport. And there's Pierre, Pierre Le Newport. Actually, I do have a French-- I do have French heritage.

My paternal grandmother was a Levelle, the Levelle family. And the Levelle family goes all the way back to the French Huguenots that came over here pre-Revolution. And they used to be Levalve. So it's French. I have a French Huguenot blood back in there. All right. What do we got next?

We have our corner. Oh, Slow Productivity Corner. Let's hear that theme music. So for those who are new, Slow Productivity Corner is the question. We do one question each week related to my most recent book, Slow Productivity, the Lost Art of Accomplishment Without Burnout. Went to Amazon. So I can see this now.

Went to Amazon's Best Business Books of 2024 and the winner of Best Business Book of the Year from the S-- something, something, something-- S-A-B-E-W. It has those letters in it. There's a huge cash prize, guys. This is an important award, somewhere between $60,000 and $600,000 award. My award-winning book, Slow Productivity, we try to do a question each week that comes from that book.

If you haven't read the book yet, come on. Get the book. Half the stuff we talk about comes from it. All right. What's our Slow Productivity Corner question of the week? It's from Madonna's Gold Tooth. Yikes. What do you think about the slow living craze on the internet? And do you think it's just a fad or will it be permanent?

All right. So do you know what this is, Jesse? There's a YouTube video. You have a YouTube video. Yeah. It'll explain what slow living is? Because I don't actually know what this is. All right. Let's load this up. Let's listen here. We're going to be learning this together, what slow living is, and then I can answer this question.

Madonna's Gold Tooth. All right. There seems to be-- One of the things I am most scared of in my life is looking back with regret. Looking back at all the little moments that I missed in pursuit of more. This year, I've been really working on slowing down and trying to be more present for the little moments that make up most of our lives and making sure that I'm actually building a life by design and not just by default.

So these are some simple, tiny habits that I have implemented that have really helped me slow down. Five minutes of nothing. This is literally a block that I have on my calendar every day just to have five minutes of nothing happening. Doesn't sound like a lot, but when you don't have any music, no podcast, no work that you're thinking of, no work that you're doing, no book that you're reading for just five minutes and you sit there and you stare at a wall, not trying to meditate, but you just let your brain do its thing and you observe it.

It's a beautiful time for me to just reset and make sure that there's actually breaks in my life to remember that my life is not online. It's not on a computer screen and sometimes I need physical breaks to make that happen. Three zones. For me, this is my sauna, my cold plunge.

All right. I think I get the idea. Well, first of all, for those who are just listening instead of watching, the video had like a faux graininess while they played like really relaxing music and he washed eggs from the chickens that he just stared at. I think with that type of music.

I like the music a lot. Yeah. I think almost anything seems, almost anything seems profound. 73,000 views. Yeah. I mean, almost anything I think will sound sort of important and meaningful and somber. You could have a video of someone earnestly trying and failing to life threatening in a life threatening way to get a bear into the car to drive it.

Played to that music. Like good for him. Good for him. The bear's mauling him as he's trying to get him into a Chevy Impala. Play it to that music and cut to some scenes of someone washing eggs. You'd be like, "Yeah, life is like a bear trying to get into a car." Drive to the insurance agency and get the insurance card.

Just him at the insurance agency, just his face ripped up in bandages trying to get the insurance guy is just shaking his head. You'd be like, "Yeah, that's... Play that music." You're like, "Yeah, it's profound." So no. Slow living and slow productivity are different. So let me tell you how, and then I'm going to tell you what slow living seems to be like connected to some other stuff we talk about.

Slow productivity is about work. It's about knowledge work. It's about how do we define what productivity means in knowledge work. The core argument of that book is that we have a bad implicit definition that we tend to fall back on, which is pseudo productivity, which is to use visible effort as a proxy for useful activity.

Slow productivity is a alternative. It says our goal in knowledge work should not be to be as busy as possible. We should instead focus on not doing too many things at the same time, keeping our pace of work varied and natural, but then really obsessing over quality. This is a better, more sustainable definition of productivity.

This is about work. Slow living seems to be about life outside of work and a lot to do with distraction, especially digital distraction. So it's probably closer to digital minimalism when it comes to the things I talk and write about than anything else. I think if you're a digital minimalist, your life will seem slower in the way that's being talked about in this video, because a digital minimalist works backwards from their values to dictate their technology use, and so, you know, if you value your chickens and washing your eggs or whatever, you are going to be careful about crafting your technological use so that you're not always looking at your phone and you can't enjoy doing that.

In general, digital minimalists do feel like their lives are slower and richer. There is a neurological reason for this, right? Your life is what you pay attention to. So if you're constantly paying attention to your phone, you perceive your life as very sort of like fast-paced, emotionally activated, sort of this like really sort of shaky, jittery world that's always rolling past, because when you're looking at your phone, everything's moving fast.

Swipe, swipe, swipe. Tap, tap, tap. Look at this. Look at that. Jump over there. Time moves fast because you're moving fast on your phone. Also time moves fast because you're doing this sort of homogenous behavior. So when you're doing sort of the same thing, you don't have a really good sense of how long time is.

Time can just sort of unfold. When you're not on your phone and engaging in specific behaviors, it is just by definition slower because everything is slower than using your phone. And because those behaviors are novel, they're different specific things in novel specific locations, your perception of time is of it being much slower.

Your day seems longer. Your experience is richer. So I think digital minimalism will probably lead you to something like slower living. Start with the digital minimalism and end up at the slower living. It's sort of a consequence of getting intentional about your life and technology. But it is quite separate from slow productivity.

They share the same word slow, but they're only connected by this idea of sort of intentionality. Slow productivity is about your work at your desk. Slow living is about your life outside of work. Does that seem reasonable, Jesse? Yeah. Maybe we should have chickens in here. I like that video.

We should do more of that music. We have kind of music, cooler music like that for the in-depth episodes. Yeah. That's a little more like a meditative music. All right. Do we have a call this week? We do. All right. Let's hear this. Hi, Cal. I've noticed I struggle with tiredness and a lack of focus in the afternoons, especially during my scheduled deep work blocks.

After lunch, my mind doesn't feel as sharp and I often find myself drifting off and daydreaming. Do you have any strategies to help maintain focus and mental clarity during these times? Thanks. Well, first of all, we have to keep in mind that there's a limited capacity to do deep work in a given day.

So if you're talking about like highly demanding focused activities, things that require you to use your full cognitive capacity, probably do those in the morning. Do those first thing before you've had a lot of context shifts so your mind is still clear. And be okay. I got in a good, hard early session.

I'm okay not having to return to these cognitively demanding activities in the afternoon because I'm just not going to have enough cognitive gas. If it's more just, "No, I have administrative stuff to do. I have to take notes. I have to send emails." It's not cognitively demanding, but I just sort of lose focus and drift and lose energy in the afternoon.

Well, that's very common as well. And a couple things that helps is time blocking. So instead of having to constantly have an argument with yourself of like, "What should I do next? Should I keep doing this? Should I take a break?" Time block those afternoons and just make the single commitment to stick to your time block schedule the best you can.

So you get rid of a lot of that decisional friction that comes from being more freeform in your approach to your afternoon. Second in that time block schedule batch. So let me do a lot of similar tasks together because even if minor, sticking within the same cognitive context makes it easier.

This can apply even to cleaning out your email inbox. I recommend if you have like a super stuffed inbox and it's like three o'clock and you're exhausted but you kind of have to get through it, create a folder or label for the current messages you're answering. And then go through and grab a bunch of messages of the same type.

So they're all relevant to the same cognitive context, they're all scheduling messages. They're all messages related to like an upcoming event. Move those all to that label or folder and then tackle those just by themselves. So now you're doing messages without having to change your context. Then go and grab another type of messages and do the same.

What happens is if you follow the alternative of just sort of doing your emails in the order they exist in your inbox, you're switching potentially your cognitive context from message to message to message and that's exhausting. So that can help as well. Third, end your day earlier. Hey, I'm exhausted by three or four that may be like stop your day between three and four.

Time block your day. If you're doing multi-scale planning, you have a good weekly plan, your weekly plan is in touch with your quarterly plan. This is a key idea from my book Slow Productivity. The second principle, it says work at a natural pace, which says this idea that like the perfect calibration for humans that do cognitive work is nine to five all out every day is preposterous.

Why would that just happen to be optimal for everyone? You might find out working till three or 3.30, this is really what's optimal. You're time blocked, you're on it, and then when you're done, be done. Or maybe four, maybe two, it could be different for different people. But don't feel like you're too stuck with it has to be this exact eight hour day.

Some people just run out of gas earlier than others. Your work might be harder than others, so you need to end earlier. I talk about in, I don't know if this is in Slow Productivity, I think this is in my book A World Without Email. I talk about this type of programming called extreme programming, and it's pair-based and it's super intense, and it produced fantastic code, but it's super intense.

I report that companies that do this type of coding, they produce really cool stuff, but they have to let people go home by like 2.30 or 3.00. It's just too exhausting. You can't do it till five. People at first have to go home and take naps. So don't assume that everyone is perfectly calibrated to work all out till five.

Figure out what works for you. If you're organized and on the ball, you'll produce good work. So I would vary it that way as well. All right, let's see here. We have a case study. This is where we have people write in where they talk about their personal experience putting the type of things we talk about the show in the practice in their own lives.

All right, so today's case study comes from Zach. Now here's what Zach says, "Recently, I've made a monumental life change for the better in no small part due to CALS, books, podcast, and newsletter. I graduated in March of 2020. While my classes went online, I decided to get my real estate license and pursue my interest in real estate investments because of the high autonomy and market activity due to the interest rate environment at the time.

I was successful. I specialized in commercial investment sales and became proficient in my field because of my implementation of deep work principles. The only problem was that I was miserable at work. My days mostly consisted of cold calling and driving all over the state for client meetings. So even though I was making decent money and had full autonomy, my lifestyle wasn't great and it was trending in the wrong direction.

On top of that, I was working mostly solo while I'm a very team oriented person. After listening to your podcast religiously on my long drives, my mindset began to shift. I realized I was optimizing for autonomy and money without much thought to lifestyle and long term life design. So I saved up some money and quit.

Believe me, this was tough. Leveraging CALS principles got me far relatively quickly. So I had a promising career trajectory. But when I looked at guys way further down the road, they had a lot of material success without intentional design. After hunting and interviewing with jobs that align with my long term lifestyle vision for a few months, they successfully landed a job at a tech startup that provides me a much better day today.

It's a short, beautiful commute to an office in my favorite part of town. My work is varied, challenging and interesting. And most importantly, I'm working with a like minded team who are all just as obsessed about productivity systems as I am. I just finished my first week and I'm blown away at what a difference this intentional change has made in my life.

For the first time in years, I'm bursting with excitement to go to work. Should I have applied for jobs while working? Probably. But I was so burnt out that I had a burn the ships mindset. I'm eagerly awaiting your next book nearly as much as I am awaiting Brandon Sanderson's the author of Name of the Wind.

I didn't include that, leave that in there. Brandon Sanderson. I still want to go down. I told you I have an invitation to go see his lair. Yeah. I have to do that. My wife is going on a trip down there. To Brandon Anderson? Well, no, not to it.

That would be weird. She's like, I'm going on a trip, I'm going to be spending a week in Brandon Sanderson's dungeon. I'd be, I'd be worried about that. No, she's going to that part of the country. And all I could think is like, if that was me going on that trip, I would be able to see Sanderson's lair.

Yeah. It'd be interesting if I get there and it's just half of it's just a sex dungeon. Probably not. I think he's a pretty straight laced Mormon, but you never know. The least popular pornographic video of all time is titled "Brandon Sanderson's Sex Dungeon." Six views. All right. That's a great, Zach, I appreciated that.

Two things I want to point out about that case study. One, lifestyle centric planning. That's the way to think about your career. It's one of the most important dials you have to turn in trying to construct your lifestyle. But what matters is the target lifestyle. What do you want the day-to-day of your life to be like?

You work backwards from that vision. That's how you help figure out what work to do or not do. This is much more effective than either following your passion or just blindly following a clear metric like money and just hoping by happenstance that will lead you being happy. The other thing I want to point out about this example though is, okay, Zach started one job, didn't work out, he switched.

Is that a failure? No. It's very common. Figuring out the components of your ideal lifestyle is difficult and it evolves with experience. He had a hypothesis, I think, built on autonomy and financial security. He had a hypothesis of a lifestyle vision that he thought would be ideal for him.

Zach pursued a job that matched that hypothesis and then learned through real-life experience, "Oh, there's these other things I care about. I didn't realize them until I had them not be present in my life. I didn't realize autonomy without X, Y, and Z wasn't so good, the money thing I don't care so much about." Through life experience, he updated his priors.

His vision of the ideal lifestyle evolved and he said, "Great. Let me now leverage my career capital and make a shift that's going to get me closer to that lifestyle." Now, in this case, the career capital he leveraged was literal capital. He was making good money, so he saved up enough to buy him time to make a switch.

He was early enough in his career that sort of skills-based career capital was less useful or less important because he was still a pretty early-stage career. Then he used that money to buy him some time to find a job that focused on other things he had discovered were important and now he's much happier.

That's lifestyle-centric career planning in action. It evolves, it's tactical, it's not sexy. It's not Brandon Sanderson's Sex Dungeon sexy, but it's what over time is going to make your life more fulfilling. I've worn my VBLCCP hat a few times now. I've been wearing my Deep Life hat regularly. No one has asked me yet or noticed what VBLCCP means.

I haven't got a reaction to it yet, but I'm still thinking we'll find our first. You get any questions about Deep Life or people just assume it's a brand? No, no questions yet. Yeah, I'll see. I'm going to keep wearing mine until I find a true believer, but I haven't found them yet.

All right, we've got a cool final segment coming up, a Tech Corner segment, but first let's hear briefly about another sponsor. You know what's not fair? The fact that Netflix hides thousands of shows and movies from you based on your location and then has the nerve to just keep increasing their prices.

Now you could just cancel your subscription in protest, or you could be smart about it and make sure you get your full money's worth, like I do, by using ExpressVPN. So we talk a lot about VPNs on the show. I'm very clear you should use a VPN. The way it works very briefly is that instead of just directly accessing a website or a service with a VPN, you instead connect to a VPN server.

You tell that server with an encrypted message the site and service you actually want to use. That server talks to it on your behalf, encrypts the response, and sends it back. So that means anyone monitoring your internet usage only learns that you're talking to a VPN server. They don't learn what site you're talking to.

They don't learn what service you're talking to. One of the advantages of doing this, beyond just the obvious privacy advantages, the hacking advantages, the security advantages, is if you connect to a VPN server in a different location and that server talks to Netflix on your behalf, Netflix thinks you're in that location.

So ExpressVPN has servers all around the world. So you can select a server in like whatever geographic zone you care about, and then you'll get that zone's Netflix content or whatever streaming service you're using when you use that app. That's an extra bonus thing you can get, a benefit of using a VPN on top of all the other ones.

The reason why I like ExpressVPN is that it's easy. You fire up the app, right? You can change your location of the server with one click. When it's on, which is easy to do, you just click to turn it on, you just use all your websites and apps like normal.

And all this happens transparently in the background. It works on phones, laptops, tablets, even smart TVs and more. It's super fast. It's got high bandwidth. There are servers all around the world. So like there's probably one nearby to get the fastest speed. You can stream in HD with zero buffering through it.

So it's got great sort of best in class speed. It's rated number one by top tech reviewers like CNET and The Verge. That's why of the VPNs that are out there, I recommend ExpressVPN. Right now you can take advantage of ExpressVPN's Black Friday Cyber Monday offer to get the absolute best VPN deal you'll find all year.

Use my special link expressvpn.com/deep and you'll get four extra months with the 12-month plan or six extra months with the 24-month plan totally free. That's expressvpn.com/deep to get an extra four or even six months of ExpressVPN for free. I also want to talk about our friends at Shopify. When you think about businesses whose sales are rocketing like Feastables by Mr.

Beast or Thrive Cosmetics or Silicon Valley's Weekend Uniform supplier Cotopaxi, you think about an innovative product or a progressive brand or buttoned down marketing. But an often overlooked secret is how these brands actually do their selling. The experience of buying from these brands online. These brands, along with millions of others, use Shopify.

All right, nobody does selling better than Shopify. It's home of the number one checkout on the planet and the not so secret secret which is ShopPay that boost conversions up to 50%. This means that way less carts go abandoned and way more sales get done. So if you're growing your business, your commerce platform better be ready to sell wherever your customers are scrolling or strolling.

On the web, in your store, in their feed, and everywhere in between. Businesses that sell more sell on Shopify. Upgrade your business and get the same checkout that Feastables or Thrive or Cotopaxi use when you use Shopify. Sign up for your $1 per month trial period at shopify.com/deep. But type that in all lowercase.

Go to shopify.com/deep all lowercase to upgrade your selling today, shopify.com/deep. All right, Jesse, let's go to our final segment. All right, for our final segment today, we want to do a triumphant return to my tech corner segment where we get into a technical topic that is relevant to the type of things we talk about today.

So I put on my computer science hat a little bit to help give us some more insight on topics relevant to living a deep life in a distracted world. All right, in today's tech corner, I want to talk about how do recommendation algorithms work. Why? Because, in part, this is really relevant to the ongoing discussion about social media and social media regulation.

So if we look at some of the new child safety legislation like COSA or COPA 2.0 or California's big law, we see that one of the things that they are pushing for is that when kids are using social media that we have to be careful about what it does recommend or not recommend.

Right? So you sort of see these arguments, okay, we're not talking about censoring information that can exist on social media, but we want to be careful about what we recommend or don't recommend. We also see this in discussions about things like Twitter or Twitter alternatives like Threads or Blue Sky, where there's often this notion of the recommendation algorithm can be tuned up or tuned down.

We saw this a lot with the discussions around Threads when it was released, that they were tuning down, they claim, the political nature of content and tuning up recommendations for others. There's this idea that there is a "algorithm" that is in charge of showing us stuff, and this algorithm is really important, and we can change this algorithm to change the experience or maybe strip it out altogether and have an experience without it.

It is at the core of many discussions around social media and its harm, so I thought we would talk about, well, how do these algorithms actually work? So what I'm going to do here is greatly simplify the idea of how a sort of machine learning-based optimization recommendation algorithm actually works.

I want to start by saying there is a spectrum on which these algorithms exist. So if we look in the social media ecosystem, on one end of the spectrum will be something like Twitter, which actually is relatively non-algorithmic. The way curation decisions are made on Twitter, a lot of it is actually cybernetic, which means it's based on individual humans' decisions to retweet or not, and when those are combined with the network structure of Twitter, which has power law dynamics, it's really good at sort of selecting for certain content to have explosive growth and start trending.

But it's largely non-algorithmic. It's actually just the aggregate of a lot of human decisions. On the other end of the spectrum is TikTok, which is essentially entirely algorithmic. It doesn't care who you follow or don't follow or what other people like. It just uses an algorithm to select what to show you next, then what to show you next, and what to show you next.

So we're going to be leaning more towards that TikTok side, where really it's like a computer is deciding, not other humans, what it is you should see. I'm going to give you a highly simplified way of thinking about this, then we can draw some conclusions from it afterwards. All right.

So let's pretend, for the sake of this example, that we're building an algorithm to recommend TikTok videos. And I am going to do a lot of drawing here, God help us. So if you're listening instead of watching, you might want to actually load up the YouTube version of this.

This is, what are we Jesse? This is episode 327. Is that right? Yep. All right. So you just go to, what, the deeplife.com/listen, look for episode 327. You'll see the video there. It usually comes up within the same day or the day after the episode lands. Okay. So we're TikTok, and we want to recommend videos to a user.

So we need ways, first of all, of describing the videos we have in our collection, and we want to describe them in a way computers can understand. So we want to use numbers. So let's start really simple. Let's say we're going to assign a single number to every video that we're going to use to help describe it.

All right. Hold on a second. Let me press again. There we go. All right. So now I can draw. All right. So we're going to have videos, we're going to describe each videos with a single number. Let's say, you know, this number, for example, here is going to describe for each video, the number of cats in that video.

And so we have this a single number on which we can categorize videos. I drew a line here, and we can imagine this as this is our space in which videos can fall. And I'm sort of adding numbers to this line. And in this very simple example, kind of numbered from, you know, zero up to eight.

We could use yellow dots in this example to be different videos, and we can just sort of place them on this line, depending on how many cats they have in them. So there's a couple videos with a lot of cats, and some five, a couple over here, and I don't know, we have fractional numbers of cats, whatever.

So we're describing all these videos by a single number. Okay. In this simple example, now let's let a user come along. And what we do is we want to look at the videos that you are looking at. And let's say we want to categorize them simply as a video you like or don't like.

So in like the Facebook days, there'd be an actual like button. The way we think TikTok works is that it actually looks at how long you watch a video. So if you quickly swipe to the next video, you don't like it. If you watch it long enough, then we can consider that you do like it.

So we're gonna start showing you videos, and we are going to start, let's say, let's just keep track of the videos you like, so the videos that you actually watch for a little while. And I'm gonna plot those on this same one-dimensional line here with a purple dot, and so maybe you like a couple videos with five cats, you like one with zero cats, six cats, there's a three-cat one, another six-cat, maybe one eight, maybe another couple more zero ones.

So I'm just keeping track of, okay, these videos you liked, where did they fall in this range of number of cats? Now after we've done this for a while, what I can do, and this is how these sort of basic algorithms work in a very simplistic way, I can say, okay, where on average, where on average are these videos you like falling on this single value I care about?

And there's different ways to do this. You can think about what we're trying to do here is basically find the average point, think about this as like we're trying to find a point that has like the best overall distance to all of your points, it's interesting, my controller is weird.

In reality, the way this is typically done is actually trying to minimize the average square of the distances, don't worry about that here. What I'm trying to do here is sort of find a point on here that's sort of in the middle, it's the average, it's minimizing distance to all of your likes.

So you have a bunch of zeros you like here, but you have like a bunch of fives, sixes and eights. So, you know, maybe your average is like right where that X is, that's kind of the center point of where the videos you like fall. So now when it comes time for me as TikTok to show you another video, what I can do to be smarter is say, great, I'm going to randomly select you a video from all the videos that exist, but I'm going to weight the probability that I select a given video depending on how close it is to this point that we said was kind of at the center of your preferences.

So in here, right, this point is like somewhere between four and five cats is kind of like the center of your preferences when we measure videos by cats. So you know, it's possible that I could select you a video out here, but I'm much more likely to select the videos around here.

So you're gonna get a lot of videos with like four or five cats and sometimes some videos with less cats and sometimes some videos with more. But pretty soon you're gonna be like, wow, this is eerie, TikTok has really figured out that there's kind of, I like videos that have, you know, like a couple armfuls of cats in them.

Again, this is simplified, but it roughly gets to how these type of things work. All right, so this is a single number. Of course, these videos are going to have more dimensions on which we're going to want to measure them. But that's okay, because the same thing works even as we go to more dimensions.

So maybe we say, okay, there's two numbers, let me select this here, maybe there's two numbers by which we want to describe all of our videos. So one number is the number of cats, and then the another number is like the number of skeletons. So we could just draw this if you're looking at the screen here as just like another axis, like now we're in a two dimensional space.

And again, we can do the same thing. All the videos fall somewhere. Every video has a spot somewhere in here. You know, a video with seven cats and one, two, three skeletons would show up right here in this space. A video with like zero cats and four skeletons might be over here.

And again, we see, we plot every time you like a video, we kind of plot it in this two dimensional space, and we can do the exact same thing we did before, where we find like roughly speaking where the center is by some sort of center metric, okay, roughly speaking, this is the center of all the videos you've liked.

And so now when we randomly select videos, they're going to be kind of roughly in this range, like you're gonna see a lot of stuff with a good number of cats and a fair number of skeletons. And like, you're very rarely going to see something with like a bunch of cats and a bunch of skeletons or no cats and you know, whatever, right?

We could do this with three numbers. Now we would be in three dimensional space and you could imagine there's regions where you have lots of videos you like in that three dimensional space. And when we randomly select videos, we select them near there, we can expend the number of numbers we use to describe these videos, they can get much larger.

And something like TikTok is going to have probably thousands of different numbers, each describing different parts of these videos. Now we can't draw this, once we get past three numbers, we can't really draw these in a way that makes sense to us, but the same mathematics works. So the videos are described by a ton of numbers.

We keep track of the videos you like. And then we can select for videos that are in some sense close to the clusters of videos that you like. All right, two complications here. What if you like multiple, there's, if we look in this region where you have a bunch of clustered videos you like, what if there's like multiple types of videos you like?

This just shows up as like multiple clusters, kind of like multiple clusters in this multidimensional space of videos that you like. We have ways of finding a bunch of different center points. We do things like k-means averaging. Okay, there's a bunch of different center points that each correspond to like a type of video that you like.

And so now when we randomly select a video to show you, we're giving extra probabilities to anything near one of these clusters, and the bigger the cluster, the more likely we are to show you a video from there. The other complication, well, how do we know if you like something that you've never seen before?

TikTok answers this by alternating between just purely showing you something weighted towards the things you like versus showing you something new. So it will opportunistically show you new things just to see, give you a chance to show a preference for things you've never seen before. That's why when you use TikTok at first, it kind of drifts over time until you finally stabilize into the clusters you like.

It'll show you a lot more random stuff at first to try to see what you like. It's like very roughly speaking, something like this is going on. All right, so here's some conclusions about this. These algorithms are automatic and agnostic to content details, right? It's not computer code where you can come in and it has in there political content, unsafe for kids content, sports content, and you can turn a knob, let's turn down politics and turn up sports content, or turn down controversial and turn up non-controversial.

It's agnostic to that. It has all these numbers, most of which, by the way, are figured out using embedding tools that are machine learning tools so that you don't know what they are in advance, right? You're not choosing what these numbers are. The software just figuring out what numbers matter.

That's just automatically plotting your videos that you like or don't like and finding these sort of center spaces in the space and randomly selecting. The algorithm has no idea what these spaces are from a content point of view. It's agnostic to that. It's selecting vectors that are weighted to be near other vectors that you've expressed preferences for.

So it can be remarkably effective. That's why when you purify these algorithms, like TikTok does, it seems eerie, like how did TikTok figure out that, you know, I like videos about, you know, bears working on crafts? But this type of exploration of the space and weighted selection will pretty quickly cluster these things together and the intersection clusters will have a lot of weight.

It will just automatically find these things. It seems very eerie, but it's actually quite simplistic mathematically what's happening underneath. But because it's automatic, they're not nearly as controllable as we think. Controlling these type of recommendation algorithms is difficult because of their automatic content agnostic nature. What we end up needing to do is things like human-in-the-loop dead zone definitions.

So we show a lot of content to real people and we say, "Here's the type of stuff we're worried about." And when they see the stuff that, to their human intuition, matches things we're worried about, they kind of hit a button. Okay, that's bad. That's bad. That's bad. And they create what you can think of as a like dislike plots in this space.

And then you can find the sort of centers of these spaces of stuff that people or testers said was bad, and you can reduce weight for videos near those. It could give you sort of negative probability weight if you're near one of those zones, right? But this is, again, it's kind of indirect.

It's not just you coming in saying, "Don't do this type of content." You have to have humans calling stuff bad, and that translates into this inscrutable multidimensional space, and it sort of affects the weight. So it's kind of an imperfect way of trying to tame this algorithm. Stuff that the human testers haven't seen or clicked on is going to be treated like anything else.

And so these algorithms, we just, we have to keep this in mind. Algorithms are automatic and mathematical and not easily tamable in a sort of human understandable way. So when thinking about reforms of these technologies, do not think about like a newspaper editor who's making decisions. You could just say, "Hey, do less of this." It's much more automatic than that.

It can give you like eerily successful results in terms of honing in on your interest, but it's also very hard to keep an algorithm like that successful. And somehow have it avoid lots of stuff, because it doesn't know what stuff means. Humans have to get in there, and it's messy at best.

So anyways, I hear a lot about algorithms. They're often discussed to be these like highly tunable, understandable things. They're simple algorithmically, but complicated in their effect and complicated to tame. So there you go, Jesse. We did philosophy and computer science in the same episode. We just kind of got our nerd bona fides up here, probably also lost half our audience.

You got a professor, guys. You have a professor podcasting. Sometimes you're gonna get some of that. Anyway, so thank you all for listening. We'll be back next week with another episode. It'll be a little bit less heady next week. We'll see. See what the feedback is. But until then, as always, stay deep.

Hey, if you liked today's discussion about the philosophical underpinnings of digital minimalism, I think you'll also like episode 298 about intentional information in which we go deep at understanding the role of information and human flourishing. Check it out. I think you'll like it. So today I'm going to argue that we misunderstand the impact of how we obtain information on the overall quality of our lives.