Description

C. Thi Nguyen (Website, Philpeople.org, X) is a professor of philosophy at the University of Utah focused on values, games, agency, art, aesthetics, and data.

His new book, The Score: How to Stop Playing Somebody Else's Game is out now. Thi is also the author of Games: Agency as Art, in which he explores how game designers work in the medium of agency, but sculpting a players abilities, goals, and obstacles to create "harmonious action."

I first learned about Thi's work via his interview with Ezra Klein, which is one of my all-time favorite podcast episodes. In it, he discusses Agency as Art, How Twitter Gamifies Communication, Why Q-Anon is game-like, and more.

The Score is a marriage of his work on games and on data and metrics. He explores how scoring systems in games allow for playfulness and agentic exploration of our values, while scoring systems in real life produce what he calls value capture. In an effort to make the world more quantified, comprehensible, and trustless, metrics are flattening our values and sapping the meaning out of our lives. One way he describes his work is that James C. Scott's Seeing Like a State also applies to the human soul.

In this conversation, I aimed to cover the most compelling ideas in the book in two parts.

First, we explore the local side: personal agency and values, attention and the difference between recognition and perception, process vs. outcome, and why playfulness and openness allow us to have richer lives. He also shares how games are a compelling template for this kind of exploration.

Second, we talk about the societal level: what we miss in a world of values dominated by what is easily measurable, how we can scale trust and enjoy the benefits of collaboration, science, and technology while not delegating our understanding to the wrong people, and why objectivity and truth are not always the same thing. Thi makes the case that technology is value-laden, not value-neutral, and that we must be more vigilant and nuanced in our approach to the ethical decisions that exist everywhere. I hope this conversation is a prompt for you and I to think more deeply about what we truly care about, to "move lightly" between agentic and value-laden worlds, and bring a perceptive playfulness to our lives. Remember, we are all grasshoppers in disguise. If you enjoy the episode, please support Thi's work and check out The Score.

Dialectic is presented by Notion. Notion is an AI-powered connected workspace where teams build their best work. Notion is also where I compile research for episodes and the home of my new site where you can find all links and transcripts. You can read more about why Notion embodies Dialectic’s values and our partnership announcement here.

Timestamps

0:00: Opening Highlights
1:39: Introduction to C. Thi Nguyen
5:13: Thanks to Notion
6:31: Start: What Does it Mean to Be Playful?
13:41: Starting Local: Agency, Scoring Systems, and Games
23:36: Value Capture: Incentives, Values, and the Collapse of Meaning
36:28: What is the Shape of Good Values?
49:45: Attention, Recognition vs. Perception, and Aesthetic Openness
58:46: Process vs. Outcome, Striving Play vs. Achievement Play, Recipe vs. Dish
1:10:00: Aesthetic Value & Autotelic Pursuits in Life
1:16:59: Metrics, "Measure What Matters," and What We Miss
1:24:16: Quantitative vs. Qualitative Ways of Knowing and Different Conceptions of Rules
1:38:01: Scaling Trust, Data, Experts, and Legibility
1:54:37: Objectivity & Truth, Value-Laden Technology & Decisions, and "Objectivity Laundering"
2:07:57: Advice for Technologists: Ethics, Maps, Value-Neutrality, and Playfulness
2:18:52: Closing Thanks to Notion

Links & References

Transcript

0:00: Opening Highlights

1:39: Introduction to C. Thi Nguyen

5:12: Thanks to Notion

6:31: Start: What Does it Mean to Be Playful?

Jackson: C. Thi Nguyen, it's a pleasure.

Thi: What's up? Welcome to my insane home.

Jackson: One of my favorite things about this is that I get to inhabit people’s worlds. First off, congrats on the book.

Thi: Thank you.

Jackson: In the future, when this is released, you will have successfully published the book. You have a bit early on in the book where you tell the story about how a student of yours influenced this book. It reminded me of a Kwame Appiah quote that I love: “In life, the challenge is not so much to figure out how best to play the game. The challenge is to figure out what game you’re playing.” What a great place to start. I want to take that last word because it’s such an important word that we’ll come back to a lot. My first question is simple: what does it mean to be playful?

Thi: This is a really good and deep question. We say that you play a game, but this is not true in other languages. The word play is always appended to games, and we assume they go together, that doing a game is always playing. The philosopher Bernard Suits pointed out that this is not always true. You can be playful and not be involved in a game. You can be at your job and starting to make up new stuff. You can be fucking around with new ways to approach your daily life or cooking. You can also play a game when it’s not really play; it’s work. If you hate it, or someone’s making you do it, or you used to be interested in poker and now you hate it, that’s not playful. That’s work, even though it’s a game. I’ve been trying to figure out what play means for ten years, and I don’t have a good single stable foundation. But there are two really good stabs. Bernard Suits says that being playful is redirecting normally useful resources to autotelic activities. Autotelic means valuable in itself. Normally I use this logical capacity to fix things or to argue with people, but then sometimes I do a puzzle. That’s just exercising the logical capacity for the sheer pleasure of doing it, often uselessly. This house is surrounded with weird skill toys like yo-yos and kendamas. There are balance and physical things you would do with your body that you would normally use for survival or getting food or getting from place to place. Now you’re just screwing around with it for the sheer joy of motion. That's Suits's sense.

Jackson: One thing I’m hearing is that it’s almost like a sacrifice of resources. Is that too strong of language?

Thi: That’s great. There’s a perspective by which the quick way to put what Suits is saying is that playing is taking normally useful resources and wasting them for fun.

Jackson: Wasting them for fun.

Thi: It’s only a waste if you thought that stuff was pointless. When you say that, you’re already implicitly adopting the attitude that clear outcomes, making stuff, creating goods, and accumulating resources are what’s actually valuable, and action for its own sake is useless. Suits actually thinks that is what’s important. It’s not actual waste from the philosophical perspective. The actual waste is if you spend all your time on useless activity and have a miserable life and a pile of goods at the end and you die unhappy. The actual fulfilling life is one where you found activity that’s valuable to you. To say that wasting useful stuff for fun makes sense is already using the language that is the trap.

Jackson: Waste is amazing. I love that you use that word because you have a visceral reaction to it. It's so not useful.

Thi: Right.

Jackson: That’s the first one. I think you said you had a second?

Thi: The other notion of playfulness that I find really useful is from Maria Lugones, the great feminist philosopher, in this beautiful paper called “Playfulness, World-Traveling, and the Loving Gaze.” She says that playfulness is the ability to move lightly between worlds. There are normative worlds of rules, landscape, and meaning. Sometimes you might be in the business world and you’re focused on profits, and sometimes you might be in the family world, and sometimes you might be in the artist world trying to be expressive. One way is to just inhabit one of these worlds permanently. Another thing you can do is to realize there are different worlds you can shift between and not be stuck in them. To be playful is to inhabit the world lightly and creatively, to both be able to move between them and also be able to change the rules of the world that your soul is inhabiting. Somewhere between those is some deep sense of what it is to be playful.

Jackson: What I like about the second one is it captures what you point at when you describe someone as playful. It’s this sort of dancing. You used the word creative very early on in your answer. I’m curious if creativity is just common in this idea of playfulness or something more paramount.

Thi: This is when the two of them separate. Bernard Suits is the philosopher that I learned the most about games from. You can see the big difference in that. If you’re hyper-focused on a sport, for example, like rock climbing, you are trying to perfect a movement and get exactly this movement exactly right. You get your body all in a line, and you might know exactly what you’re supposed to do. I know what the technique is. I know how I’m supposed to balance my hips, and I know I’m supposed to move my hands. I can see other people do it; I’m just trying to find it. In a sense, that version isn’t very creative, but it is refining my movement and taking actions for the sheer pleasure of it. That’s the case where they come a little bit apart.

13:41: Starting Local: Agency, Scoring Systems, and Games

Jackson: The way you lay out your ideas in the book is quite compelling. I wanted to attack them from two vantage points: the personal and the local, and then the global. The name of the book is The Score. Foundational to both sides of this is the notion that we have scoring systems all around us. One thing I have been thinking about regarding playfulness is that we live in this highly gamified world, and yet we are decreasingly playful. Scoring systems tell us what to desire. Challenges in games give us freedom, while in the real world, they constrain us. To start on the personal and local front, I want to start with agency and talk about agency in the context of value. There is a central theme here about starting with “Who am I?”—explore, then exploit. You have a Carol Rovane quote about agents that I liked: “An agent is some entity that considers reasons, makes choices based on those reasons, and acts.” If you change the reasons that you act on, you change your agency. This is notably not close to the popular conception of agency in the technology world. There, agency means you can just do things; it is actually less about the reasons behind what you do and more about just doing. You focus much more on general exploration. I am curious what you think about that idea of agency.

Thi: Can I take a long-ass step back and do a run-up to this? This is going to be full of weird details that you may or may not care about, but I will give you my personal history of working with the concept of agency. It comes from trying to figure out games. The reason I started working on games—which is not a topic philosophers are usually allowed to work on—was because I was teaching a philosophy of art class and wanted to do a case study on whether video games are art. A lot of the material I found emphasized that games are art because they are like movies. They have dialogue, scripts, and characters. They would celebrate the games that were the least like a game, where you had the least freedom and were locked into cinematic cutscenes. It felt like someone was clutching for a familiar sense of importance and ignoring play, freedom, or choice. I would read a 300-page book on the art of games and never hear those words. I was clutching around for it. There was a moment—as with a lot of other moments, I was pretty drunk—where I realized what games are. They are like governments; they are rule systems, but for fun. Games are art-governments. Reiner Knizia, my favorite game designer, has a moment in a talk where he says the most important thing in his game design toolbox is the scoring system. It tells the players what to care about and what to want in the game. To a game player, that seems completely obvious. You look at a rulebook, and it tells you not only what you are trying to collect, but it also tells you your basic conception of victory. If we open a board game, we might find out that we are trying to kill each other, or we might find out that we are cooperating and victory is shared. We willingly opt in. What I ended up trying to articulate was that the heart of games was not that they told stories, though they can. The heart of games was that they shaped your actions so that actions, decisions, and stories came out of you. Games are the art of agency; they work in the medium of agency. I realized people often misunderstand this because agency just means activity and freedom to them. That is not a good conception of games. Some games are incredibly good because they give you enormous agency, but some are really good because they hyper-constrict your agency. Soccer is interesting because it takes away your hands. Tetris is interesting because it is like limit poker. At each stage, you are trying to do so much with such a tiny action space. That is actually what is beautiful about it. Reiner Knizia borrows from poker and has an incredible game called Ra. You are trying to affect people’s incentive structure, but you only have three coins, and you get to bid on them or pass. You are trying to do so much with so little. I realized I was using an older notion of agency. Carol Rovane’s quote is a natural version of this. We see this in terms like literary agents, legal agents, or AI agents. Exactly. For you to have a real estate agent or a legal agent means that when they are performing their job, they are acting on your reasons, not theirs. They are shifting the reason structure. At first, this might look weird, like a bizarre one-off thing you do in games. But Carol Rovane points out that we do this constantly. Her example is a search committee trying to hire someone. I am not usually using my own reasons. I am not supposed to hire who I like, who I want to hang out with, or who would be good for my projects. As a department, we decide what we care about. When I am on the search committee, if I am acting really in that role, I look at those reasons, and I switch. In games, we cancel out a lot of our reasons. If my spouse and I are playing a game, normally, my standard reasons are that I love and support her. In that game, we are going to try to kill each other. We do this all the time. For example, this will perhaps be embarrassing and revealing, but I think it is very true—when I started teaching, I brought my full human self to the classroom. This meant I was kinder and more open to students who were of my tribe, who had my sense of humor, who had my politics, or who liked the same music as I did. That's a bad teacher. When you are in a particular role, you cancel out a lot of reasons. Imagine going to a government office and having the person treat you as a full person, giving you better treatment because they like you. You don't want that. John Dewey, the great American philosopher, says that in a lot of the arts, we take something we do in normal life and we concentrate it and crystallize it for beauty or interest for its own sake. I think there’s this thing that we do always in normal life, where we enter a role, we cancel out some of our reasons, and we change the reasons we act on. In games, we do the same shit for fun or richness.

Jackson: That makes us see the thing that maybe we were typically subject to. We're seeing the water a little bit when we play games.

Thi: We become more aware of it.

Jackson: Usually, we're not thinking about the ways we're taking on different forms of agency in most of our daily lives. Most people's perception would be that, of course, they are going to be a little different when they are at work, but they haven't fully considered it. I would argue many of the people who are most successful in modern society are the ones who are really good at this. Maybe it's sociopathic.

Thi: There are two different things: being really good at role shifting versus being willing to completely transform yourself and get yourself stuck in a role forever. One worry might be that there's playfully being able to shift into different roles. There is another thing that this world might reward, which is psychopathically committing yourself to a hyper-simplified role and never being playful with it.

23:37: Value Capture: Incentives, Values, and the Collapse of Meaning

Jackson: We will get to that later. On this note of agency and values, because I think it’s critical, one conception of this would be that games can make it easy to act. This is the “you can just do things” form of agency. The other is that they make it easy to explore the possibility space of actions.

Thi: One thing that is important here is to keep in mind the two levels of agency involved in games. There is one kind of agency in which a specific game provides you with one particular form of agency, fixes it, and you plunge into it. During the game, you suddenly become a being of only balance, calculation, or deceit and lying. Each game focuses you. Then there’s another tier of agency, the freeing sense or the exploring sense where games as a whole let you move between them. They give you the freedom to have a choice. In a lot of my work life, I don’t have this choice. After this interview, I’m going to be grading. I don’t have any choices about this; there is a kind of agency that’s going to be pushed into me. But for my break, I can choose: am I going to play with yo-yos? Am I going to run around the block? Am I going to have a quick game of online chess? Each of these is a completely different kind of action and a completely different feel of action. I get choice of them because of the enclosed nature of games and because the clarity makes that role shifting easier. The hyper-clarity of the rules and the points makes the role shifting easier.

Jackson: This is pointing to something important. What you just described is part of the difference between incentives and values, which are two very different kinds of motivators. The first one is “I have to grade.” The “you can just do things” form of agency is not acting based on incentives; it’s acting in the world of value. Incentives can provide some motivation, but they don’t change your core values. Why do you think modernity tempts us so much towards incentives rather than acting based on values?

Thi: I’m not sure that’s the right formula for me because my worry is about when incentives become values. There are always going to be incentives. A value, roughly, is whatever is your core motivator and the ultimate guide of your actions. Whatever sets your choices and how you’re going to change yourself, your values are where all this springs from. Incentives are things the world gives you where it says, “If you do this, you’ll get those resources.” There is one way that you can keep the world at arm’s length. You can say the world is giving me certain incentives. It’s saying I have to do things this way to make money, or for people to listen to me, or to get enough people to tune into this show. Those are the incentives. The reason I’m entangling with them—my reason for gathering those resources—comes out of my actual values. The thing I’m worried about is when they collapse and we suddenly forget to think about what our values are beyond the incentives. This is the simplest damn thing. One frame of mind is to make enough money to do what you actually want. The other frame of mind is to say, “Here is the scoring system. I’m just going to max out on making money,” without thinking about the thing it’s for. That difference should be familiar to everyone. That’s the difference between having a firewall between your true values and your incentives and letting them be collapsed.

Jackson: I think maybe I was making too big of an assumption there. When I said the difference between incentives and values, I was thinking of values as the things we do despite clear incentives. What do you think of willpower?

Thi: Tell me what you mean by that question.

Jackson: Do you know who David Goggins is? He is the ultra-marathoner. He is the guy who says to wake up at 4:00 AM and carry the logs. One of my jokes is that I don’t really believe in willpower. I think willpower is this grittiness, despite maybe even clear incentives, that says I’m going to do something hard because in the long run, it’s going to pay off. I’ve always related willpower to the idea that the people who require it are the ones who haven’t found a way to harmonize their real values with what the world wants.

Thi: That’s pretty romantic. I get to be a philosopher, but let me tell you, I have to grade ten more papers tonight, and that’s going to need a lot of willpower. Let me try something that might be interesting to you. One of the core ideas of my book is value capture. Value capture is what happens when you have values that are rich, subtle, and developing, and then you are put in an institution or a social setting that feeds you clear, simplified versions of those values, like a metric.

Jackson: The incentives and the values you were just talking about collapsing.

Thi: They take over. Value capture is what happens when you go to school out of an interest in learning and you get focused on GPA. It’s what happens when you go on social media to connect to people and you get focused on likes. Or you start a podcast to get ideas, and then you become focused just on subscriber count—not as a means to communicate what you really want, but changing what you’re trying to say to just max out your subscriber count. I’ve been worried about value capture for a while, and I’ve been trying to figure out what’s wrong with it. One thing you might think is that the wrong of value capture is that you’re losing control of your values, that the world is forcing your values out of something you freely choose. I don’t think that’s actually the right account. If that were the right account, then if someone were forced and brainwashed into putting on a watch and just caring about their steps, or forced to pay attention to their BMI instead of their health, that would be a problem. But if someone freely chose and devoted themselves to weight loss over all else, that would be fine because it was a matter of choice. I think actually one of the most worrisome things is cases where people enthusiastically and freely choose to be value captured. They embrace it and they lose the plot because, in some ways, it’s easier. The world is giving you a quick, easy tracker and everyone else understands it. I’m going to do some abstract philosophy now that I think will be interesting to you. One conception of what human well-being is involves flourishing and what makes a good human life. We have to talk about that because what we’re talking about is how oversimplified values screw that up.

Jackson: That is part of the reason I wanted to set the table on the personal side. We’re going to talk more about all the global problems we have, but at the root, these things are intuitive, even if we forget them. The point of living is to live a meaningful, flourishing life. We can all agree on that.

Thi: There are so many things I want to say, but here is one quick way to do it. I’m worried about someone capturing your definition of meaning. This is why the willpower question is so interesting to me. If the world can change your sense of meaning, it can reorient your will. It can make you send all your willpower and all your grit towards something simplified. Let me give you a bit of technical background that isn’t in the book because it was a little too technical, but you might find it interesting. There’s a standard view in a lot of economics and rational choice theory that guides social policy. It says that people’s well-being is just the satisfaction of their expressed preferences and desires. Whatever they say they desire, give them what they want and they will be good. The view is that we can’t be intrusive; we have to let people be autonomous and listen to what they want.

Jackson: This is like neoliberal late-stage capitalism.

Thi: This is also underneath all of economic theory and a lot of pxolitical science. Very progressive, positive, “help the world” political scientists often work with this view that well-being is satisfying people’s expressed desires.

Jackson: It is also about not being paternalistic.

Thi: That’s true. It respects people’s autonomy and it is non-paternalistic if you just listen to what people want. One of the most interesting criticisms of this came out of 1980s and 90s feminist theory regarding a worry about what are called adaptive preferences. Psychologically, a lot of this comes from a sociologist named Jon Elster. If I start to limit your opportunities, your preferences will adapt to the opportunities that are available to you.

Jackson: I’ll lower my bar.

Thi: You’ll lower your bar. Exactly. It turns out that if systemically women can’t find work outside the house and they’re not permitted in the workplace, then people will adapt their desires to domesticity. Here is the worry: we create a world where half the people are not allowed to work. They adapt their desires to fit the limited situation we have. According to desire preference satisfaction theory, we’ve won. We’ve succeeded, and the world is great.

Jackson: Then you have a falling knife.

Thi: What do you mean?

Jackson: If you wanted to paint a really cynical vision of the world, it is that the scenario you just described just seeps into more and more of it.

Thi: Yes, exactly. I just did this for my students. If your view is that a good world is one in which people’s expressed desires match their situation, imagine a case where most people get $20,000 a year but their desires are to have $60,000. There are two ways to satisfy everyone’s preferences: make the world better or get people to decrease their preferences. This is the background of the worry. It might look great if we just get people to collapse their values into incentives, because then they will have everything they want. If they want something simple and straightforward that the world can provide for them, then things are great. But I have been convinced that desire preference satisfaction theories of well-being aren’t right. This also means that strict autonomy about values isn’t right—that you can have the wrong values. I’m worried about the systems by which you might be convinced to fully commit yourself to very simplified values that diminish yourself.

Jackson: Yes. And it’s reflexive.

Thi: Are we in the ballpark of what you want to talk about?

Jackson: You said that we might have the wrong values. This is a thorny question. It’s clear that we don’t want flattened values. What is the shape or the texture, directionally, of good values?

Thi: This is the most interesting question. There is one way I can imagine this conversation going in which you ask this question and someone else would try to give a specification of what the right values are. That is not the way this conversation can go. I have been convinced that the right values for a particular person are incredibly dependent on details about the particular person, their particular context, their particular psychological profile, and their particular place in the world. Elijah Milgram, a philosopher who has been really influential to me, has this view that you don’t calculate the right values for you from the top down by thinking about some abstract conception of the good and then deducing it. You have to try them out and see if they work for you. We get these signals. If you have a value that works for you, you thrive. If you have a value that doesn’t fit you in your situation, you fall apart. Some of it is dependent on your personality, but some of it is dependent on the place you are in. One of my favorite examples comes from Jane Jacobs in “The Death and Life of Great American Cities.” She talks about how a lot of people will come from rural areas to New York. In rural areas, they come in with a value where they highly value friendliness, making eye contact, and chatting with everyone they meet. Then they come to New York and everyone seems like an asshole because no one’s making eye contact. If you try to talk someone up in the subway who has their headphones on, they will bite your head off. After you spend some time there, you realize the value of friendliness does not work when you’re in the subway with people all day long. The big city is so packed that you need to actually deeply value respecting other people’s privacy.

Jackson: Values like an evolutionary fitness for that environment.

Thi: Exactly. It’s a fitness. It’s not that one of these values is better than the other; it’s that some values are suited for some contexts and some values are suited for others. Cross-multiply the context sensitivity of values with the fact that different values will be good for different personality types.

Thi: The view is that values should be tailored to you. This is not the same view as having the freedom to pick whatever values you want. It’s that your values need to be carefully tailored to your environment and place, often using the particular signals of your emotions and how you feel as a guide.

Jackson: I’m glad you answered this way. When I ask what the shape of good values is, I think it is inside that last statement. So much of this is running against a broad feeling that we have a meaning crisis, especially for young people. Your central contention is that one of the main reasons is people need to choose their own values. The history of the world is actually that values are issued to you top-down. Not necessarily in this postmodern flattening of metrics and incentives, but in how the state tells you what to value, or the church tells you what to value, or your family tells you what to value. I’m feeling some tension. It’s really profound, and maybe it’s because I’m a young person in 2025. I tend to agree that we should choose our own values, and yet that doesn’t mean you should just choose any values. There are good values.

Thi: You're getting a particularly wonky, techy version of this.

Jackson: Just what I was looking for.

Thi: We should be really careful about the difference between choosing your own values and tailoring them to the specific context. It’s not just a matter of choice.

Thi: A lot of people want to collapse this to a distinction between objective and subjective. If they're objective, they're universal, and if they're subjective, they're just you and your feelings.

Jackson: Almost like morals and feelings.

Thi: I’m saying you can have better and worse values. You can get the wrong values, but whatever the right ones are, they are deeply tailored to what you are. It’s less about choice and more about sensitive detection. It’s a mixture of both invention and listening. Elijah Milgram has a beautiful paper called “On Being Bored out of One’s Mind.” He says you might have a theory that this is the right value for you. You go to grad school and think you’re going to do this thing, and then you’re just miserable. That misery is a detection system.

Jackson: Then you start relying on willpower.

Thi: Then you start cranking through. At some point, you have to listen to the signal that this is a terrible value. You are not flourishing.

Jackson: Yes.

Thi: What was the question?

Jackson: I think what I’m trying to square is values from the outside.

Thi: One of the reasons I really like this kind of feminist 90s value literature is that before this, there’s a fantasy inherited from existential philosophy that your authentic values come from you and values from the outside are alienating and terrible. That can’t be right. We soak up our values.

36:28: What is the Shape of Good Values?

Jackson: That’s the “choose your values” idea: sit in a cave and choose your values.

Thi: Like pure choice. Philosopher-economist Aurel Kolnai says that some conceptions of value are too heavy and thick, and don’t have any freedom in them at all. But then the existential conception of value is too thin. It’s just pure choice. Ignore the world and just make something up? That can’t be it. We get our values all the time. A lot of the way we learn to value things is by learning from other people who guide us through activities. This is a lot of what games do. You show up and start doing some weird new activity like climbing. You think climbing is for getting ripped and getting a workout. Then someone tells you that you’re climbing really brutally and to try a little more sensitivity in your feet. You suddenly learn that so much of the beauty on offer and the value in this activity is something you never realized before. There could be delicate poetry in your own movement. I learned that from other people. Pure existentialism of that kind doesn’t have room for learning from other people.

Jackson: It’s more top-down, too. To go back to Jane Jacobs or Christopher Alexander, these ideas are about finding it through emergence and experience.

Thi: There are two extremes. Pure top-down is where the world tells you your values and you take them on. That can’t be right. Pure bottom-up is almost like an immaculate conception. In between is something about learning and negotiation. One of the differences is that when they’re not fixed from the top down, you receive a lot of value candidates and proposals from the world. You can balance them and reinterpret them for yourself. The world may tell you—it’s a bit like taste.

Jackson: It’s a bit like taste.

Thi: Explain what you mean.

Jackson: Taste has been a big discussion and something I’ve talked about with people. People in technology figured out taste mattered a year or two ago. The conception of taste is often just talked about as judgment—judgment in isolation. He has good taste because he knows what things are good. Taste, intuitively, is also something we forget: it’s eating a lot of food, first and foremost. It’s a lot of inputs. Perhaps what games can do—and other things might do, too—is allow us to refine our taste and our values.

Thi: I see where you’re going. Taste is great in the following way: when you’re developing taste in something, going it completely by yourself isn’t going to get you there, and accepting external authority’s taste isn’t going to get you there either. When you learn about something like jazz, you listen, you learn, and you let people point things out. Then you slowly start to find your own way and refine your own tastes. You do it through intense exposure and careful attention to many examples. If the world forces a very specific value conception on you rigidly, that is not good for human beings. But something that often happens, which I think we’re losing in the metrified world, is more open. The world might give you a bunch of values—honor, courage, loyalty, family, community—but it doesn’t tell you exactly how to apply those concepts. It doesn’t give you the precise borders, and it doesn’t tell you how they count against each other. Even if the world is generally communicating values to you, you have a lot of freedom in interpreting particular open-ended terms and finding balances. That’s very different from saying, “I will value a higher subscriber count.” That is a non-interpretable, non-plural, unidimensional thing that has no free play in it. That is really different.

Jackson: You said precise values embody a closed-minded spirit about what’s important in the world. Conversely, you write about poets using a meaningful inarticulateness and how values can have these imprecise edges. How do we move toward these fuzzy values? How do we sit with them?

Thi: They’ve been there all along; you just have to let them in. We’re basically talking about the value of art. This is what poetry is like. The philosopher Elisabeth Camp has a beautiful set of essays about what metaphors are. She says metaphors are when you know this thing is like that thing, but you don’t know exactly how. You’re gesturing roughly, saying, “I’m not sure, but somehow your soul is like the ocean.” It’s a way of pointing at something without defining its edges. It’s common to think the more precise the language, the better. There are two cases where this falls apart. One is when the world is actually vague and fuzzy at its boundaries. In debates about whether something is a sandwich, a taco, or a hot dog, the answer is that those aren’t well-bordered concepts. We have to accept their fuzziness and not pretend we know things. The other case where we get fuzziness is when we’re uncertain and we want to express and mark that uncertainty. We remind ourselves we’re uncertain and we want to sit in that uncertainty.

Jackson: We want to sit in the uncertainty.

Thi: A paper I’m writing right now is about the term “vibe.” I think “good vibes” is very openly fuzzy. It’s like there’s something good going on and I can’t put my finger on it.

Jackson: I think “auras” was the word last year, right? It’s interesting that society, or at least young people, seem to be trending toward words that are like the finger pointing at the moon.

Thi: New slang terms are often people trying to find language to express something important. The rise of “vibe” and “aura” is people wanting to point to the need to sit in unclarity.

Jackson: And it’s not because it’s not real. It’s actually very real.

Thi: It's very real.

49:45: Attention, Recognition vs. Perception, and Aesthetic Openness

Jackson: This leads into one of the earlier conceptions we were talking about: agency. Your central idea of games is that they allow agential fluidity. We humans have an enormous capacity for agential fluidity. The board game Imperial trained me in a mode of getting people to do what you want by giving them a piece of the action. It wasn’t just training a technical skill; it gave me a whole outlook and a whole attentional focus.

Thi: I love that.

Jackson: Why does this begin with attention?

Thi: Attention and value are so interlinked. What you value is what you pay attention to. One of the core ways games work is that the scoring system guides what you’re trying to do, which deeply guides your attention. Setting the scoring system is a way for a game designer to sculpt your attention. In rock climbing, the goal is to go up and the rules are not to use any rope. Suddenly, your attention has to be on tiny details of the rock and the way you balance.

Jackson: Details you’ve never paid any attention to. You never looked at that crevice before.

Thi: You never looked at it. It’s like an amplifier. If you’re inattentive to your balance on the rock, you’ll fall. That game is constantly not only telling you what to pay attention to, but refining your attention by slapping you over and over again.

Jackson: You were talking about this earlier with the yo-yo. It is a great way to lock in.

Thi: My spouse is very good at attending to herself. I am not. Godfrey Devereaux, one of my favorite yoga writers, says something that I think got cut out of the book.

Jackson: It might be in it.

Thi: Oh, yeah?

Jackson: Unless there’s another yoga part.

Thi: I think this is different from the yoga part in the book. Godfrey Devereaux says the reason we do hard yoga poses is because it is a tool for meditation that amplifies a wandering focus. If you are trying to meditate seated and your focus wanders, you won’t notice because your focus has wandered. But if you are in a hard pose and your focus wanders, suddenly you wobble and you feel it. A lot of games take this approach. There are deception games where all the normal board play is taken away from you, and all you have is to stare at someone in the face and try to read them. There are other deception games where you are not even allowed to look or talk to other people. In a game like The Mind, you are moving stuff around the board and trying to signal and deceive people. Each of these things focuses your attention on one particular modality of the world and refines it. The downside is if the world or my university says what matters is student graduation rates. If that is what we measure, and we don’t measure happiness, wisdom, or ethical growth, then the entire institutional attention becomes hyper-focused only on those features that immediately produce a measurable outcome.

Jackson: Lock in.

Thi: Lock in is such a useful term. Scoring systems can give you the most beautiful parts of lock in and the most soul-deadening, society-destroying parts of lock in.

Jackson: What’s the difference between recognition and perception?

Thi: Thanks for hitting that one.

Jackson: This is obviously very related.

Thi: This is from Dewey.

Jackson: My favorite bit is where you say “we recognize and categorize something, then we stop. Perception keeps going.”

Thi: A lot of what this is about is how we categorize things and the clarity of our categorization system. The whole point of a fuzzy term like “interesting” or “rich” is that you have to ask yourself whether an activity fits. You have to deliberate, fight, and interpret. Other terms associated with metrics are very mechanistic and easy to apply, so it is very easy to stop. The example I have been thinking about is screen time. If I target that my child should be involved in creative or interesting apps, I have to make a complex decision. With screen time, it is automatic. Screen time is a lousy category. It lumps together my kid coding on Minecraft with the worst YouTube shorts ever. The difference between recognition and perception is that in recognition, you apply a category and then stop looking at the thing. The category is the end of thought. With perception, you apply the category and that helps you look at it. You keep looking further and further. I have known people where if you do something odd, they look for an explanation for it. When they can put a name to it, they just stop thinking. They decide you are just goofing off or that you are on vacation. They have categorized it and they are done. Perception is the ability to look at something, have a name for it, put it under a category, and then keep looking at its peculiarities and differences.

Jackson: And really see it.

Thi: The problem of recognition is that categories are very abstract. It assumes that when you say someone is a coder, a business person, or a Gen Zer, that is all the information you need to know.

Jackson: There’s no more information here.

Thi: It stops at a very specific level of abstraction, typically a level that has been established socially. This whole book is about the puzzle of why scoring systems are fun in games and so soul-deadening in institutions. In games, we often have a playful, exploratory attitude where we keep looking. The big interesting puzzle is why metrics in large-scale institutions promote this closed, dismissive vision that categorizes things very quickly and then stops looking.

Jackson: Playfulness and perception are maybe on some kind of continuum or similarity of vector space.

Thi: Let me vomit a bunch of associations at you because this is super interesting. The modernization I really like is Jerome Stolnitz’s. Stolnitz says that in normal life we have a very practical vision. We have a goal and we go through the world and classify things according to our goal. We stop looking at features of things that aren’t immediately related to our goal. It is a very filtered vision.

Jackson: It’s evolutionarily…

Thi: I think that gives animals a bad rap. That is very corporate.

Jackson: I’m thinking of survivalist behavior.

Thi: It is very survivalist, but it is not necessarily natural. I suspect that many animals play far more than many modern humans.

Jackson: That’s a good check.

Thi: Stolnitz says that aesthetic vision is vision that is stepped back from purposes and goals. It is a tension that roves freely over all the features of everything, open to whatever there might be to find. He says that aesthetic vision is where you let the object take the lead and show you what there is to love about it instead of having your own goal. The more hyper-clear the goal and focus ahead of time, the more your vision seems to be closed, dismissive, and non-exploratory. The more you pre-specified what’s important to you. There is this other thing—call it art vision, aesthetic vision, beauty orientation, or play—where one of the characteristics seems to be an openness of vision to new value.

Jackson: To be surprised!

Thi: Yeah, to be surprised. Why the fuck not? Fuck around and find out. But good.

58:46: Process vs. Outcome, Striving Play vs. Achievement Play, Recipe vs. Dish

Jackson: What a nice lead into a critical part of what you talk about: how games allow us to explore this fluidity through process and outcome. A couple of things from you: “In games, the value of the outcome is inseparable from the value of the process. In normal life, we struggle to attain some goal that we really want. In striving play, we adopt a goal to get the struggle we really want.” And then: “When I started trying to exercise, I had only the barest conception of what it could do for me. I didn’t realize how much else I would find in it.” What is the link between process and values? What about process allows us to fluidly explore values?

Thi: This harkens back to something we said at the very beginning of the conversation. I’m worried that already adopts a modern frame that has crapped on processes too much. Let me take a few steps so I can take a running start again. One of the core ideas animating this book is Bernard Suits’ definition of a game. For Suits, to play a game is to voluntarily take on unnecessary obstacles to create the possibility of striving to overcome them. In the quote that you read, Suits says the struggle and the constraints are an intrinsic part of the goal.

Jackson: It’s not the end themselves.

Thi: There’s a little wrinkle there. Suits says in practical life, what we care about is the outcome in and of itself. If I’m trying to get to a particular spot in the city, I just want to get there. Any method will do. If I’m running a marathon, I have to get there by prescribed means. I’m not allowed to take a taxi or a shortcut. It only counts as crossing the finish line if I did it a particular way.

Jackson: You have a great line in there where you say “cheating in a cosmic sense,” which is so good.

Thi: In games, there’s an incredibly powerful connection between the struggle and the goal. That’s stage one. I’ve added my own little wrinkle to this. In games, there are two kinds of play: achievement play and striving play. Achievement play is playing for the value of winning. Striving play is playing for the value of the process. An achievement player in a game still cares about the struggle. They want to have won that struggle.

Jackson: And they wouldn’t cheat because it would disqualify the achievement.

Thi: Someone who just wants to win the marathon is not going to take a taxi because then they wouldn’t have won the marathon. But they still want to win. The striving player is the person that wants to win for the sake of the process.

Jackson: Winning is actually a constraint that has to exist for the process to be good.

Thi: I think about how I play board games with my spouse. We are good at very different things in board games. Once in a while we’ll find a board game where we’re matched and have a delicious struggle. If I find a strategy guide afterwards, the argument from the book is that if I’m an achievement player, there’s only one logical thing to do: read the guide and win.

Jackson: Crusher.

Thi: But I don’t, because that would make the game boring. Galactically, I’m avoiding making moves in life that would make it more likely to win. During the game, I have to try all out to win to have fun. I don’t care about winning because I’m not reading the strategy guide. In order to have the process I want, I have to try really hard.

Jackson: You’re taking on a form of agency.

Thi: I am adopting a win-oriented agency to have the pleasure of an interesting struggle. To go back to your original question about what helps us explore values, the process is where all the values were in the first place. It’s only this weird delusion…

Jackson: That there are values without process.

Thi: This is from Aristotle. Bernard Suits, the game scholar I love, was an Aristotle scholar and attributes his notion of games openly to Aristotle. Aristotle says the value in human life comes from the rich exercise of our capacities: our intellect, our body, and our social capacities. Outcomes, tools, and resources are useful insofar as they allow the exercise of the capacities. In many cases, we are aiming at an outcome, but the valuable thing is the activity and the process. I have theories about why we have been persuaded that processes are unimportant and that only countable outcomes matter. We focus on the products we make, the money we make, and the measurements applied to the scale. If I wanted to rack up the numbers of success in games, I would play intellectual games because I am physically mediocre. I’m a terrible climber for how long I’ve been trying to climb. I cannot justify it in any outcomes-oriented way. The only way to justify it is to think this is valuable because I get to be moving. The process of skill refinement, even if I never end up more than mediocre, is valuable in and of itself. My response to why that’s the place to explore value is that’s where the value is in the first place. How did we get to think that you had to look in piles of things that you made for the value?

Jackson: One of my favorite examples you use is the difference between a recipe and a dish. I think this comes from John Thorne. Can you talk about that briefly for the listener?

Thi: John Thorne, one of my favorite food writers, was a philosophy major as an undergrad.

Jackson: There’s some kind of connection.

Thi: Thorne has this marvelous distinction between a recipe and a dish. He says a recipe is a dead thing, a writing down of how something was made by someone once. A dish is a live thing, an idea of balance in a creative cook’s head that gets remade anew each time. Creative cooks who are making dishes are typically better cooks. A lot of people in my generation and your generation mostly cook from recipes, and we wonder why our parents’ generation was better. It’s because they don’t have a fixed recipe; they are adapting.

Jackson: You have a great line in the book where you say you were mad that your mom wasn’t giving you a “real recipe.”

Thi: She was giving me this thing where she said, “You have to taste this and you have to adjust. You have to taste the pineapple and see if it’s sweet or sour today and adjust your vinegar and sugar.” I thought, “This isn’t real cooking.”

Jackson: Give me the answer, Mom.

Thi: Give me the recipe that I can follow. Of course, now that I cook, I realize that’s the real thing. The recipe is not the real thing. I think the reason this is related to processed foods is the interest people have in the perfect cookbook with the perfect recipe for the perfect dish. What that often has you doing is following something precisely and measuring things precisely. It’s been engineered so you don’t have to make decisions. You often get a good outcome, but it’s a very fixed outcome.

Jackson: I think there’s something profound in that sentence that applies to much of what we’ve been talking about today.

Thi: Yes, exactly.

Jackson: That’s the tradeoff. It would be nice to have somebody make the decisions for me to simplify things, to reduce the complexity, and be able to get a pretty good outcome.

Thi: Let’s think about what you get and what you don’t get. For me, cooking is the perfect metaphor. Here is what you get from a clear, mechanical recipe: it’s easy, you don’t have to make decisions, and you don’t need any experience. If you follow it correctly, you will get a pretty good, semi-reliable result. The more the ingredients are fixed, the more it will be reliable. The more you use the same standardized, stable canned tomatoes and flour, the more you’ll get a reliable outcome, but it will be the same one. When you cook from a dish, here’s what you get. A: It will take a while to learn. B: You will slowly develop the capacity to see the decision space and move around it to see when you can make it crispier, saltier, or softer. C: You’ll be able to adapt to changing ingredients and roll with the fact that the tomatoes today are really sour and not sweet. You know how to compensate for that. And four (D): instead of being in the mode of rigidly following someone else’s rules, you’ll be constantly engaged in your senses and making decisions. You’ll be tasting and thinking, “That’s good, but it could use a little more.” You’ll be engaged in that process.

Jackson: You might enjoy the cooking.

Thi: You might enjoy the cooking.

Jackson: When I was hearing you list the pros and cons of the first approach, I thought about how one of the most valuable parts of the recipe is that you’re way less likely to waste the food and your time. It goes back to what we talked about earlier. We’re so afraid to waste.

Thi: Yeah. I love that. There’s a perspective from which you’ll get it right each time, and you won’t waste something. But here’s another perspective: you could have had a lovely hour being engaged, making decisions, and being free with your senses while tailoring something to yourself. Instead, you spend an hour following someone else’s rules like a robot.

Jackson: You don’t get to cook. You don’t “have to” cook; you “get to” cook.

Thi: John Thorne says we have persuaded ourselves to adopt a restaurant mindset where we’ve turned ourselves into our own menial laborers. We go into the kitchen, close the doors, and follow the recipe precisely in order to have a guaranteed good outcome. Instead, we could invite our friend into the kitchen, cook together, get a little drunk together, make decisions together, and taste together. The outcome might not have been perfect…

Jackson: But we lived.

Thi: But you lived! You did the thing with the person.

Jackson: You were there.

Thi: We have somehow been persuaded that the value isn't there.

1:10:00: Aesthetic Value & Autotelic Pursuits in Life

Jackson: We’ve covered this a bit, but I wanted to hit it briefly. You talk about object beauty and process beauty. You say “in games the beauty shows up not in the game, but in the player. Game designers work a step back. They shape the general contours of our action, not the precise details.” I know you’ve also done a lot of work on aesthetics broadly. How do we underrate the virtues of aesthetics?

Thi: With games, it’s doubly underrated. We are underrating aesthetics, and we’re underrating the aesthetics of doing instead of objects. We live in a world where people who are really good at optimizing industrial processes make a ton of money, while people who have soaked their souls into making beautiful, moving, emotional, personal indie comics can barely survive. As a world, we underrate it. My suspicion is that aesthetic value is one of the harder things to count in an objective way. There’s not a good metric for aesthetic value, and I don’t think that’s an accident. Whatever aesthetic value is, it is by its nature subtle and variable.

Jackson: And it’s also not recognition; it’s perception.

Thi: Aesthetic value is about the value of perception. The thing about perception is that it is slow as fuck. If you want to be really efficient at hitting some simple target, you should be a recognizer and not a perceiver. You’ll move quickly, you’ll ignore everything that’s irrelevant, and then you’ll be able to optimize for your target. That is great, as long as there’s nothing of value in what you threw away and decided to ignore.

Jackson: You mentioned “autotelic” earlier, which is a wonderful word and one of the more compelling philosophies for how to live: finding love for its own sake and finding your passion. You reacted to my willpower comment about it being a little unrealistic or idealistic. Bernard Suits writes about a theoretical future utopia where we live in a world of abundance. We live in a pretty abundant world, but not a profoundly abundant one. Great things are clearly not always playful. You do all kinds of things that suck and are hard to do, whether it be for the aesthetic process you want or outcomes that are important. Yet, aspirationally, we all hope to get a little closer. To me, doing this now is closer to the autotelic than things I’ve done in the past. I’m curious what your relationship in your career has been to finding the love, the joy, and the internal motivation while also managing the messy external things. Is this just a thing we strive for until we die? People want to believe it’s incrementally more possible. Perhaps the answer is not to naively think you can just purely be autotelic.

Thi: I think you’re asking me about whether technological progress is good. Is that what you’re asking?

Jackson: It certainly leads into what we’ll talk about next. I suppose I’m asking for a personal reflection on… 19- Henrik Karlsson - Cultivating a Life that Fits Kevin Kelly talks about how you shouldn’t be the best; you should be the only. He says the goal of life is to become yourself by the time you’re on your deathbed. A playful orientation to life, a way of finding all this stuff we’ve discussed, feels like the directional way to get there. As someone who expresses these ideas in a profound way and also has to live in the messiness of being a professor, I’m curious: Is that just the struggle? How do you relate to this?

Thi: Every profession has its grind, and you have to do the grind. The question is whether you devote yourself entirely to it or whether you make space for other things. My career through philosophy and this value capture stuff is, in part, coming out of me pulling myself out of a bad trap. I went into philosophy because I loved it, and then the same thing happened that happens everywhere else. I went to grad school, and I encountered the metrics. In philosophy, there are status rankings for universities and journals. You aim to get a lot of journal articles published in highly ranked journals. It becomes clear how you would game that out: you hyper-specialize in a small domain and write very technical, small-move articles. A lot of really good work happens that way, but that path made me almost die of boredom. I was walking that path. In the course of my professionalization, I went from someone deeply excited about philosophy to someone who, by the time I left grad school, was grimly doing philosophy. I didn’t look forward to it. I had to work on and write things I wasn’t interested in. I had a dark night of the soul, and I think I was saved by my impatience and intolerance of boredom. I realized I couldn’t do it anymore and had to do something else. It’s really fun to talk here, but there’s a fantasy that I spend all my life talking about these ideas. At best, I get a quarter of my life to think about this stuff. The rest of the time is administration, teaching, and grading. It’s going through university policy changes to see if a trap has been laid for us. There’s a lot of grind. You always have to do it, but a previous version of myself threw the rest of me into the grossest part of the grind instead of carving out a spot to do what I wanted. Does that answer the question?

Jackson: Yes, but what a privilege to have that 25%.

Thi: Getting 25% of my life to think about this is pure play. I recognize that I’m one of the luckiest people in the world, that I get to play and roam through weird ideas for some part of my life.

1:16:59: Metrics, "Measure What Matters," and What We Miss

Jackson: We’ve talked around it, but I think the best place to start is metrics. In the tech and business worlds where I’ve spent a lot of time, there is a classic guiding principle: measure what matters. Of course, the implication of doing so is that we tend only to value the things we can measure. You have a few choice excerpts in your book on metrics: “Metrics are technology that standardize attention.” “Data is engineered information, and metrics are engineered values.” “Public metrics get rid of intuition. They force us to justify ourselves in the cold light of general comprehensibility. They kill opacity.” Finally, “Scoring systems don’t just discover a convergence that was already there; they produce convergence.” There have been many benefits to living in a quantified world, including our speed of progress, our collective understanding, and collaboration. Yet you spent a third of this book ripping into the ways that metrics are ruining us. Maybe that’s too strong a language. Why can’t we just tolerate a quantified world at a global level and leave the subjective, qualitative stuff for our personal lives?

Thi: That would be great if we could do it, but the worry is that there’s a constant, intense suction. I’m not saying we should get rid of metrics or that they are bad. Metrics are clear, comprehensible, accessible, and portable. We can unpack what all of that means in a bit. That is good for a lot of things. If we treated them as simple, low-quality but useful proxies, that would be fine. If they were used in limited ways as heuristics or guidance mechanisms for large-scale activity, that would be great. But instead, we find that when you put a metric in a space, everyone starts to care about it and hyper-orient to it. Body Mass Index (BMI) is a really interesting example. We know BMI is a terrible health measure that varies from person to person. It has been horribly abused. It was originally proposed as a useful proxy at the population level. We know some people are perfectly healthy at a high BMI and others at a very low BMI. BMI was introduced as an epidemiological measure to see if a whole country suddenly had a huge shift, which would be a sign that something was going on. It works great for that. It’s a rough first proxy for national populations and shifts in nutrition. You can use it to identify large-scale food deserts. But once that number is out there, people tend to hyper-orient toward it.

Jackson: It’s like a personal scoreboard for health.

Thi: It becomes a personal scoreboard, and its “proxiness” is lost. People forget that it’s just a really rough population-level measure. That is the worrisome thing. I want to talk about the idea of “measure what matters.” I can provide many bad examples of metrics, and the standard response is that they are just bad metrics and we should get better ones. The thing I’m worried about, and what I argue in this book, is that metrics are by their nature unable to measure certain things. Not that metrics are bad, but that they’re very good at measuring particular kinds of things. They are good at measuring things that are easy to count together, where everyone can recognize the borders and count things in the same way. Lifespan measured in years is easy to measure. Whether or not an intervention leads to death or saved lives is easy to measure. Changes in graduation rate are easy to measure. Another example: in the history of attempts to diversify various institutions, early people in these efforts wanted to diversify along many dimensions. They didn’t want it to become a quota of women or minorities. But diversity along lines of intellectual style, creative style, or cultural background is very hard to measure. The number of women hired is very easy for everyone to measure together. The basic intrinsic nature of metrics is very conducive to targeting some things and doing it really well. It is no accident that large-scale data systems have given us miracles like antibiotics. Antibiotics lead to a highly measurable result: the bacteria goes away and you stop being sick. But there are whole other swaths of human life, like art, beauty, wisdom, happiness, richness, connection, and friendship. All these things are very hard to measure.

Jackson: Critically, I would add—as you talk about in the book—health. It is important to acknowledge that there are a lot of things that are really important, even if you view them as not that subjective, that apply.

Thi: We should navigate to health last; it is one of the trickiest examples. One of the things I am really concerned with is when you have something really rich like flourishing, well-being, or physical health, and then you have a simplified proxy near it, like how long your life was or how low the heart attack rates were. That proxy begins to capture our sense of health, and then we start forgetting about the other stuff. It is not that lives lived isn’t important, and it’s not that we can’t target it. In the war of trade-offs, what the metrified world seems to do is get us to hyper-focus on what is easy to count together and drown out the subtler, quieter, more variable things.

Jackson: Yes.

Thi: You are teeing me up for the vomit, so let me give you the vomit. When I was trying to figure this out, I spent a lot of years trying to figure out whether there is something intrinsic about institutional metrics that makes it really hard for them to capture a lot of what is really valuable for human life. I found a bunch of scholarship that really helped me. Two key figures here are Theodore Porter and Lorraine Daston. Theodore Porter is a historian of quantification culture. He was deeply influenced by a philosopher of science named Ian Hacking. Porter tries to figure out why politicians and bureaucrats become so motivated to justify things quantitatively, even when the quantitative measures are obviously bad. His answer is to think about the two different kinds of knowing: qualitative knowing and quantitative knowing. Qualitative ways of knowing are rich, open-ended, context-sensitive, and dynamic, but they travel really badly between contexts because you need a lot of shared background to understand them. Quantitative ways of knowing—and here, big asterisk—we are not talking quantification in principle. We are talking about institutional quantification: how bureaucracies and large-scale organizations count. When we identify a context-invariant kernel, a little nugget that everyone understands the same. We create this by removing high-context detail to create something portable that everyone can understand. Now we can all understand what each other means by this single thing, and it can aggregate. My standard example is qualitative assessments, like the long paragraphs I write in response to student essays that talk about their relative originality or clarity. It can assess what they are doing in particular. It can pivot based on what they want. For a more creative person, I can focus on their creativity. It can do all kinds of things—

Jackson: Really meaningful to that student.

Thi: It is very meaningful to that student, and maybe to other philosophers, but very hard to interpret for someone in the business department, law school, or someone hiring in Silicon Valley. Crucially, it doesn’t aggregate. This is really important. By nature, if a lot of different people are open and context-sensitive and issue thousands of different qualitative responses, those responses don’t aggregate. The variability and open-endedness make them unable to aggregate. The fact that we fix the meaning of A, B, C, D in a letter grade and hold it the same is what makes that information travel well and permits automatic aggregation. Porter’s insight is so compelling and terrifying. Because institutional quantification is socially powerful by design: nuance, context, and sensitivity have been stripped out of it. It is an artificially created, easy communication mechanism. What it is missing is the source of its power.

Jackson: That’s the feature. It’s not a bug.

Thi: It is the feature and the bug. This is the trade-offs view. The thought that keeps me up at night is that the thing that gives it social power is its very insensitivity. That is stage one. Any kind of value that requires a lot of context to understand is blotted out.

Jackson: Or a lot of experience is systematically removed.

Thi: It is systematically removed. That's one.

1:24:16: Quantitative vs. Qualitative Ways of Knowing and Different Conceptions of Rules

Jackson: And Lorraine?

Thi: The other main lesson is from Lorraine Daston. She is an incredible intellectual historian and one of the most important intellectual figures in the last hundred years. She has this incredible early work about the nature of objectivity and how it changes. Her book, On Rules, basically broke open the secret heart of things for me. When we are talking about scoring systems and metrics, we are talking about rules for evaluation. There are rules for counting. There are rules for saying here is what we pay attention to, and here is what makes the number go up. There are rules for what counts. She says there are different conceptions of rules. The dominant older conception of a rule was what she calls a principle. A principle is an abstract generalization that admits of exceptions and requires discernment and judgment to apply because you need to know when you might be in an exception case.

Jackson: The exception proves the rule, structurally.

Thi: I have never understood the phrase “the exception proves the rule.”

Jackson: The essence of the point is that when you find an exception, it helps you model that the rule is almost always typically taken.

Thi: I get that. One way to put it is through one of my favorite examples of a principle: in creative writing, I learned show, don’t tell. People break that. But if you understand why they break it and when, you will understand why the rule generally holds.

Jackson: Yes.

Thi: And also, I need to break it.

Jackson: The point isn’t to throw out the rule. The point is that the rule should hold until it doesn’t.

Thi: Right. And there’s no mechanical way to say.

Jackson: You have to get a vibes thing.

Thi: Exactly. In some sense, wisdom is vibes. That is the next paper I am writing: Wisdom is Vibes. That is exactly the point.

Jackson: That’s a principle.

Thi: That is a principle. The second kind is a model, like a role model. She says the Rule of St. Benedict was just the person, St. Benedict. To rule yourself that way was to think about what he would do in that situation. Both these rules require judgment and expertise.

Jackson: Every single time. There is no automatic way to apply it.

Thi: There is no automatic way to apply it, like “What would Jesus do?” The models and rules are often complicated and subtle.

Jackson: Those two have some travel and some scale. There are principles, but they are harder to repeatedly scale.

Thi: For someone to really learn “Show, don’t tell,” they have to learn it in a context. I learned to apply it over several semesters in a creative workshop or apprenticeship. It is the kind of thing that travels in an apprenticeship culture.

Jackson: Yes.

Thi: The third kind of rule is an algorithmic or mechanical rule. This is something that is meant to be applied unthinkingly, automatically, with no exceptions, exactly as written. People thought that mechanical and algorithmic rules arose with computing machines, but they didn’t. They arose about 150 years earlier in an attempt to cheapen labor. You didn’t have to hire experts who had judgment. They make labor fungible. They make it so you can hire and fire anybody because the rule has been made explicit. Again, think about recipes: McDonald’s has mechanical rules, which means they can hire anybody and slot them in, and fire anybody.

Jackson: In Korea, in Mexico City, or in LA.

Thi: It doesn’t matter where you are. And note, also, it works particularly well when you standardize the inputs, like the bread and the flour. In that case, a mechanical rule is a rule that is applicable consistently by anybody.

Jackson: When you say mechanical rule, you mean algorithmic.

Thi: It is the same thing. In my book, I use the term mechanical because the term algorithmic has shifted in its use.

Jackson: It also harkens back to the incentive stuff we were talking about earlier. When values and incentives combine, you get mechanical value sets.

Thi: A lot of her examples are mathematical. People freak out because they think mathematics is by its nature mechanical, but you have to understand how much of mathematics is not mechanical. In many cases, you get a complex choice of which procedure or formula you are going to apply. You have to decide if you are going to do statistics, Newtonian mechanics, or modeling.

Jackson: They might be mechanical within those.

Thi: Right. Here’s an example: let’s say that you want to split a pie in half. What methodology do you use? You could split it by weight, or you could split it by angle. You could split it with “I cut, you choose.” Each of them is a different procedure. Splitting by weight is good for nutrition if you are counting calories. Splitting by angle is good for visual appearance. “I split, you choose” is good for a sense of equity and fairness between two kids. Each of these is a different procedure. Half is already a subtly complicated thing. Half by angle is different from half by weight, which is different from half by deliciousness or appealingness. There is a complex choice of procedure.

Jackson: If we lived in a nuanceless world of perfect detail, you could mathematically figure out half of a circle. But no pie is perfect.

Thi: The fewer complex overlapping details there are, the more everything is evened out. If all pies were completely identical and there was no variation in texture, there wouldn’t be a problem. But they are lumpy and delicious

Jackson: Almost like people!

Thi: (Laughs): People are like pies.

Jackson: Lumpy and delicious, yes.

Thi: Metrics typically prefer mechanical rules for counting. They tend to recognize differences between things that are easy for anyone to recognize and highly accessible. They tend not to pick out distinctions between things that require high discernment. One example tells us that mechanical procedures are sometimes great. You might think the right to vote tracks something complicated like intellectual and emotional maturity. But that requires a lot of discernment. We offer a mechanical rule: age 18. Being 18 does not track intellectual maturity.

Jackson: We have made a set of compromises. This is the best we can do.

Thi: In this case, it is great that we make that compromise. The cost of not getting intellectual maturity exactly right is low, and the cost of bias is so high. But it is not always like that.

Jackson: Right.

Thi: Michael Endicott, a philosopher of law, says that when you are switching between discerning judgment and mechanical judgment, you are actually trading off between two kinds of arbitrariness.

Jackson: I remember that.

Thi: With discerning judgment, you let in individual bias. With mechanical lines, you shut that out, but you introduce the bias of sharp lines where things are fuzzy, complicated, and gray.

Jackson: What a profound example. That example travels so wide.

Thi: It’s—

Jackson: We are deluding ourselves into thinking that one choice is arbitrary and the other is objective.

Thi: Over and over again, I’ve ended up thinking that metrics aren’t inherently bad, but there is a complex set of trade-offs and costs. We often have a fantasy that there isn’t a trade-off. In this case, we’re talking about the trade-off of accessibility. You can see it in the recipes trade-off again. If you make a procedure more accessible and make everyone follow that procedure, you also cut off expertise and sensitivity.

Jackson: Yes.

Thi: Sometimes you want to do that. Sometimes it's worth it.

1:38:01: Scaling Trust, Data, Experts, and Legibility

Jackson: I want to talk about illegibility, legibility, and trust. You discuss this as one of the core dilemmas we face in a world where science is so complicated. You have amazing work on conspiracy theories about how their appeal lies in making the world fit into your head. In the real world, there’s so much complexity that we have to specialize. You say, “From the outside, posers and visionaries can be hard to distinguish,” which illustrates the problem. You also have a great section on how transparency is surveillance. You say “we limit the harm that bad and incompetent people can do, but we also limit what good and competent people can do. Transparency leashes both kinds of people, forcing them to operate within the public’s comprehension.” Finally, “{we often assume that expertise is just technical—that experts are just there to run their machinery,” like McDonald’s workers. “But the goals and values guiding it all are not always obvious and accessible to everybody. This is a mistake. Expertise involves seeing more deeply into what our true goals should be and grasping the subtle values of the terrain. The reason you go to a doctor is that they understand things you don’t.” This is a profound problem without obvious answers. It’s about how much we can compress and make accessible. We are at an all-time low in institutional trust. While there are benefits we get from living in a metrics world, it also makes us want to trust philosophers, academics, and doctors less. What does trust need to scale?

Thi: Oh my God.

Jackson: That feels like the antidote to the metric.

Thi: The reason we were talking about Daston and Porter is to understand what metrics miss about values. Porter teaches us that large-scale institutional metrics typically remove high context. Daston teaches us that large-scale metrics typically remove expertise. They both are about different kinds of sensitivity and specificity. They emphasize the things that everyone can see and recognize consistently. In a metrics world, there’s no room for context. I’ve experienced a simple version of this. Our philosophy department was just cut again, defunded in favor of AI programs and the business school. We’ve been labeled an unproductive department, even though we are a highly ranked department with important publications.

Jackson: Lots of waste, dare I say it.

Thi: The reason is that the measure of productivity at my university has become research expenditures through grants. Philosophers don’t need grants. In a sense, we’re very efficient. You just need a book budget; you don’t need a lab. Because the metric has become research expenditures, it can’t make the contextual shift to realize that this group shouldn’t be measured by grants. That’s an oversimplified example. This is why metrics can’t capture values. Values are context-sensitive and require expertise. Metrics by their nature target what is portable, consistent, and accessible. They target what is legible. Another example I found fascinating is Charity Navigator. It is a nonprofit watchdog designed to tell us which charities were good and which were wasteful.

Jackson: The root of this is that if you don’t quantify, people will abuse systems.

Thi: Exactly. The reason we want transparency is because there are fakers out there. We impose a method of transparency to root them out by seeing who can succeed on a publicly legible target. That takes out two groups: the fakers, and the people trying to target something important that isn’t easily quantifiable or legible. Then the fakers will simply adapt and game the clear targets.

Jackson: You have this great concept in the book of the gap, which is the gap between what we can measure and what actually matters.

Thi: The gap is the difference between what is easy to measure and what actually matters. The Charity Navigator case is interesting because for a long time, they used a throughput rating. I was convinced by it and let it guide my charitable giving. The throughput rating measures efficiency—the ratio of how much money is given to the nonprofit versus how much it expends. It assumes that a nonprofit is just there to redistribute resources and not to hire experts or make decisions. In order to actually judge a nonprofit’s efficacy, you would have to understand its specific domain. If you wanted to compare nonprofits working on housing versus food, you would have to understand the complexity of improving an ecosystem or nutritional delivery services. But a throughput rating is a matter of accounting. That is the same for every nonprofit. It provides a single way to judge them all in an apparently objective, mechanical accounting system.

Jackson: And do it efficiently and fast.

Thi: It highlights what is mechanical and similar, which is accounting. It ignores what’s actually important but highly variable.

Jackson: So how do we scale trust?

Thi: The Charity Navigator example is super interesting because it’s an attempt to scale trust. It was an attempt to eliminate trust.

Jackson: Which is very different. Trustless systems like markets are interesting and can be really valuable for the world, but they can’t replace trust.

Thi: A transparency metric is made to make institutions accountable to the rest of the world to see if they’re biased. They’re trust eliminators. Instead of trusting an expert about their specific domain of what’s important, they look for some public way of counting the goods. They’re now forcing institutions not to use their expertise about what’s important, but to only focus and target.

Jackson: Only be automatic.

Thi: The simple, easily accountable values. The simple, easily accountable target. One of my favorite examples of this was when Congress put the National Endowment for the Arts under oversight because they were afraid of bias. They started measuring artistic success by box office receipts.

Jackson: Engagement. What a way to measure art. Thanks, Netflix.

Thi: Exactly.

Thi: Box office sales, ticket sales, page views, and engagement hours are all mechanically countable. The problem of the world is that there are different domains where people understand the special value of different things. To actually access that, we need to trust people beyond where we can go. That is the whole point of trust.

Jackson: And the point of specialization.

Thi: The whole paradox of transparency and accountability is that we want experts and specialists because we don’t understand them. Then you can only attempt to explain the things you can understand. That’s not possible. There is a tension between transparency, accountability, and accessibility versus trust, expertise, and sensitivity. I understand how trust works on a small-scale intimate life. I understand why it’s hard to get trust in people at a large scale and why we substitute things like metrics. I understand why that creates an enormous degree of loss. I don’t have an answer, but let me tell you two things I find fascinating. There are two moments from my favorite texts that I’ve been obsessed with. If I can understand them, then we will understand the heart of the modern world. One is from Theodore Porter, the quantification guy. He says data and information is a special mode of understanding and knowledge that’s been pre-prepared to travel and be understood by distant strangers.

Jackson: You even use the word engineered to do that.

Thi: There’s this other text I love. One of my favorite pieces of philosophy is Annette Baier’s “Trust and Antitrust.” She starts by complaining against social contract theory. There’s this mistake people make where they think that morality can be bounded on contracts, where contracts are envisioned as voluntary agreements between free people. She says that’s something only rich dudes in a gentleman’s club could have imagined would be the root of morality. Morality begins in dependence, vulnerability, and relations between parents and children. The heart of trust is vulnerability. Part of the mistake is trying to secure that vulnerability perfectly, because what it is to trust someone is to be vulnerable.

Jackson: I need a contract to trust you.

Thi: Exactly. The real reason social contracts are a weird place to build your morality is because it’s a very specific metaphor. Contracts are a specific social technology to make finds and expectations explicit to ease and secure one-off transactions between distant strangers.

Jackson: Which is beautiful. It allowed immense scale, but the whole premise is actually not about trust.

Thi: Annette Baier’s comment about social contracts eliminating trust and using one-off transactions against strangers, and Porter’s comment about how data and information has been prepared and engineered to travel to distant strangers, are two pointers to the heart of the modern world.

Jackson: My reaction to the scaling trust thing would be that we trust the other experts. What are the things that go wrong when we defer? Theoretically, the doctors should all regulate the doctors, but we run into the same problems you’re describing even in those cases.

Thi: The problem is that we have to trust experts from distant expertises. When I was a graduate student, the problem I was obsessed with is as old as Socrates: how does a non-expert recognize an expert? The dumb way to think about the problem is to think there’s some class of special experts. Really, most of us are non-experts in 99% of the world and experts in at most 0.01% of the world. We’re constantly having to recognize non-experts and figure out who to trust. We’re constantly having to overextend ourselves and become vulnerable. We try to secure it with hyper-accessible metrics which imagine away the complexity. They suppose that there’s some easily accessible moment, some test for real expertise. If there’s not a mechanical easy test for expertise, then we’re plunged into the true awful existential dilemma of the modern world: we are surrounded by people that are true experts and sensitives about forms of value and forms of life that we know nothing about. We’re also surrounded by posers, fakers, and exploiters. (Understanding) the difference requires expertise that we don’t have.

Jackson: The challenge is that we want security, objectivity, and answers. We don’t want to consider the possibility that we might get scammed.

Thi: I had a moment where I was talking about trust and vulnerability. I had a student in the back of my class—a big guy in a tank top. He said, “This is why I never trust anybody. You can’t trust anyone; they might screw you over. You always have to take care of yourself.” I asked him how he got to class today. He said he drove. I asked if he went on the highway, and he said yes. I asked, “How many other drivers and car mechanics have you trusted with your life today?” He actually had a meltdown. Annette Beyer says that trust is so intense and deep that we forget how much we’re trusting. Trust to us is like water to a fish. We swim in it all the time, so it becomes invisible to us.

Jackson: Are you an optimist in the cosmic sense?

Thi: I think so. Sometimes I walk my students through an exercise of trying to figure out how many people they’re trusting with their lives in this moment, sitting in this building. It introduces vertigo because you suddenly realize how big your trust is. There is a fantasy that we can secure it and know for certain. In philosophy, this is Descartes’ fantasy: that you could start over from the beginning and only believe in things you’re sure of by trusting only yourself.

Jackson: If we just rebuilt the whole world…

Thi: Without science! It’s funny—this mentor, Elijah Milgram, in his book The Great Endarkenment, jokes that the Great Enlightenment undid itself because it started with the idea of intellectual autonomy and rethinking things. That created so much science that intellectual autonomy became impossible. The idea that we can think for ourselves is an old, out-of-date illusion.

Jackson: Have you ever come across David Deutsch’s articulation of good explanations? It rhymes with your articulation of good science. You talk about it as having a good error metabolism. It is something that is atomically verifiable by an intellectually reasonable person. Maybe that’s our best hope. Otherwise, part of what I’m left feeling is that objectivity is not something we’re ever going to have. We are going to have trust, and we are going to have failures. We should also try to make more of the world have a good error metabolism.

Thi: Right. Okay.

Jackson: There is a lot there.

Thi: I think I have found the best tool to fuck with your mind.

Jackson: Okay.

Thi: Lorraine Daston’s first book is about what objectivity means. She says there are very different senses of objectivity, and they mean very different things. The notion of objectivity she thinks we have settled on in the modern era is what she calls aperspectival objectivity. To be objective is to be a kind of fact that is recognized no matter what person is looking and what kind of person they are. In this case, objectivity and truth come completely apart. Objectivity is the land of highly accessible, consistent judgments. There are some things that are easy for us to get objective about—the world of the easily countable. There are other things in which aperspectival consistency is incredibly hard, but they might still be important things.

Jackson: They might still be true.

Thi: They might still be true. The reason we’re talking about all this is because the dream of metrics is that by narrowing things down to entirely objective mechanical rules, we can secure our intellectual behavior and our judgments. We will always know for certain we’re right. Theodore Porter says the reason bureaucrats and politicians reach for numbers is to avoid responsibility. It is so good to not have to make a judgment or exercise discretion. They take themselves out of the apparent stream of judgment and say, “It’s not me; it’s just the numbers.”

Jackson: I don’t have to trust you because we have a contract. It is the same thing.

Thi: It’s a dream. If all that was important was easy to count together instantly and we could recognize it, then we could be secure in making decisions together. There would be a mechanical method. But it’s not. If you have subtle values that require context and sensitivity to detect, but you are so obsessed with the dream of security and the hope of objectivity that you will only reason using easily countable metrics, then you achieve security at the price of sensitivity.

1:54:37: Objectivity & Truth, Value-Laden Technology & Decisions, and "Objectivity Laundering"

Jackson: We have a little time left. This ties well into this last section about the way that value judgments are hidden everywhere. It is another thing adding to the point you were just making about…

Thi: The real life of objectivity.

Jackson: Yeah. First off, it is worth noting you have a great line: “Standardization may crush souls, but it also saves lives.” You are decidedly not anti-progress or anti-science. And yet, you say “when we start using any technology, we’re always outsourcing some of our values.” Why are technologies not value-neutral?

Thi: This is important because this is the second attack on the dream of objectivity. Many of the metrics we think of look objective, but there’s a value choice hidden at their core. This is a basic insight from the philosophy of technology and from a field called science and technology studies. There’s a presumptive view in our culture that tools and technologies are value-neutral. The idea is that scientists and engineers just make tools and empower people, and then the people make the choices. Once you look at the actual details, this is obviously wrong. Technologies can be biased in all kinds of ways. One of my favorite classic examples is Robert Moses and the New York bridges.

Jackson: I know a bit about Moses, but I’m not sure I know this specific one.

Thi: He managed to keep out what he perceived to be the rubbish from the good parts of New York by carefully setting the standard height for bridges and overpasses as slightly lower than the average bus. You can encode a bias into technology.

Jackson: Let’s back up for a second. The counterargument would be that that’s not a technology.

Thi: Exactly.

Jackson: Maybe buses are.

Thi: I was going to say decision. That’s an easy first example, but that’s not the real thing. Langdon Winner, one of my favorite philosophers in this space, has a beautiful article called “Do Artifacts Have Politics?” He says one of the interesting things is that technologies actually shape and push society into certain directions, often irrespective of what their designers and users hope for. One of his first examples is the printing press. He thinks it is interesting that oral communication is deeply decentralized. If I say something to you and you say something to them, there is no centralizing power. Each of us can change the news. No matter the democratic and anti-elite hopes of its inventors, the printing press concentrates communicative power and communicative authority.

Jackson: It requires capital.

Thi: Whoever has enough capital to have a printing press has that power. It instantly becomes the official news.

Jackson: This is the challenge of technology. So many technologies start this way and then they democratize. We can convince ourselves of these ideals and all the ways that we will democratize access to power. And yet, you could jump ahead to 2025. The slope of how artificial intelligence is adopted matters a lot, not just its ideas.

Thi: In a very abstract sense, many technologies have world-shaping powers. In the specific case of metrics, one of the things I find interesting is that people think metrics are neutral and objective. To have a metric to count something, we need a categorization system to count it together.

Jackson: Counting can’t be value-laden, can it?

Thi: Geoffrey Bowker and Susan Leigh Star have a great book called “Sorting Things Out” where they look at classification systems. They are interested in the “interestedness” of the classification systems that are the foundations of data and metrics. Every category involves decisions. We can’t keep track of everything in the world. We need to reduce the granularity in order to store and aggregate it, but there are decisions about where to reduce that granularity. For example, the U.S. Census has categories for Black, Caucasian, Asian, and Latino. That system is very interested in the difference between Asian and Latino, but uninterested in the difference between South Asian and East Asian. It forgets the granularity.

Jackson: That has seeped into how people believe. We just say “Asians.”

Thi: We just say “Asians.” Similarly, they are interested in medical record keeping. They look at the ICD-10, which is the classification manual for mortality and accident statistics used for large-scale epidemiological and medical research. In the falls category, there are separate codes for various urban falls: fall from a balcony, fall from a stair, fall from an escalator, fall from a bed, fall from a commode, fall from playground equipment, and fall from hospital equipment. There are about ten. For rural falls, there is “fall from a cliff” and “fall other.” You can read the interestedness. Metrics seem objective, but they track specific values.

Jackson: We have forgotten the value judgment that was made way back when.

Thi: We have forgotten. There are so many cases where we forget the value judgment made way back when. Theodore Porter’s book is about the history of the cost-benefit analysis. He notes that cost-benefit analyses look objective, but at the starting point, there are all kinds of interested decisions. If you are trying to do a cost-benefit analysis of how much you spend on a national park, you have to insert the value of a recreation day to a visitor. You have to attach a dollar number to it. In 1914, they said it was 14 cents per visitor day. That is the value of their aesthetic experience.

Jackson: That is the value of one person visiting.

Thi: You run that analysis and it looks objective at the other end, but one of the inputs is something we just made up. It expresses our sense of valuation. One of the most important numbers in any cost-benefit analysis is the discount rate. The discount rate is the relative amount you discount the future compared to the present.

Jackson: It is essentially how we keep the future from being wildly overvalued.

Thi: If you set the discount rate to zero, you get long-termism. Seventeenth-century economists like Adam Smith worked this out. If you set the discount rate to zero, you get absurd effects. Any minor change in the present would be justified because of the swamping effects in the future through compound interest. You have to have some discount rate.

Jackson: The present has basically no value.

Thi: You have to set a discount rate, but there is no correct one.

Jackson: We made it up.

Thi: The discount rate represents a value choice about the relative value of the future versus the present. Someone sets it at the bottom of your calculation, it gets hidden, and then things look objective.

Jackson: You could easily imagine how arbitrary discount rates might affect our broad sensitivities for how much the future matters. It goes back to the adoption of technology. Is it okay if we build this technology because in 20 years everything will be great? So much of our thinking is almost like we are avoiding the messiness of these ethical questions.

Thi: That is exactly the point. The phrase I have in my book for this effect is “objectivity laundering.” Like money laundering, where you take dirty money and pile on calculations, here you take a subjective choice and pile on calculations. The opening thought for this book was that games are beautiful because they simplify things. For once in your life, you don’t have to worry about conflicting values. There is a victory point scale, you know exactly how much things are worth, and you know exactly who won. It is so tempting to export that.

Jackson: Oversimplify the rest of the world. Simplicity contained is great.

Thi: Simplicity contained is great. If you manage to convince yourself that your metrics measure all that matters, then you won’t experience the tension of what you’re missing. You’ll think it’s easy. You’ll be much more efficient at achieving and optimizing for your metric because you don’t get any drag from worrying about all the other stuff. All you had to do was delete from your vision everything else that was important.

Jackson: Many people in my audience are technologists. I joked with you when we first met that you are a philosopher and think a lot about games. Both are wildly low status in an area of technology that should almost certainly be learning a lot from those two domains. Why can’t we do ethics from first principles? Maybe it’s a super simple starting point. You joked to me about that and I think it’s profound.

Thi: That’s a really intense question.

Jackson: What is your caution to people who think ethics can be simple?

Thi: If ethics is about treating people well and fairly, then doing it well will involve deep attention to the particularities of people and their context and sensitivity to the emerging complexity of what matters. This is from Aristotle. I learned this from Martha Nussbaum’s version of Aristotle. She says practical wisdom is being soaked in the moral and value complexity of particular situations.

Jackson: It’s perception, not recognition.

Thi: Exactly. There are some abstract things I could say, like morality is treating people well. But as an actual thing that guides action, what actually matters to people, what actually hurts them, and what actually helps them is highly dependent on interaction with particular contexts, personalities, and values. It’s going to require enormous sensitivity and context. If you expect not just to have a general vague principle of morality, such as act well and be sensitive, but a decision procedure that will resolve ethical debates and you expect it to be mechanistic, then you will start to concentrate on those moral features that are easily countable.

Jackson: And you’ll ignore the others.

Thi: And you’ll ignore the others.

Jackson: You give an example in the book that I love about how maps are value-laden as a technology. Maps have elevation because somebody made a decision 100 years ago for battles and they don’t have sound quality. The answer here is not to not use maps. The answer is not to stop participating in the world. We’ve benefited so much from compounding and a ladder of abstraction, particularly in technology, benefiting from the work that others have done in the past even if they were value-laden. Technology, to a fault, loves to solve for hero metrics and simplicity and solving for X. But even for technologists who are mindful of the ethical thing, who want to build quickly and with impact and global coordination, what do you say to those people? If I could give you 30 minutes in a room with Sam Altman, Mark Zuckerberg, and Elon Musk, what would you try to persuade them of? I realize this is a different question than for the thoughtful person who’s listening.

Thi: Don’t give me that question.

Jackson: What would you try to have there? You have another bit where you talk about reflexive control. Part of this is remembering to put on a mindset that gives way to these things while still operating in the real world and benefiting from all of these.

Thi: A lot of the times I’ve been asked this question about how we’re supposed to survive as individuals, not as technologists. What I say about the individual question is very much like the maps thing. We have to use maps. What we should hope to do is be aware that different maps reflect different values and choose our maps with care and sometimes make our own. At the structural level, if you have controls over the structure, one thing to do is to try to make a map for everyone and force it out. Another thing to do is to help create a variety of maps and help people figure out there are choices.

Jackson: Choices, or that there could be choices.

2:07:57: Advice for Technologists: Ethics, Maps, Value-Neutrality, and Playfulness

Thi: Make tools that let people build the maps they need and want and to be careful. I want to focus on a simple case. I’ve been thinking about people from the technology space who are concerned that polarization has screwed up politics in the world. They’re trying to solve for it by optimizing. They try to reduce polarization by algorithmically boosting content that’s equally agreed upon by both sides. I think they think this is a value-neutral way of proceeding, but it clearly emphasizes the politically centrist… I mean, here’s the uncomfortable version: imagine you went back to 1830 America where half the population still believes in slavery and you had a bridging content moderation algorithm that boosted the things that both sides agreed with. That is very value-laden and very choicey. There’s this fantasy in the technology world that you’re going to be able to avoid making moral or political decisions, but somehow improve the world and change it for the better and never have to think any complicated thoughts about what better is.

Jackson: I think it’s a little bit oversimplified, but directionally I think you’re right.

Thi: Maybe we might be thinking about different people.

Jackson: Anthropic and AI labs talk about ethics. It’s not that it’s not part of the discussion at all. But the push that I hope a conversation like this helps people with is that you need to perceive this way more. You need to be way more thoughtful about all of the places that values are laid out.

Thi: If you think there are some values that are easy and you can optimize for them, cool. But if you think values are tangled and complicated, and you might be missing a lot of it, and anything you do is going to vastly change the value landscape in a substantive way, you might have to be a lot more careful. There is no easy, neutral way to make the world better without making value and political choices.

Jackson: I have so many other things I want to ask you. I think we have time for just two or three more. You mentioned to me that you have another idea that you think you’ll probably spend the next five to eight years working on. Could you give us a tease?

Thi: The biggest, most interesting question for me is the question about data spun up whenever we coordinate, and we have to communicate. This will require making decisions together about what to ignore together and what to track together. How do we think about the essential pluses and minuses of coordinated communication—not just data, but for everything, for our concepts?

Jackson: You said we give things power by believing in them. What do we need more belief in?

Thi: We’ve answered this: play. I was going to say play or humility, but I think they’re the same thing. One way to put my worry about metrics is that the attitude of optimizing for a metric encodes the attitude that we know what’s important and we just need to max out for it. There is no process by which we might wander the world figuring out what we’ve missed.

Jackson: Yes.

Thi: The spirit of play is encoding a spirit of humility because you’re trying out shit even if it looks weird and silly.

Jackson: Ready to be surprised.

Thi: And you're open to being surprised.

Jackson: I have just one last question which obviously relates. Near the beginning of The Grasshopper, there’s a line where he says to two of his disciples, “I have the oddest notion that you are grasshoppers in disguise. That everyone is really a grasshopper.” How do you remind yourself that you’re a grasshopper?

Thi: I’ve had some professional success lately. I used to be the person that walked into the room and no one wanted to pay attention to me because I was the one working on the dumb stuff like games. Now I am getting to the thing where I can see the path of becoming a senior person where everyone wants to listen to you. I can start to trust myself too much and become overconfident. The way I remind myself of the fact that we are all just clumsy, playful, idiotic mud things is that I bring really weird toys. I bring some yo-yos. I’ve got these really silly spinning tops and I play with them with people and fail. There’s something really weird about doing something dumb and hard that you’re bad at with other human beings.

Jackson: Being a beginner.

Thi: Being a beginner. Being soaked in the oddity of it and not being particularly good at something. Being soaked in an activity where you don’t know exactly what it’s for and exploring it with people that I don’t know reminds me that I don’t know much about anything.

Jackson: What a great way to end it. Thi, thank you very much.

Thi: Thank you.

Description

Timestamps

Links & References

Transcript

0:00: Opening Highlights

1:39: Introduction to C. Thi Nguyen

5:12: Thanks to Notion

6:31: Start: What Does it Mean to Be Playful?

13:41: Starting Local: Agency, Scoring Systems, and Games

23:37: Value Capture: Incentives, Values, and the Collapse of Meaning

36:28: What is the Shape of Good Values?

49:45: Attention, Recognition vs. Perception, and Aesthetic Openness

58:46: Process vs. Outcome, Striving Play vs. Achievement Play, Recipe vs. Dish

1:10:00: Aesthetic Value & Autotelic Pursuits in Life

1:16:59: Metrics, "Measure What Matters," and What We Miss

1:24:16: Quantitative vs. Qualitative Ways of Knowing and Different Conceptions of Rules

1:38:01: Scaling Trust, Data, Experts, and Legibility

1:54:37: Objectivity & Truth, Value-Laden Technology & Decisions, and "Objectivity Laundering"

2:07:57: Advice for Technologists: Ethics, Maps, Value-Neutrality, and Playfulness

2:18:52: Closing Thanks to Notion