SPENCER: Uri, welcome back on the Clearer Thinking Podcast.

URI: It's my favorite place to be.

SPENCER: This is a very special episode. This is our 300th episode. Can't believe it. So we're going to go a bit meta. We're going to talk about what makes for a great conversation. Obviously, it's relevant for something like a podcast, but also just in people's lives. Many people who listen to this want to have great intellectual conversations. I think many people feel they don't always have the caliber of conversation they want to have. So that's what I want to dig into with you.

URI: Amazing. Let's start.

SPENCER: All right. So we'll start by discussing one-on-one conversations. Later on, we'll talk about group conversations, which is a whole other thing, and I think you and I have tons of complaints about how group intellectual conversations go. But suppose you're talking to a friend one-on-one, you're both interested in intellectual conversation, interested in the same topic. What's one of the things you would point to that could go off the rails or go wrong, preventing it from being a great conversation?

URI: Yeah, so the way I view conversations is that you're trying to build a kind of tower together. I've never found the perfect metaphor for this, but it's roughly, I put down a brick, and you put down a brick, and then we're climbing over these bricks together, higher and higher into the sky. Because we're both contributing things, we both get to climb to new places we wouldn't have otherwise. The first thing that I see that so often goes wrong is that you're just not on compatible pages. So you're not building a tower together. You're building two separate towers or attempting to and then flitting back and forth between them, and then neither of you gets anywhere new.

SPENCER: Now, would that be because you both have a different idea of what you're discussing, or a different goal for the conversation?

URI: I think it can happen for a lot of reasons. One obvious way it happens is when people are just saying cached ideas that they've already thought before and want to repeat them again. This is another of my least favorite things. People sometimes just want to say something they've already thought about; they're not thinking in real time. Having listened to 300 episodes of the Clearer Thinking Podcast, my favorite episodes are always the ones where you can tell that the guest is thinking and responding in real time, and they're not just saying something they've said before. They're listening to what you said, thinking about it, and then saying something potentially that they've never even thought before the conversation started.

SPENCER: Yeah, I think this is a big challenge. When you bring on experts, they've often spent hundreds or even more hours talking about that subject, which means that they have many things that they essentially memorized. That means not only are they not thinking when they're saying it, but they're very unlikely to go in a different direction than they've ever gone before. They sort of have this particular path that their mind takes.

URI: Yeah, there's a kind of danger and excitement when you talk to someone who is thinking in real time, because, as you said, they might fall off; they might say something new, they might think something new. They might realize that everything they ever thought before was wrong. I do think it is a kind of brave thing to do in a conversation, but for me, that's what makes conversations interesting. I guess I should say my goal is, can we come up with new thoughts between us? Can we get to new places that neither of us has been before? I think high level, that's my goal in most intellectual conversations that I participate in.

SPENCER: Now, obviously there's a difference between being in a conversation and listening to one, but I think you would say that even when listening, you'd rather people be having new thoughts. You could imagine someone saying the opposite, "Oh, I want to know this expert spent 100 hours thinking about this. I want to know what they concluded from that. I don't need them to be thinking new thoughts."

URI: For me, this kind of gets to the heart of what conversation is for, because I'm happy to listen to a well-structured talk by an expert, where they're telling me something they've thought about before, they're giving it to me very clearly. Maybe a book or an audiobook has the same idea. But for me, what conversation is, is thinking together. I have a friend who always says, "Friends don't let friends think alone." I guess my model of conversation is collective thought; that's what it is. Do you have a different model? Do you think intellectual conversation can be something else as well?

SPENCER: Yeah, so I guess what you'd say is, "Look, if you want the thoughts they've already had about the thing, go watch their lecture." But if you're going to watch a podcast episode with them, you want this different thing. I think there could be other uses for conversation. I think what you're describing is one good use, but another is that a podcast host or interviewer, whoever it is, can go deeper than they normally go or approach it from different angles than they normally do. For example, maybe they have a sort of standard spiel, but nobody ever asks whatever follow-up question, or nobody ever asks, "How is this relevant to this particular audience?"

URI: How much do you think the best conversations are relatively equal or balanced? Because you could imagine there's an expert and there's a novice. The novice is asking questions of the expert. Maybe the expert isn't learning anything at all from the conversation, but the novice is learning a lot, and maybe the audience is learning a lot. Would that count for you as a good intellectual conversation?

SPENCER: I think this is a distinction. If I'm in a one-on-one conversation, I think there are sort of learning conversations where it is lopsided; one person knows a lot more about the subject and may be teaching the other person. The other person could be teaching me, or I could be teaching them. That can be really enjoyable and intellectual. To me, it's not the highest form of intellectual conversation, but it can still be great conversation. Like, "Oh, you know a lot about this thing. Can I pick your brain and learn from you?" So maybe that's just a distinct type of intellectual conversation.

URI: You're an expert on all the things, and we have sometimes had conversations where I'm just asking you about it. It's clearly not exactly equal. I'm asking you about a problem I have or a situation. Do you think you learn anything from those, even just by the way that you express things? Is there potential for learning even when you're on the expert side? Or is that just a one-sided thing, which is also okay? Sometimes things in life are one-sided.

SPENCER: I think absolutely, when you're teaching, you tend to learn. Once you've done it enough, maybe it kind of locks in and you're less likely to be engaging in real thought. But certainly when you're teaching something for the first time, like, let's say you know a subject really well, but you've never explained it to someone, that's a great learning moment. I think it really helps you compress the information, think about what's relevant, and think about how to explain it in accurate terms that can be understood by someone who's not an expert. So I think that's all very useful as a learning experience.

URI: I guess, okay, I should be more concrete. We had a conversation recently. I was worried that someone I knew might be a sociopath. You know a lot about sociopaths, narcissists, and other dark triads. You've had a lot of great episodes in the last year on those on the show. In that conversation, you did something a bit like a doctor. A kind of differential diagnosis. You talked me through it. What should I look for? How would I guess? Do you think of that as a good intellectual conversation, or do you think of that as a different thing?

SPENCER: There's always the question of what we mean by terms. I think often something that stymies good conversations is that people use terms in different ways. So I would just ask, what do you mean by a good conversation? What do we talk about? Is it a good conversation? It could be good if I give you useful information, good in that way. Is it intellectual? It definitely involves thought. So, yeah, what do you mean when you say good intellectual conversation? Let's try to model here what we're saying is the ideal form of conversation.

URI: Okay, well, maybe to approach this from a different angle. Are there any conversations you've had recently that you think of as epitomizing great conversation?

SPENCER: To me, the pinnacle of good intellectual conversation matches what you're describing, building a tower together, where you actually don't know where you're going to get by the end of the conversation. You literally don't know. You're both having new thoughts, but you're also both adding to the structure so that you literally create something that neither of you could have created on your own, which I think is what you were getting at. To me, that's the best. I think it's the best partly because I think it's where the biggest learning tends to happen, but also just because I find it really fun. For both of those reasons, other people might have a different notion of what the best is. It's partly why I think that debates, while they can be interesting and you can learn, I don't view them as anywhere close to the pinnacle of intellectual conversation because I don't think they have that structure at all.

URI: Yeah, so what do you think is missing? Imagine a good debate. It's quality. The two sides respect each other. There are obviously ways that it can go wrong, where they're just disrespectful and angry, and it's not fun. But imagine a respectful, high-quality debate. What's worse about that than your ideal conversation, or what's different about it?

SPENCER: I think it almost always involves mostly cached thoughts, where they've already thought about it a lot, and especially a good debater. A very amateur debater is less likely to be having cache. A really good one knows all the things their opponent might say and knows what they're going to say in response. It's really more of a performance art at that point. There's also very little openness to changing their mind, and that's something we didn't talk about. In a really good intellectual conversation, both people are coming in. In Scout Mindset, as Julia Galef would describe it, where they're really trying to understand the world. They want to understand what's true. They're not coming in from a soldier mindset, where they're trying to convince the other person of something or trying to win the debate. A debate sets up almost the exact wrong setting, where you're trying to win. You're not open to new ideas, you're not open to changing your mind. The fact that people are watching a debate adds to this. You can have a one-on-one debate when nobody's watching, but if someone's watching, it's even worse, because then there's a social element where you might feel embarrassed if you lose or get owned by the other person.

URI: Do you have any tricks or formats that set up good scout conversations?

SPENCER: I think one thing I recommend is that people keep this in the back of their mind. If they're having a challenging conversation where someone's more likely to be in a soldier mindset trying to win, they should remember that before the conversation can really get going, they have to make sure the other person is in scout mindset, and they have to make sure that they themselves are in a scout mindset. That means trying to understand the vibe and how both people feel. If either of those is not met, and you want this to be a real intellectual conversation that can make progress, then the first goal is to ensure both people are in a scout mindset. If the other person is not, then you might focus on lowering their defenses, making them feel at ease, and making them feel connected to you. If you yourself are not in a scout mindset, then it might be more about thinking, "Why am I feeling defensive around this? Is this a conversation I actually want to have? If I do want to have it, what kind of assurance do I need to feel comfortable here?"

URI: I really like that. I think vibe is surprisingly important. Even when the topic is intellectual, getting on the same wavelength and tempo. We have a mutual friend who I think is incredibly smart and very interesting, but I can't talk to her for more than half an hour because she's too fast for me. It feels like our speeds are not aligned. She's going faster than I can keep up with, and I often think that if she could slow her down somehow, we would have multi-hour amazing conversations. But instead, I'm just kind of interested but unable to keep up.

SPENCER: What happens by the end of that? Do you feel exhausted? Or do you feel like you literally are not understanding what she's saying?

URI: Genuinely, physically exhausted. It's tiring in the same way that running is tiring; I could do it for some amount of time. It would be interesting to ask her, maybe after the show we can get in touch with her potentially, and see how she feels about this, whether she feels everyone else is too slow, whether it slows her down from her thinking. Yeah, that's an interesting one. And have you ever had this experience?

SPENCER: Well, it raises a question, which is, what makes for a good vibe in a conversation, right? So you point at one factor, which is, there's something about it mentally exhausting? And maybe it has to do with the speed of the conversation. Do they give you time to pause, things like that? But what else do you think affects the vibe?

URI: There's some amount of goodwill, just a sense of goodwill, good energy, positivity, et cetera, the feeling that someone likes you and will continue to like you despite what you might say and disagreements you have. I think there's a feedback loop to conversations that is often underrated. In the best conversations, you can tell from the way the person is reacting whether they're with you or not, and you can often skip steps based on someone's micro-expressions and reactions. That seems really important to me because I think you can get a lot further. I can basically say something and trust that if you're nodding, then you're with me, and I can jump ahead. If you're not nodding, or if you're actively shaking your head, it means you can see where I'm going wrong, and you need to interject so that I can get back on the right path, or the path that I know I've misunderstood something. All of this seems like a really important part of the conversational vibe to me.

SPENCER: Yeah, that also speaks to this idea, the term social monitoring. Are you familiar with that?

URI: No, I'm not.

SPENCER: So it's basically how do people pay attention to the social situation? Some people are high social monitors, where they're constantly evaluating what's going on in the conversation. How is everyone feeling? How am I being perceived? Other people are just kind of not aware of that layer. But it does seem like there's some awareness layer that really helps the conversation, noticing, "Oh, wait, the other person seems confused right now, or the other person seems unhappy about something," or the other person seems really engaged when I said that thing, and that I think can really make the conversation go better because it gives you this extra information layer to integrate into what you're saying.

URI: Yeah, and I've never made these norms explicit with anyone, but I sense that there's some agreement on meta, conversational meta that you're allowed to interrupt at certain times if you think someone's misunderstood you, for example. Normally it would be rude to interrupt, but if someone is responding to something, and you think that they're responding to the wrong thing because they caught the wrong end of the stick of what you're saying, you're allowed to interrupt because otherwise you would just be wasting time. It kind of reminds me of Robert's Rules of Order. It's a very official way of running meetings, big organizations, and the UN kind of places. They follow this very big rule book. One of the things in the rule book is there's an order, and you speak in order, but you're allowed to interrupt with a point of information or a point of order when you think something has gone wrong. I think we kind of have informal versions of that in our everyday lives, and I think that's really important. It's one of the reasons I think verbal conversations are much better than written conversations most of the time.

SPENCER: And what about video versus phone? Do you think that being able to see the person or even being in person with them really matters, or do you think it doesn't?

URI: So I feel weirdly mixed about this. Video is much higher bandwidth, and I just said that I think the information you get from visual cues is really important. But I sometimes feel like I have better conversations with voice only, and I don't know why that is. I think in some ways, you can be more vulnerable when you're not looking at someone, or it's easier to be vulnerable with someone when you're not looking at them. People also have really good conversations in cars, I think. If you're both sitting in a car, you're facing ahead, you're not staring into each other's eyes the whole time. I think that can be a great setting, sitting on a park bench. There are all these real-life settings where you're not looking at each other, and I feel like that contributes somehow.

SPENCER: But it seems like a trade-off. You lose certain social information that could help you realize, oh, that person's confused or that person's really engaged. At the same time, I don't know about you, but I find there's a certain amount of energy and focus it takes to track all the social information. Maybe some people do it more naturally, and it just doesn't take any energy to do that. But by being on a phone call, it frees me from that. Instead of focusing entirely on the conversation in person, I'm like 10% focused on the social dynamic or something.

URI: That's really good, that makes a ton of sense to me. So you're not wasting or spending so much energy on monitoring, and then presumably there's more chance that it can go completely wrong because you miss a cue or whatever. But conditional on it going well, it goes very well. I am in high school. I once had a conversation sitting with someone in a corridor that was completely dark. I think it's the only time I can remember doing this, just sitting somewhere completely in the dark but just voice only. I do feel like that was surprisingly effective, surprisingly positive, in a way that I didn't anticipate.

SPENCER: It's so funny you say that, because that's something I used to do with my friends when I was young. In high school, we would actually hang out in a completely pitch-black room. It's actually something I've been thinking of doing for my social experiments group — doing a "voices in the void" event where it's like a completely pitch-black conversation. I do think it removes a lot of social layers, which is really interesting. I also think, as you hinted at, that people might be more vulnerable in situations like that; somehow removing that social layer makes people more okay with talking about what they really feel.

URI: So even though we're talking about intellectual conversations, I think being emotional is really important. Often you're talking about something political or social, and if people aren't saying their true views, the conversation can't progress. Even though it's an intellectual conversation, emotional vulnerability really contributes.

SPENCER: I think emotional vulnerability and tracking emotions in general matter for another reason too, which is that oftentimes conversations go off the rails when one person becomes emotional. There could be other reasons they're feeling emotional that are unrelated, but if they're feeling emotional about what the other person said, that's often a way that the intellectual conversation can fail. Let's say one person is abstractly considering an idea, and the other person is horrified by that idea. The fact that you've been discussing it is upsetting them. I sometimes teach a workshop on disagreement, how to have productive disagreements, which is not the same as trying to have a good intellectual conversation, but it's related. One of the things I teach is that it's critical to focus on the emotional tone. If the emotional tone goes negative, that should become your priority, much the same way that if either person loses scout mindset, that should become your priority. If the emotional tone of the other person becomes negative, the conversation has to shift to being about that, or at least addressing it, before you move on. If you don't address that, it can really go off the rails. The other person is getting increasingly agitated, or you're getting increasingly agitated.

URI: Yeah, for sure. I have a couple of topics related to intellectual conversation that I'd love to discuss with you.

SPENCER: Let's do it.

URI: We can model it in real time. The first one is this tower metaphor. It's always somewhat frustrating for me. It doesn't feel quite right. I think there's something related that's right. I have a second metaphor that doesn't feel quite right either, which is a walk in the forest where the idea is that there's no particular path you need to go down. There's no highway. You can go in any direction, but you do want to go in a direction. You want to both be moving along, and at every break in the path, there are multiple options where you can go. There's something about how moving together collaboratively is important to conversations that doesn't really work for me either to explain the joy of great conversations. Do you have thoughts, ideas? What is the best?

SPENCER: To elaborate through the tower metaphor, you're each adding blocks to build this tower, and together you're building the tower. You're not sure exactly what the tower will be, but you both are contributing to it. In the end, you kind of built this tower, right?

URI: Yes. What I appreciate about that is this idea that you're both bringing in ideas, information, and thoughts that the other person hasn't had, and that, therefore, you can kind of climb together. But there's something that feels wrong to me about maybe it's the difference between building and discovery. Maybe it's about directions in space. Maybe it's about the fact that you're not actually both on the same tower; you're both in reality, building separate towers in your own separate areas behind a big wall that separates.

SPENCER: Because you can't ever know what it's really like for the person to experience this thought, so we're just getting our own version of it.

URI: Yeah, and so you're just hoping you've put the same bricks the other person gave you in the right place. I don't know.

SPENCER: What does it matter? If you both, at the end of the conversation, feel like you got a lot out of it. Does it matter that you have the same tower?

URI: I think it matters that it's roughly the same in the long run, but I could be wrong about that. If you both built a tower that you felt good about, but it was actually two different towers.

SPENCER: I feel like there are some people I know that are intellectuals, but they are more vibey in the way they deal with ideas, where they might watch a talk, and it might stimulate interesting ideas for them that may be totally not what the speaker intended, but they still got something out of it, and they appreciated it and valued it. Whereas I think I tend to be the sort of person that's like, "No, I want to know what that speaker was actually trying to tell me. And evaluating that and whether I think that was something I want to know. And what do I think about that?" But I know intellectuals are both types, and I think that the first type of intellectual wouldn't be bothered by having two different towers, as long as everyone was happy with their towers.

URI: Something about it really drives me nuts, and I can't show it. So I will say, I write book thoughts. After I read a book, I'll write my thoughts. I think I got this from a blog post you wrote many years ago. You said you get a lot more value from a book if you spend an extra 10% of your time writing about it. But I don't write reviews, and often my book thoughts are just wildly divergent from the thing I read. So I don't really force myself there. I just think, "Oh, here's some source material. It inspired some thoughts in me." If I like the thoughts, then that's good. It doesn't really matter what relation it had to what the author wrote. But I aggressively dislike this idea in conversation. I don't know why, but it feels wrong to me. It feels like if we're not building the same towers, then what are we doing here? I don't know if it's because I think it's bad or just because I think it's impossible in the long run. If you keep building things that are not the same, then when you reference those ideas to each other over time, it's not going to work in my mind.

SPENCER: Maybe it works for the individual towers, but it doesn't work for building the whole kingdom. It's like, "Okay, we both had a good experience. We built different towers, but we're not living in the same place, right? And so when we come back and reconvene, we can't pick up on that and build more. We're not even on the same wavelength, right?"

URI: I think you have just, in real time, identified — not what the right metaphor is, but why this tower metaphor bothers me. It's partly because each of these towers is itself a chunk that can be used to build other towers. Each tower is a building block towards future towers in a way that doesn't really make sense for towers; it needs some other descriptor.

SPENCER: Well, I sometimes think about beliefs with this metaphor of a temple, where you're building temples with different columns and different floors. So you've got, at the very bottom of your belief system, really core beliefs. Those are some of the columns, and then you've got a level you put on top of that, and then you build more columns and put a level on top of that. I wonder if something like that would fit here.

URI: I see what you mean, but it feels like a good idea is infinitely replicable. Once you have a good concept, you can use it in tons and tons of different places.

SPENCER: It's not rooted in one sort of structure.

URI: And this is actually one of the reasons that within an academic discipline, people can talk to each other really easily if they all studied economics together, and now they know a bunch of economics terms, and they're fairly confident that when they say it to someone else, they're meaning the same thing by the term. Philosophy is maybe a better example where if you go to philosophy seminars, the philosophers are talking at a speed and at a level that I can't comprehend because they all know the references. They know the objections. They know the previous work.

SPENCER: They could just reference an objection. Everyone knows what they're talking about. They don't have to re-explain the objection.

URI: Yes, yes. And they can say, someone says something, and then somebody says, "Oh, but you know the Spencer paradox, of course." They kind of understand what that means, which seems to be one of the best ways. Some of my best friends, I feel like I can move at that pace. We have a lot of shared ideas and contexts, and that replicability seems really important.

SPENCER: What about a crafting metaphor in Minecraft, where you're combining ingredients to create a special material? We don't exactly know what we're going to create, but we're combining our ingredients now. Okay, now we have new material. We can use that as part of the material for the next thing we craft.

URI: I have not played Minecraft, I have to admit.

SPENCER: I've never played anything. I might be someone who plays, but basically, I think you can combine ingredients to make new ingredients, and then those can be used for further crafting.

URI: This is funny because I haven't played it either, but I believe I understand your idea and I agree with it. I think this is much better. You're developing technologies, you're developing new combinations of materials that create new materials, then you can use those materials in the future, and once you have the material, you can keep it. I really like that. I think that is a better metaphor.

SPENCER: Okay, maybe metaphor upgraded question mark?

URI: Potentially.

SPENCER: But are you building the same material? There's still the problem of you don't necessarily know that you have the same exact idea.

URI: So I have this idea, if not, could not, which is that oftentimes you can tell if someone understood what you said and is on the same page as you because if they didn't understand, it's implausible that the next thing they said would be the next thing they said. So, if I explain some problem I'm having to you, and then you jump in with an example that fits perfectly, it's possible that it's just a coincidence, and you completely understood something different from what I said, but it's very, very unlikely. Whereas, if I say something and you say, "How does that make you feel?" I have no idea if you understood what I said. I have no idea if you were even listening. Famously, ELIZA, this chatbot from the 70s, could do quite a lot of this and quite successfully. Just say, "Tell me what you think about that." So yes, I do think, in theory, we have this problem of never knowing whether we're on the same page. But in practice, I think we often have some sense.

SPENCER: Yeah, I think that's right. And this actually goes to something else that we haven't discussed, which is that you have to communicate accurately with your emotions or with your body language or your voice, because there are people who have encountered where they will lead you to believe that they understand what you're talking about when they actually have no idea. And it can be a vulnerable thing. "I actually am totally confused. I'm not following you, or I spaced out for a moment. Can you repeat that? I completely missed the last paragraph you said." That seems really important in a good intellectual conversation, that level of, "Hey, I'm not following you here," not just nodding along because you might do that. Let's say you were at a barbecue, and someone's rattling on about something you're not that interested in. You space off a little bit, you might just nod along; it's the least awkward thing to do, but that doesn't work in an intellectual conversation.

URI: I completely agree. And I think there should be no shame in saying, "Oh, sorry, I spaced out." There's no shame in saying, "Oh, I didn't understand," even if you've just been nodding, because a lot of the time, if you nod while someone's talking, you can catch it up; your brain can figure out later what they said earlier, and it'll all make sense. So even if I'm nodding and then you say something and I'm like, "Okay, actually, I don't know. I didn't understand. Can you say that again?" I think we should celebrate this. I think this should be much more common.

SPENCER: And a related point there is definition. So many conversations just don't get off the ground because people are using different definitions and not identifying it.

URI: Interesting. So you think people are using words, they think they're using the same words, but they're using the same words to mean completely different things.

SPENCER: Yeah. Will it not necessarily be completely different? Often the meanings we use are related, but they're not sufficiently related. Your meaning of a word and mine will never be identical, but they should be similar enough that we're having the same conversation. Part of the reason I know this is true is that I've run my disagreement workshop. We'll have people sometimes map out what was the root of the disagreement, can they work it out together, essentially working as a team to figure out why they disagree, and then we lay out different common reasons for disagreement. One might be that you have different values. One might be you disagree about certain facts, and maybe you could figure out a way to check those facts. But another is you literally just disagree about definitions. I would estimate that maybe about one in three of those conversations, they realize by the end they do not disagree, even though at the beginning one said strongly agree and one said disagree. At the end, they're like, "No, we actually never had a disagreement here."

URI: That is wild. Oh, that's so wild and so frustrating in some way to think that potentially one-third of the disagreements I've had in my life were just completely unnecessary.

SPENCER: I don't know about yours. This is a specific context where people are assigned to disagree with a stranger on a topic they say they disagree about. I think it's fascinating. Even if you're on the same page of definition, tightening it up and saying, "Hey, wait a minute. Can we make sure we're on the same page here? Here's what I mean by this," and making sure that they consent to that. The other thing I noticed is that it can be really mentally taxing to use a different definition for the same word.

URI: Yes.

SPENCER: Let's say you and I don't agree about the definition. The solution is not to debate the definition. That's almost never the right solution. One of us has to switch definitions. Sometimes it's best to just make up a new word. You know, Eliezer Yudkowsky's idea of taboo, like, "Can we just come up with a different word for this that we can both call it?" As a generous conversational partner, you can agree to use their definition, and now it's on you to track what it means, not to expect them to adopt your definition, which may be hard for them.

URI: I think that makes sense. I read a book once. I want to say it was Foucault, but I'm not sure, where he said that up until the 1700s or 1800s, everyone believed that each word had a true definition, almost in a religious sense, that everything had a true name and a true definition. The invention of modernity was the understanding that there is, "What is a planet? Is Pluto a planet?" The answer is kind of maybe. "What do we mean by planet? There are several different things we could mean by some of them; it is, by some of them, it isn't." There isn't actually a truth here. There isn't some objective reality — I'm sorry. Don't send me emails if you're big Pluto fans. I hate upsetting the Pluto community — many disagreements are like this. My feeling is that secretly, a lot of us, in many situations, act as if this is still true, as if either Pluto is or isn't a planet, and this is worth debating, rather than just saying, "Okay, you know, it has some of these properties. It doesn't have these properties. There's kind of this rough thing we mean by planet," and that's about all that you can say, I think.

SPENCER: Yeah, I think dictionaries, although very useful, are incredibly misleading. You have to think of them as retroactive attempts to be like, "Huh? How are people using this word?" So often they're at least a little bit out of date, not always, but often. Even when they're not out of date, they add a false precision because real usage is going to be this big cluster of stuff, and they're trying to stick that definition somewhere near the center of that big cluster, but that's the best they can do. They're never going to capture all the different usages and all the nuances of real usage, and they're never going to capture the full range of how it's used.

URI: Do you think in words?

SPENCER: I think I perceive myself as thinking in words, except that I view it as a metaphor. There's this garden that's my consciousness, where I can think about things, reflect, and work through logic. Then there's this wall, and on the other side of the wall, there's this dark forest, and the wall is really high. I can't see it. What I can do is throw things over the wall and wait to see what comes back. Sometimes things will be thrown back over the wall from the dark forest. I have no idea what's happening in the dark forest, but when something comes back, I can say, "Hey, subconscious, I needed an idea for this." Then I wait, and then it comes back over. "Oh, that's cool, but actually, I need to be more like this." Then I wait again, and it comes back over. So that's kind of how I think about it.

URI: When you throw things over the wall, how specifically do you think you're getting the thing? How specifically can you query this dark forest? That's kind of my question.

SPENCER: I find it really interesting because not all queries work equally well. For example, if I say, "Hey, would you mind coming up with words that have the third letter C?" Some people might find that really easy, but my dark forest is not good with that kind of query. Other people, I think, are much better at that kind of query. On the other hand, there are certain kinds of queries it seems really good at. If I say, "Here's a list of things. Generate more things that are like this," it seems to do a good job often. Over time, you become aware that it's running some kind of really complicated algorithm in the dark forest. You can begin to learn about what sort of things those algorithms can do and what sort of things they cannot do, and in what terms they are error-prone.

URI: I presumably have a dark forest, and then I have a garden, and you can communicate into my garden, perhaps by throwing things, but in a slightly different way.

SPENCER: Different kinds of... Yeah, we need a different metaphor. What am I doing? Just drop it. I'm giving you a note or something.

URI: Yeah, are there kinds of queries that you can ask your dark forest, but you can't ask me?

SPENCER: Yeah, absolutely. There are kinds of queries I can do. For example, with my dark forest, I can imagine a scene, and then I can have it auto-complete visually, and I can have it auto-complete what happens next visually. So I don't think I know how to do that to you.

URI: Yes. I think I have a version where I can sometimes have a thought, like, "Who is that guy who was at that thing with that person? Where each of those is the abstract noun?" And then my brain goes, "Oh, yeah, so and so." And there's no way I could ask you that even if you were at the same ones with the same people, because somehow multiple missing pieces are getting filled in at once.

SPENCER: Yeah, it's filling in a template. Does this way of thinking about your mind resonate with you? Do you feel yours is similar?

URI: I like it a lot. I tend to think of the dark forest as the Muses. I much more believe that I'm a conduit for ideas that exist outside of myself, but that's entirely metaphysical.

SPENCER: Do you really believe that, though?

URI: Do I genuinely believe that?

SPENCER: Don't you think they're coming from your brain?

URI: In a sense, I genuinely believe it. It is how it feels to me. Leonard Cohen was asked once about why he used a particular song in one of his lyrics, and he said, "Buddy, I don't write the words," which is to say that they come from God, and they just get channeled through him. Genuinely, this is how it feels to me. Even though I understand intellectually it is not currently likely that panpsychism is true and that I am feeding from the universal consciousness. But I guess, again, it doesn't particularly matter. I think I'm doing the same process. There's a big wall, and there's something on the other side.

SPENCER: Your dark forest, you model as being the entire universe or something, whereas I think of my dark forest as being some recesses of my mind that I can't access.

URI: Yes, yes, yes. I specifically think of them as being the Greek muses. I think there are some kind of angry gods.

SPENCER: So you have a bunch of muses over there, and they're just hanging out.

URI: But yeah, I guess one of the frustrations of talking to people is that you have this complicated cluster in your garden, maybe a complicated set of plants, and trying to communicate about that set of plants, this note process that we mentioned, you're throwing a note out of your garden into someone else's garden, and somehow it has lower dimensionality, or something. You can only do it in lower resolution, I guess. How would you describe the difference?

SPENCER: Yeah, I guess my mental model of what happens in conversation is that my mouth makes a bunch of sounds, your brain interprets the sounds, turns them into chunks, and eventually turns them into higher-level concepts. Ultimately, what happens when you turn to higher-level concepts is it triggers a huge set of associations. If I say the word "duck," you might, depending on what sort of mind you have, imagine a visualization of a duck. You might hear the quack of a duck. You might also have loaded conceptual elements connected to ducks, like ponds. It all is kind of popping into your mind. Of course, the duck you imagine might not look exactly like the duck I imagine. Maybe my duck's bigger and has an orange beak, or whatever it is. If we have enough shared culture and knowledge, there's enough similarity between all the different associations that they're sort of compatible, and that's why we can communicate. But if, when I said "duck," you were imagining a moose, then communication is currently off the table until we resolve that.

URI: There's some set of questions. If I'm thinking of a moose, I've somehow confused mooses and ducks, and you say you're talking about a duck, and I say, "Can your duck walk?" and I'm like, "Yeah, of course it can walk." And you're like, "Does it make sounds?" and I'm like, "Yeah, of course it makes sounds." There's some amount of distance we could go. But then if you say, "Does it fit in a bread box?" I'd be like, "Absolutely not. What are you talking about?"

SPENCER: Then I'd be like, "What's wrong with your definition of a duck?" We can fight over what the definition of a duck is for a long time. But it is actually plausible that I am thinking of an actual duck, but I think it can't fit in a bread box because I'm thinking of a kind of larger duck. You can go quite a long way without being on the same page. This is one of my great frustrations. I actually don't know how to resolve this: how can you tell whether you're talking about the same things at the same time?

SPENCER: Yeah, I think you kind of hinted at this idea that they might say something they couldn't have said unless they understood it. I think that's a good insight. Another way of restating that is you should have some expectation of what responses occur if they actually really get it. Let's say they say something that doesn't quite make sense to you, and then you're like, "Okay, help me connect that to what I'm saying?" Then they should be able to explain it and tie it back in a way that connects to what you mean by the concept.

URI: This gets harder and harder the more abstract the concept is. I see how to do it for a duck, but I don't actually know how to do it for democracy or consciousness, or all of these things that people have to debate.

SPENCER: Oh, my God, consciousness is such a good example because people have such different ideas of what it means. Even if you're using the same definition, my preferred definition is what it is like to be a thing. The way a rock is, there's nothing it's like to be a rock, presumably, but there is something it's like to be a person, and maybe there's something it's like to be a duck. Another way to put it is that it actually has internal experiences. If there's something that experiences redness, then it's consciousness. But even if we're operating under the same definition, I find it's hard to ensure, "Okay, we really mean the same thing by consciousness." I usually spend a few minutes at the beginning of a conversation making sure we both know what consciousness — that we're using the same terms. We're using the same terms.

URI: Absolutely.

SPENCER: Maybe there are just some landmines; there are some things that are so hard to discuss that the first part of the conversation has to be about definitions.

URI: Yeah, well, what's your model? Why is consciousness specifically so hard? What properties does it have to make it?

SPENCER: I think part of it is very abstract, but a big part of it is that there aren't good words in English to refer to it. It's sort of the same challenge you have when you're talking about meditation. When you're meditating, do this thing with your mind. We don't have words for that thing with your mind. That's not something that language is used to describing. You see this in martial arts sometimes; I think this is partly why you get these funny vocabularies that develop. You have a word like mindfulness. Mindfulness is actually a pretty complex word. It's hard to understand what it is. It's because it's trying to talk about this fairly complex state of mind that people are not used to being in, but it is an important element of some types of meditation.

URI: But these are all internal experiences, so I can understand this. This category of anything that's highly internal is very hard to talk about. All kinds of feelings are hard to talk about. It does feel like there are lots of other things that are hard to talk about, but for some other reason.

SPENCER: What's the example of something it's really hard to talk about?

URI: I've heard that children really struggle to learn colors. I haven't tried this myself, but apparently, young children, we always insist that they say, "Oh, my favorite color is red, pink, whatever." But if you point to an object and say, "red," they don't know whether you're talking about the size, the shape, or the color. These kinds of things, I guess that's just an abstraction.

SPENCER: That seems solvable. If you give them a bunch of red things, the only thing that is common is the redness, they could learn that pretty quickly. Yeah, I bet it might be in practice that it's hard given the stimulus they get. It might take a while for them to figure that out.

URI: I once played an incredibly long game of 20 Questions. I want to say it was genuinely four hours long, where the answer was left-handed, so we were trying to get to it using only yes and no questions, and it's just like, "What even category is that in? I don't know what left-handedness is?" But I know it when I see it.

SPENCER: Yeah, so that somehow it's hard to pick out in the space of concepts, or figure out what binary way to cut up the space of concepts gets you there.

URI: Yeah, there's something there about concept space, maybe things that are further from other things in concept space are more unique. I don't know. Does that make it harder to pick out?

SPENCER: What about political topics? Because I think this is something that a lot of people want to have intellectual conversations about, but they're sort of famously unlikely to go really well. I think the two main failure modes are: failure mode one, the person is not on the same page, and it gets heated quickly, and you're just trying to win. Failure mode two, you're very aligned, and it's just agreeing with each other session. The really interesting intellectual conversation there is when you're both open to changing your mind. You have substantive disagreements about politics, and neither of you gets upset. Your relationship is stronger, not weaker because of the conversation.

URI: The soldier mindset version is just super calm. All kinds of words trigger people into defensive postures, and then the conversation goes awry. I remember in undergrad, I made a point in a seminar once, and everyone in the group seemed, "Oh that's interesting," and nodded along. Then I said, "Of course, this is just coming from Hayek," this famous kind of right-wing thinker, "No, no," and everyone suddenly disagreed. Everyone was suddenly very disagreeable because I had attributed the point to this taboo figure. I think that happens a lot, and you can often see that happening. But if we put that case aside. I'm sorry, what was the second thing you said? You said this was the first failure mode: people getting into soldier mode.

SPENCER: The other, well, the other is that they agree so much that they're just kind of in agreement. What about the other failure mode, where people are in the same political group, and they just spend the time agreeing with each other and maybe bashing the other side?

URI: This feels like it's not an intellectual conversation. It's more of a camaraderie situation. It's tribal. You have this where maybe you have a friend who has just broken up with someone, and they're saying, "Oh, this person was terrible in all these ways." And you're thinking, "Actually, they seem quite nice. I think they're fine." But you're on their team, and your job right now is to say, "Yes, yes." That's kind of what politics often feels like: I'm with my people, and we're here for camaraderie.

SPENCER: Yeah, that's what I call light-gassing. I coined that phrase. It basically is the opposite of gaslighting. Gaslighting is getting people to doubt their own perceptions when their perceptions are correct, whereas light-gassing is getting people to believe their false perceptions.

URI: Yes, yes.

SPENCER: You're reinforcing their false perceptions. They're like, "Oh yes, the other side is all evil and stupid," and you're like, "Yes, yes, they are." Or, "Yes, your boyfriend really was the worst."

URI: Light-gassing is great. Okay, so I think those two cases would probably take up most of why politics is hard to talk about in my estimation, but they seem less. I'm interested if there is some version with intellectual conversations where both people are trying to have an intellectual conversation. They're not just trying to agree or disagree preemptively, and yet it still is very difficult to talk about. Do you think there is a third reason?

SPENCER: Or in space in particular?

URI: Yeah, or if you can avoid light-gassing, and you can avoid kind of combative tribalism, you will actually be able to have it easily.

SPENCER: I find that the standard default framing on a lot of topics.

SPENCER: Well, there is this other thing that happens in political conversations that I think can really prevent them from going well, which is that people often adopt the sort of default standard framing of the question, and I think actually trying to shift frames and approach the topic totally differently can often make political conversations go better. So an example of this is immigration. Very often, if you just look at how it is framed in political spheres or by the media, it's often framed as immigration good or immigration bad. And you could have that discussion. You could debate that or try to have an intellectual conversation about that. But I would much rather have an intellectual conversation about a different framing on the same topic. One question, for example, is suppose that we had to let in one person and not let in another. You just didn't have a choice. How would you make the decision of who's getting in? Or another framing on that, which takes a different tack, is suppose you were tasked with deciding exactly how many people to let into the country every year? How would you decide the number? Would it be everyone who wants to get in? Would it be half? Would it be nobody? And how would you make that decision? To me, those are much more interesting questions than the default framing good, bad.

URI: I think your nuanced thinking tool on your website is very good for this and very helpful. I often feel when talking to people that they don't have this concept of marginal thinking. The question is not zero or infinite immigrants. How many? Are we currently a little too much, a little too little. This is kind of coming back to your Minecraft materials. I think if someone doesn't have in their conceptual toolkit this idea of thinking at the margin, or nuanced thinking, or rejecting binaries, then you might have this conversation where you think of disagreeing, or they're upset with you for talking about it this way, but actually, there's just a missing concept that you don't share.

SPENCER: Yeah, I think you're hitting on an important point here, which is that intellectual discussions about binary categories are usually not very good, not very satisfying. Is it in this binary category or not? Because usually the answer is, categories are fuzzy, and have fuzzy boundaries in real life. They might work well for a lot of cases, maybe 90% of cases even, maybe even 99%, but often the ones we're debating are exactly the things that have fuzzy edges, and then it ends up being a secret debate about definitions without even acknowledging it. Or it could be a secret debate about values, depending on if it's good or bad. It's probably a secret debate about values. If it's like, is it in this category or not? It's probably a secret debate about definitions, and neither of those usually are very good. When it's a secret debate about definitions, I think, "Okay, let's just decide on a definition. Then let's see what's left over in a debate. Let's just agree on a definition." When it's a secret debate about values, I think that's also interesting, because you can switch it to a discussion of what values are promoted and what values are harmed by this thing. So let's say one person thinks immigration is good, one person thinks it's bad. You could switch to, let's work together. Let's build a tower. Let's make craft materials together about which values are promoted by more immigration and which are promoted by less immigration. Maybe we can agree on that, even if we have different values, even if I like value one and you like value two, we can agree on which values are promoted and attracted.

URI: There are often people who will deny that there are any downsides to the thing that they like. It's hard to get people to say, "Yeah, this would be better for compassion, but it's worse for justice, and I prefer justice. So, here we are."

SPENCER: That goes back to sort of getting a soldier mindset, because I think if you're in that win mindset, then you're not willing to admit any bad things about the thing you prefer, or acknowledge any good things about the other side. Once you're sort of really trying to see the world as it is, I think hopefully you can say, "'Okay, yeah, maybe there are some good things about this thing that I don't like," and I can acknowledge that without saying, "I like the thing," or, "That the thing is good." I think there's a tricky aspect to this too. There is this self-censoring that happens often, where people don't want to say there's something good about something they think is bad, or something bad about something they think is good, because it sounds like a promotion of the thing, or a demotion of the thing, and they don't want to be associated with that, or they feel like that's wrong, or others would judge them for that.

URI: Do you think that the fact that we generally can't implement our political ideas, there's relatively little feedback in politics? I have all kinds of ideas for voting systems that I would like to implement. I like approval voting. I think it would be great, but I'm not, in my lifetime, going to get to implement it in any way, anywhere. So I actually don't know what the outcomes would be, versus intellectual topics that are closer to your direct experience. You can try things and then, for instance, that kind of habits work you do, or the psychological work that you do.

SPENCER: Do you think that being grounded in things you can try and iterate on maybe? Are you saying that that might make for better conversation in some way, because it's less abstract?

URI: I think potentially it's not even the abstraction so much as the fact that I can get feedback. I could stand here and make claims like, "Spencer, I can never be happy. It's impossible." And then you could say, "Oh, have you tried gratitude journaling? It's a technique." I can go away and try it, and I'm going to get some feedback from that experience, that makes it harder for me to just hold beliefs in the abstract without kind of tethering them to a feedback loop, if that makes sense.

SPENCER: Well, there is that funny thing where, when it feels like a belief will never be tested, it's easier for people to cling to a wrong view without allowing it to come in, like it doesn't need to be right, because it's never going to be tested anyway.

URI: My idea that all my ideas come from the muses, and you're just, "That's probably not true."

SPENCER: But I feel we could probably disprove that because if you think that other people's ideas also come from the muses, then do you think? Well, yeah, are there a finite number of muses?

URI: I believe so. There's nine.

SPENCER: There's nine. Okay, yeah, do you believe you know what the nine are?

URI: No.

SPENCER: Oh, okay, so I think if we can pin down the hypothesis about what exactly the nine are, I think it could be testable.

URI: But wait, in what way? Suppose, whatever it suppose I had said there's only one muse, but she gives everyone all their ideas. What's the test? Why does the fact that she gives ideas to everyone matter?

SPENCER: Well, it totally depends on the nature of this muse. If you then define that muse to be exactly identical to the subconscious of the mind, then of course, you're not going to be able to test it, because there's no prediction you can make that would differ. But if there's any aspect of the muse that differs from the subconscious mind, we could potentially latch on to that and say, "Okay, you're claiming the Muse is like this, well, my concept is like that."

URI: My Muse gives me ideas that relate to ideas I've already had on topics that I've already thought about. Why doesn't the Muse give those ideas to you? Yeah, good question.

SPENCER: Could the Muse give you an idea about something you know nothing about? Why couldn't she if it's just a Muse?

URI: That's very good, yeah.

SPENCER: But I think this is slightly off topic, but I think it generally is very interesting where often people claim to me that something is not testable scientifically, and I think almost always they're wrong. Now, it might be harder to design that experiment, but I think almost always they're wrong. Because all you need is some aspect of the claim that differs from the alternative and some aspect of the alternative hypothesis that has an outcome in the world that can be measured. You don't have to test the entire claim. You say one piece of the claim predicts a different thing in the world than the default.

URI: Is it frustrating in conversations? Are you often going around and people will say, "This is so easily testable, we could just test it and be done." But, you know, you can't say that, or they wouldn't accept that.

SPENCER: This mostly came up when I was working on designing a test for astrology because a lot of astrologers are resistant to the idea that it can be tested scientifically. I had to find advice from them about how to design it because they have a lot of expertise on how the systems work, which was actually really essential in designing a test. So I found astrologers who'd be like, "No, I think it can be tested scientifically so that we can have that conversation." But the ones that said it could, I didn't find their arguments convincing at all, because they claim it does something, it can allow some kind of prediction or some kind of assessment. So if that's true, that data has contact with the world, so we should be able to check in the world whether that is happening.

URI: That's interesting. If I remember correctly, you put something out there, and it was only using the month or the day of someone's birth, right?

SPENCER: Yeah, originally, just using their sun signs, Aries, Pisces. Which is the very most basic way of looking at us.

URI: And they were saying, "Well, this is invalid because you're not using moon signs." And you were like, "Okay, cool, yeah, let me incorporate the moon signs. Tell me how to build this experiment in a way you would like," but it feels like the bar was infinitely moving. And ultimately, a lot of people were like, "Well, you just can't test it." Is that the sort of thing?

SPENCER: Well, there was a mix. We got everything from it's impossible to test scientifically to, "This is awesome. I'm an astrologer. Let me help you." And we ended up getting six astrologers who helped us design our final study, for which I was very grateful. And it's like, yeah, putting your ideas to test is also pretty bold. It's a thing to do that I respect a lot. So it was really, really cool. Actually, our first design experiment was not very good. After the astrological sun signs one, we then did one that was not very good, and we got feedback from astrologers, and they helped us improve it a lot.

URI: So that's super cool. I have one more question for you on this general topic, which is, do you think you've been shaped as a conversationalist by your background as a mathematician? Because mathematics is an area where I think it's unusually objective. It's unusually easy to agree on definitions. Or perhaps that's not true. Let me say that as a question. But do you think in that way, or in any others, mathematics has shaped how you converse?

SPENCER: I don't think so much about how you talk or communicate, but I do think that studying mathematics does facilitate a certain level of precision in your thinking because you have to get it exactly right for the proof to go through, and that's a kind of precision that we get in almost no other type of thinking. Probably some kind of philosophy is like that. Some logic is like that, obviously, but some programming is like that, but there are not that many domains where you have that precision. It also just affected how I think about how I mentally model the world a lot. And so that probably indirectly influences conversations, but not directly.

URI: And I want to say something that I appreciate about good conversations is entering flow states. I realized for the last 20 minutes or so I forgot I was on a podcast. I was just having a really good time. I was very immersed. And yeah, I think that is not a cause of good conversations, but a consequence. It's a sign you're having a really good conversation.

SPENCER: I totally agree. Yeah, that's a great sign. When you're just in the zone, you're not even thinking about the fact you're not going meta. You're just locked in. Absolutely. And that's something that's definitely a consequence of good conversations, for sure. We've talked about individual conversations a bunch. Let's talk about group conversations. I think you and I both have a certain hatred of certain types of group conversations. What is a bad group conversation about intellectual topics? Obviously, there can be other kinds of group conversations.

URI: I'm going to start by apologizing to anyone who's ever invited me to a group event, and specifically anyone who's ever invited me to a book club, or anyone who loves book clubs, because I love the idea of book clubs, but I have a personal hatred of them, and I think they exemplify everything I hate about group conversations. Firstly, you go into a book club, the topic is a book. It's much too big for anyone to grasp in their mind at once. As a group, you start talking about it; one person will say something. Someone might react to that. Bob might react to what Anna said. Charlie might react to what Bob said, but Danielle is still thinking about what Anna said. As the conversation goes around a large group, you often have this kind of stop-start effect. I don't quite know how to describe this, either visually or audibly, but you're just going back and forth all the time. Instead of having this beautiful walk in the forest, you're just standing at one point in the forest, making small movements back and forth. No one is building on what anyone else said. That's my first complaint.

SPENCER: Hmm, yeah, I've definitely seen that happen. I think the version that I find even more disruptive is that someone will have a random thought that's loosely connected, and then they'll be like, it would be like, you go to your first metaphor. It's like, "Let's suddenly walk into the middle of the woods over there, right? Go off the path." So, you could either be making no progress because you're bouncing around, or you're just going in a random direction because someone had a random thought about that thing. I think it's very hard in a large group to sort of make forward progress. I think that can be corrected with a moderator, but the moderator really has to keep it on track. What do you think?

URI: Yeah, I think that sounds right to me. It comes back to this thing where there are infinite great conversations you could have about any given book or any given topic. Whenever we have a conversation, we're kind of choosing one path out of many, and many of the other paths are valid and interesting too. It's partly just the difficulty of collectively deciding on something as a group. When many of the options are good, trying to do all the options at once doesn't work. So yeah, the moderator seems like a great idea, or someone who's leading you in a particular direction.

SPENCER: I think this also relates to the unwritten rules of conversation, which is that nobody ever teaches you the rules explicitly, but we all kind of know that in conversation, you're supposed to say something that at least connects enough to what was said before that people understand the connection. If you don't do that, you have to make an excuse for it, like, "Oh, sorry to interrupt, but I just thought, you know, whatever." Because that is an unwritten rule of conversation, people will extend it to group conversation, which makes sense. But that means the bar is connected enough that people can grasp why you're saying it, but that's not nearly connected enough to make forward progress.

URI: Yeah, so I have an additional model for why this goes wrong, which is that each different person has a different threshold of how relevant their point has to be. This isn't specifically intellectual, but you can imagine someone says, "Oh, I just got back from a trip to Rome." Some people think that in order to jump onto that, they have to also have a story about Rome, "I was just also in Rome." Some people think it's okay if they were just in Italy. Some people think it's okay if they were in Europe. And some people will just say, "Oh, I just watched a German film." I always sit there in these conversations thinking, "What? I was sitting here, trying to decide whether my relatively relevant anecdote was relevant enough." Because of the way these thresholds work, the more people there are in the conversation, the more it will end up being dominated by whoever has the lowest relevance threshold. Whoever thinks that they can jump in at the lowest level will end up speaking for more and more of the time once you get beyond, two, three, four, five people.

SPENCER: Because the other people will just be self-censoring, essentially, yes.

URI: And the same effect happens for people who have different internal thresholds for how long you have to wait for someone to finish speaking. Some people think you can jump in while the other person is still speaking. Some people think you need to wait for a one-second pause, a five-second pause. None of us are doing this consciously, but I think implicitly, we all have these thresholds. In a two-person conversation, once you finish speaking, I'll start speaking regardless. So there's a limit. Even if I don't like interrupting, I'll always have a chance to speak. But in a group of four or five people, if I have a five-second threshold, and anyone else there has a lower threshold, I'm never going to speak. There will never be a five-second pause because someone else will always have jumped in after one or two seconds. So, yeah, this is a big model of mine.

SPENCER: That's one of my biggest pet peeves in larger groups. I like to think of it as a game show kind of situation where the more people there are, the faster you kind of have to buzz in order to speak. Maybe that's okay with four people, but at some point with a bigger group, the buzz time is so low that the only way you get to talk is you have to kind of pre-plan, "Oh, I want to say this thing. I'm going to wait for the moment someone stops speaking, and I'm going to immediately say the thing." Then it's this stressful game of, "Can I even say what I want to say?" Or you just give up, and you're like, "It's not even worth contributing." I find that around seven, it kind of completely breaks down.

URI: Yeah, I don't actually believe you can have a good group conversation with more than four people. I'm willing to test this, but I've yet to.

SPENCER: I will say I do think it's really hard. I think I meant at seven that the buzzer thing is such a problem that you can barely even speak.

URI: I would totally believe that, and I understand it's my fault to some degree. If I'm in a group with a certain norm, I should just adopt their norm. But I find it so hard, and I'm so confused by the people who will dominate a conversation.

SPENCER: There may not be a norm. It may just be, if you're right, and the person with the lowest threshold is the one to chime in, that doesn't mean it's a norm, it just means that's that person's norm, right?

URI: Yeah, that's true. And the more that you're in kind of multicultural groups, in the broader sense of culture, people who have different norms coming together, then there's some chance that there'll be one person whose family norm, or whatever it was, was just constant interruption.

SPENCER: Have you ever had an excellent group conversation?

URI: Oh, yeah. I think I've had excellent group conversations. So what makes it enjoyable?

SPENCER: So what makes them excellent? Because we both feel that typically intellectual group conversations don't go that well. But the ones that have been excellent, what would you point to?

URI: Yeah. I think it takes a lot of the awareness trait that you mentioned. It needs everyone to have that awareness trait. This is a situation where one person who keeps interrupting or derailing the conversation will ruin it. It's no good to have five great conversationalists and one bad one. You need everyone to be on point. I think it helps if you have context on each other. I think this thing we spoke about earlier, where if you know someone well, you can understand what they're saying with fewer words. You can kind of guess what they mean. You can check that you're on the same page. I think that's very helpful, though not entirely necessary. What else do I think? Do you have anything on your list?

SPENCER: Yeah, I would add. I think it helps a lot if everyone's really, genuinely listening to what other people say, so that people want to know what other people think about the thing. That makes it more of this, like we're building a thing together, because everyone's asking, "What do you think about this?" or "What is your perspective?" Then it's sort of like you're brewing. You're putting things in. If people are focused on how they look in the conversation, like, "Am I impressive enough?" or "Do people think I'm smart?" I think that's a recipe for terrible group conversations.

URI: I think I've often had good group conversations where there's some personal element so that everyone can bring personal stories. Those are indisputable. It's like, here are some experiences I had, but that doesn't strike me as strictly intellectual. I wonder whether that's less intellectual. I'm trying to think about the best group intellectual conversation I ever had.

SPENCER: Because there are a lot of great ways you could interact in a group. Maybe you could tell funny stories or laugh together. We're really just talking about intellectual conversations where you're trying to make progress and learn and think about a topic.

URI: A thought that strikes me now is that a lot of my best group conversations are in text, which has this interesting property where you can both be writing at the same time and then both hit send. So maybe there's something about that that's interesting.

SPENCER: I don't think I've had that experience of amazing group chats.

URI: I love group chats. I actually think they're my favorite form of social media. I think a three to six person text group is one of my favorite social activities. My favorite social media. I learn a lot.

SPENCER: What do you like so much about it? Besides feeling like you learn.

URI: I think, yeah, it's kind of asynchronous. You can jump in whenever you want. If someone says something, you can jump off it. You can keep multiple threads in the air better. It's not ideal, but you can have two threads going where people are interlacing responses on different topics.

SPENCER: I really don't like it, but maybe I just haven't had really good ones yet. I think they're great for communicating with a group. If you're coordinating, but I don't think I've ever had a really great intellectual group conversation with more than three people in a thread that I can think of.

URI: Yeah, okay, what else is there in group conversations, and what makes them great?

SPENCER: Well, I think there's also a personality element here. I think you kind of mentioned that if one person can ruin a group conversation, that's a pretty brittle failure mode. Because, "Well, you've got to curate the whole group," and I think there are quite a few ways that can fail. One can be over talkativeness. Someone might be a really nice person, but they're just so talkative. If someone talks 70% of the time, and there are seven people there, it's really tough for everyone to enjoy it.

URI: I'm so intrigued, confused by the people who dominate conversations. It doesn't have to be exactly equal. It's not like, if there are n people, I must speak for one over n of the time. But there are people who will speak for 7% of the time in a five-person conversation. I always wonder what's going through their head. Do they think they're the most interesting person? Do they think that everyone else values what they say more than they would value hearing anyone else speak? Do they think no one else has anything to say? I once had a friend who would dominate one-on-one conversations, and I eventually was like, "Hey, you don't always let me get a word in." She was like, "Oh, I just thought you, you know, I felt like I was carrying the conversation. I felt you never had anything to say." I guess we had these different disruption norms; she was waiting for me to interrupt. I never interrupted. She thought, "Wow, I'm doing all the work here." I always wonder what's going on inside, because it's a huge pet peeve for me if someone takes up a significantly larger share of the conversation.

SPENCER: Yeah, we actually ran a study asking people what percentage of time they prefer to talk versus listen. We found that the vast majority of people really prefer to be speaking between 60% and 40% of the time. There aren't that many people who are like, "Oh, I want to be speaking 10% of the time" or "90%," but, yeah, you do go to a party and you'll find people who seem to monopolize the conversation. I think there are quite a few different things going on. Sometimes it's driven by a lack of social monitoring, like we were describing; they literally are just not running that loop of, "What's happening right now and how do people feel?" So they're just focused on what they're saying. They're not noticing. "I've seen that." A second thing is attention-seeking. Some people just love the spotlight, so they're getting joy out of it. They're just seeking their own pleasure in that experience. Narcissistic people tend to do that more often. A third thing, and this is when I fail at this, is that sometimes I'm just really excited about what I'm saying. During that excitement, I kind of forget how long I've been talking, and then I have to be like, "Oh shit, I've been talking for too long. Okay, gotta stop," and I'll try to tone it back and talk less to make up for that.

URI: I've seen people do it a little bit. Maybe I should pay more attention to this.

SPENCER: But I've had times where I've definitely spoken too much in a conversation.

URI: I guess I've never noticed someone speaking 50% more than they should. Either someone just dominates the conversation so much that it's incredibly noticeable. I've rarely met someone where I'm like, "Oh, I feel like you speak a little bit too much, but it's within bounds, but I wish you'd speak a little bit less." I feel like I don't remember having that experience.

SPENCER: I think there's a bell curve on this, and you probably just don't even trigger it. So I think there are a lot of different things going on here. Sometimes I think people actually believe, oh, everyone's having a great time, and there's actually more fun for the other people, right? They could be right, or they could be wrong about that; they might think, "Oh, everyone's super interested in what I'm saying."

URI: I'm actually getting physically activated. I hate this thing so much that I'm getting worked up just thinking about it. I think there's a small number of people who are so interesting that it's tolerable to hear them speak. But even if they're the most interesting person in the room, I still don't want them to dominate the conversation. I'd still rather hear other people speak, even if those other people are less interesting.

SPENCER: And why is that? Is it just because it feels more egalitarian, or do you actually enjoy it more, hearing more voices, even if they're less interesting?

URI: Yeah, so it's possible that I just have a strong fairness instinct, and that this is an almost childish need to share things and be fair. But I do think that I may be doing the exact thing you mentioned earlier, where you think that everything is good about the thing that you think is good. It may be that actually, I'm just confabulating here, but I tend to think that the conversations are also better. It's both better from a value point of view, and the conversations are better because I think there's diminishing marginal returns to any given person's input. One of the things that drives me nuts about some people who talk too much in conversations is that I often feel like I got their point at the beginning, and then they'll talk past the point. They will repeat the thing completely unnecessarily two or three or four times. I already understood. I was already ready to say my thing, but they won't let me.

SPENCER: I always had a professor who said that, I think it was the first day of class. He was like, "Just, you know, almost always after students make a point, they tend to keep talking. It's really much better to stop talking after you say that first thing." I became much more aware of it after he said that. People often say the thing that they mean, and then they just keep talking afterwards. I'm sure I do it too, but it's just a funny quirk.

URI: So this is the thing, in some sense, we've both just done it. And yet it's hard not to do it at all, but it seems bad to do it excessively or something, right?

SPENCER: Yeah, some redundancy probably makes sense. You want to make sure people really understand it, but a lot of it seems like just a kind of funny language quirk that people have.

URI: I've had a couple of times in this conversation where I said something I know that I should be done. I can see in your face that you've got a thought and I can't stop myself.

SPENCER: Can be difficult, yeah. Well, okay, so I went to this event that was, we're gonna have you answer questions, and we're gonna analyze it, we're gonna set you up for meeting people we think you'll find interesting. It was a kind of friend matching. But I was like, "Oh, that sounds intriguing. I'm curious." They were gonna curate a group conversation. I was like, "Cool, I'm down for trying a weird social thing, sure." At the event, there was one very friendly, nice extrovert. I was so glad they were there because they saved it from being a super awkward hour and a half dinner or whatever. So glad that they were talking a bunch, but I think it's because there was a lack of gelling. The group wasn't cohesive, you know, it wasn't working. We needed someone to just keep the conversation flowing. Yeah, I could do that, but it'd be really tiring and effortful. So I was like, "Oh, thank god this person is handling this."

URI: There's definitely a context where I'm glad someone decided to entertain, but it's always a context where I didn't want to talk to the other people, and it's just, how do we fill this time? If there's someone who could talk and tell stories.

SPENCER: It's like, "Yeah, it's saving a particularly bad situation rather than what you want to aspire to."

URI: I'm wondering how much this thing that I think I hate about group conversations is really just that I immediately exit one-on-one conversations with people who I have this issue with, but in groups, I'm more likely to end up in a group with people I can't escape.

SPENCER: I remember you once saw me escape a group conversation you ran up. How did you do that?

URI: Yes, it was inspiring.

SPENCER: So, buddy, I have a technique for leaving group conversations. I think it works really well. Some people might disagree, but the thing about a group conversation is that you're not the focus unless you've just been speaking. So what I do is I make sure I haven't spoken for a minute or two, right? I'm not the focus of the conversation. Other people are talking. Then I move my body language slightly further, just a little bit outside the other people, which generally is not noticed by anyone because I'm only very slightly outside of the circle, but it tends to make people close in a little bit. Then I will sort of make a hand gesture and just say something like, "Excuse me," or "See you later," or something like that. Then I'll make a little hand gesture to the group and walk away. Although, in theory, it could be awkward, I just never experienced it as awkward. It seems to work fine. They just keep talking; it's like they don't even notice it, and then I'm out of the group conversation.

URI: Yeah, genuinely an act of genius. I have copied it many times since.

SPENCER: What do you think? Does it work? Does it feel awkward?

URI: Essentially, I find it surprisingly easy and enjoyable. If you're in a room with a bunch of standing groups, there's a party, and there are different groups standing. The conversations I get stuck in, and find it harder to leave, are when you're seated and there's a fixed group, whether it's dinner.

SPENCER: Oh there's only one group.

URI: There's only one group.

SPENCER: You want to leave, like you are in a home. Is that what you mean?

URI: No, I want the conversation to go back to being good. The thing that most often happens is...

SPENCER: That's a totally different problem.

Yeah. There are three-four people conversation. Someone joined, and they started dominating. Or you start with three or four, and it seems promising, but one person just dominates the whole thing, and I don't have the social skills to navigate back to the direction that I believe most of the group wants. It could just be me, but I think that's where we wanted to go. I find it harder to leave, though, so I should probably just leave more often. I should just be like, "Well, I'm not enjoying this, I'm gonna..."

SPENCER: You could leave. One thing I sometimes do is fragment the group in those cases. However, you do have to be a bit careful because some people don't like that. But if you get the vibe that people are okay with it being fragmented, then I'll start a conversation with the person on my left or whatever, and then try to create a two-person conversation or try to split the table in half. Yes, because if it's six people, you might be able to split it three and three.

URI: Right, right, right. The physicality of it matters surprisingly much. I probably need to learn better body language. Maybe I should go learn from a salsa teacher or something. How do you create physical space that implies a new group formation?

SPENCER: Absolutely. Yeah. If you want to split half the table, then you want to talk to the person you like, essentially with the person you're talking to, create a barrier momentarily, and then you can make your body kind of mimic that to some extent. When that person responds, you now have sort of a segmented conversation momentarily. Often it will stick, not always, but often it will stick that way, and it will kind of keep going in that direction. It's funny, though, because some splits don't work. There are certain splits that you can make happen and others that are hard to facilitate.

URI: Something else I've seen you doing in groups. I think you mentioned this is jumping to a new topic completely. But I think you have better techniques than most people of just saying, "Hey, we've gone down this conversational alley. I want to move to a new topic." I think you don't pretend that it's on the same topic. I think you're much better than most people at just saying, "Hey, on a completely different note. I think that's sometimes what a conversation requires."

SPENCER: Yeah, I think the smoothest, the most comfortable thing is just to follow the normal social rules of, "Oh, I'm going to say something that's related," but if you really want to bring up something different, it's actually better just to acknowledge that. It is slightly awkward, slightly uncomfortable, but I think it's just better. It's actually the least awkward to sort of acknowledge it and be like, "Hey, I have this really random thing I wanted to get your thoughts on. Do you mind if we talk about XYZ? Hey, I was thinking about this the other day. I wanted to see what this group thinks of this," and people I find are pretty cool with that, as long as it's acknowledged that you're kind of deviating from the standard social norm.

URI: You're acknowledging it and kind of asking permission. If someone had one more thing to say about the previous topic, they would be able to say it given the way you presented it, which is a big difference from someone who just suddenly changes the topic midstream without doing that.

SPENCER: Yeah, if you're talking about one topic and someone just suddenly asks, "What do you think about hamburgers?" Everyone will be really thrown off and confused. So you have to make it clear that a transition is occurring. Asking permission is nice. I think there's also this element of taking responsibility for a group having fun. I don't know about you, but if I'm in a small group, even if it's not my event, I feel some responsibility. If I'm in a group that's having a bad time, I feel like, "Ahh, I should do something." What do you think about that?

URI: Completely agree. A class that I took in college that I did not expect to matter at all was called public speaking. It was one of those classes that people took as a kind of fun fourth class on top of their others. It's actually one of the ones that I've thought about the most in the 15 years since. This professor, Tamsin Wolf, gave us this advice: always believe and act like you're the host of a dinner party. I hope I'm remembering this right. She said, when you go into job interviews, imagine that you're the host of a dinner party. Your job is to make these other people comfortable.

SPENCER: When you're interviewing them, or they're interviewing you?

URI: Yes, yes. When you're being interviewed.

SPENCER: What?

URI: Well, I think it's interesting because she said, most people — let's say my memory of her, which may or may not be accurate, but the version of her speech that exists in my head now — said that most people go into job interviews and the interviewer seems annoyed with you, or you think they're bored or whatever. If you treat it in the normal way, you get nervous. You think, "Oh, I'm doing something wrong. They don't like me. I'm not going to get the job, whatever." But the idea that I took away was, "No, if you go into that situation and act like you're the host of the party, your job is to make this person feel comfortable." Currently, they seem kind of irritable. That's not great as the host, so I'm going to try and charm them. I'm going to try and make them feel comfortable. I'm going to be accommodating, polite, friendly, and warm. I do think that in most situations, that framing seems to work. So yes, I can imagine a situation where that comes across as improper or like you're taking too much power on yourself or something. But I found that in most situations, that framing seems to work.

SPENCER: That's interesting. I think part of that is taking agency, that the conversation is not happening to you, but you're like, "Oh, I am responsible for the conversation." If the other person is not having a good time, or seems annoyed or bored or whatever, I can do something about that. I do know some people are sensitive to other people stepping on their hosting. When they're the host, they want to be the host and don't want other people to act like the host. So I'm like, "Ugh, maybe I'm around them. I don't know if I would do that."

SPENCER: Okay. As our final topic, before we wrap up, if you were to make a suggestion to people listening to this on how they could have better intellectual conversations, what would you have them think about? What would you have them focus on?

URI: Okay, so I think this is a situation where just the desire to have it is really important. I do think Julia's book Scout Mindset, that concept is really important. Being able to think in that way using some of the Clearer Thinking tools is also very helpful. But once you have the desire and the rough skill set of openness and curiosity, I honestly think just finding one person you really enjoy talking to and talking to them a lot is really helpful. I suspect that this is actually one of the big benefits of university, getting some time with other people and finding some people who you enjoy talking to at length. I think it's the best thing about dating. I'm obviously a conversational obsessive, and maybe I'm assuming too much that other people share my goals here. But for me, this is one of the great human experiences: thinking out loud with another person. If you can find someone that you really connect with, you can go far just talking to one or two people, whoever it is that you converse with best. This is, to go back to value, slightly in tension with more egalitarian goals. There are other reasons to talk to lots of different people and to try and talk to people you disagree with and people who you don't easily speak with, but if you can find someone who you gel with really well in the conceptual sense, it's not particularly helpful if you're talking to someone where you just agree all the time and you're constantly telling each other that you're right. But if you can find someone where you share the vibe, the energy, the ground rules, the shared understandings that let you have complicated intellectual thoughts together and build on each other, I think that's really great. That's a great place to start.

SPENCER: That's good advice. I find that there are people who just love intellectual conversations, and you could talk to them about a really wide range of topics, but that can be harder to find. Sometimes the best you can find is someone who is passionate about talking intellectually about a particular topic, and then you might need to piece that together topic by topic with different people.

URI: I'll add one more thing, which is, sometimes I have amazing conversations, and afterwards I can't recall what I thought or said. I'm left with the feeling that it was a great conversation, but no specific recollection of what I've learned or thought or changed my mind about. So I think either just reflecting in your head about that after having a conversation, trying to remember, "Oh, what did I learn? What did we say? What did I think about?" Or writing it down, those can be really useful, too, to help you develop the tool set.

SPENCER: That's cool. You could even recap it with the other person, give them a message or whatever. One trick that I found, it's very easy to do, but it can help improve intellectual conversations. If you have a friend who loves intellectual conversations, before you meet up with them, be in advance. Be like, "Oh, here are some things I was thinking about. I'd love to talk to you about it." I find that can be really cool because it can get you thinking in advance about what you really want to talk to this person about. It can be fun having a kind of random path dependency of conversations, but it can be cool to be a little more proactive. Like, "Do I really want to talk to this person about?"

URI: The founder of Intel says that all one-on-one manager-employee meetings should be one hour long because the first half hour you don't get anywhere, and if you make them half an hour, you'll never get anywhere. Only after you've talked a bunch do you then go, "Oh, actually," and then you get to the real thing. I think this technique of preempting by saying, "Okay, we're going to beforehand say, 'Oh, I'm thinking about this,'" just increases the chance that you'll get into the flow of a really good conversation. That's qualitatively different. I think having a long block of time where you really explore the ideas and grow them, and talk about the things that really matter is crucial.

SPENCER: What if someone wants to have a great group conversation, should they just give up?

URI: They should do some experiments and then email me and let me know because I'm still working on this one. Do you have any? Be very careful about who you invite.

SPENCER: Curate the group, I would say six people max, maybe four people instead. Yeah, that's even better. And have a moderator. Make it understood that there's a moderator. Make it understood that one person's job is to stay on track. "Oh, we're here to talk about this thing." It doesn't have to be that way, right? It could be free form. But if you want to make progress on a topic, then say, "Oh, I want to talk. Today we're going to talk about this." This person is in charge of making sure we stay on track. They'll just direct, acknowledge when it's gone off track, and say, "Okay, let's bring it back to the main focus." You could even have sub-themes that you want to address throughout the night on the topic. That can be fun to bring it back to the topic in different ways and different themes.

URI: I think that's really great. Because, as I said earlier, I think there are many good conversations in conversational space, and if everyone understands we're going in this direction, this is where this evening is going, and these are the subtopics we want to hit, then it allows you to collectively go in the same way. And that's really great.

SPENCER: Uri, thanks so much for coming back on the Clearer Thinking Podcast.

URI: It has been such a delight. 300 episodes, incredible. How does it feel?

SPENCER: That's pretty surreal that I've recorded that many.

URI: Looking forward to 300 more.