JOSH: Hello, and welcome to Clearer Thinking with Spencer Greenberg, the podcast about ideas that matter. I'm Josh Castle, the producer of the podcast, and I'm so glad you've joined us today because today's episode is going to be a little bit different than usual. You may have noticed that this is our 100th episode. Woohoo! We on the team want to express our heartfelt thanks to you, our listeners, for listening each week, for rating and reviewing, and for interacting with us over email and social media. Your enthusiastic support has enabled us to grow which, in turn, gives us greater ability to invite new and interesting guests on the podcast and thus to bring you even more ideas that matter. We also want to say thank you to the amazing guests who have come on the podcast so far. Your interesting and thought-provoking ideas have helped to bring the show to life and make it what it is. Finally, I want to give a shout out to our production team, Uri Bram, Ryan Kessler, and Janaisa Baril. Your tireless work behind the scenes is what makes this whole thing possible. So, in celebration of having released 100 episodes, we've changed the format up for this episode just a little bit by reversing the roles. Spencer will be in the guest seat and Uri Bram will play the role of host and interviewer. Uri is the podcast factotum by the way, and he was on the show back in episode ten. In their conversation, Uri and Spencer discuss Spencer's approach to conversation, utilizing listener feedback, and the podcast as a medium more generally.

URI: Welcome, Spencer!

SPENCER: Thanks so much for having me. What a role reversal!

URI: It is an honor to have you on the show! This is the 100th episode of Clearer Thinking podcast, and we wanted to talk to you about how you hold conversations and how you think about podcasting.

SPENCER: It's fun to be on the other side of the conversation.

URI: I'm glad. Something interesting about your show is the way you handle conflict and harmony. Can you tell us a little bit about that?

SPENCER: Yeah. I am a very agreeable person in the sense that I always want people around me to be happy. I don't like to make people angry. I think that tends to come through when I'm interviewing people. I want the person I'm interviewing to be happy while they're talking to me. That, of course, has both upsides and downsides. It has the upside that, obviously, it's appealing when someone wants you to be happy. The downside is that some disagreement can be inherently conflictual, or at least it can feel that way. So it can have the downside that I may not challenge guests enough. In our listeners' survey, one of the common themes was that people wanted me to challenge the guests more. I've actually been making a concerted effort to do that, since the listener survey.

URI: That's really interesting. Can we just clarify quickly, when you say agreeable you mean that in the Big Five personality sense, right?

SPENCER: Yeah. In the Big Five sense, I'm probably a 78th percentile agreeable. So I'm not 99th percentile but I think I'm above average.

URI: What is agreeableness consisting?

SPENCER: It has different components. One of them is empathy or compassion, another is politeness. But I think for me, the trait that I tend to have a lot of is just wanting the people that I'm interacting with to feel good while we're interacting, and never wanting to create negative mental states in people.

URI: That's really interesting. We put out a call for questions to ask you in this episode to all the listeners of the podcast in celebration of Episode 100. We had a couple of questions that I thought sat really interestingly together on this topic. One listener, Annie, noticed that you tend towards agreeing with your guests, as you said, and that you seem to prioritize harmony. But another listener, Joseph, said that you seem to have a healthy sense of skepticism. And so he makes a claim, you're inclined to ask questions or challenge their assumptions until you understand it and then form an opinion of it. I want to ask you how these two things hold together. The second one seems a little bit in contrast with the first, but I do think both are actually true of you.

SPENCER: It's an interesting dichotomy because I am very agreeable as a person. But that doesn't mean that I actually agree with people. In fact, I'm extremely skeptical, generally, about what people say. So I'm agreeable in the sense that I want people to feel good, but I'm skeptical in the sense that I tend to not believe ideas that don't gel with existing evidence I'm aware of. Basically, what I often feel like I'm navigating during podcasts is, a guest will say something, I won't agree with it but I'll try to flip into a mode of just curiosity, of "Okay, that's interesting. Can you tell me more about why you're saying that? Or what are the justifications behind that? Or how did you come to believe that?" Really what I want to do is not necessarily disagree during the episode, but explore the reasons behind what they're saying to get to the point where the listener can actually make their own judgment. Because I don't think it really matters so much if I agree with the person or not. What I really care about is that the listener gets enough information that they have an ability to judge.

URI: Do you feel a tension there or a trade-off? Is it hard to also present what you see or why you see things the way you do?

SPENCER: What I try to do is say things like, "Hmm, interesting. I would have thought that..." Or, "What would you say to the argument that...?" I think what I'm trying to do with statements like that is, without creating a sense of 'we're debating each other,' which can be a fun thing but it's really not what I'm going for on the show. I really am going for, 'we're exploring the truth together,' not 'we're debating each other.' So while trying to avoid triggering that 'we're debating each other' feeling, I'm trying to give a counter position that they can then use to explain their view more.

URI: This is really interesting. I remember the very first time we Skyped, just on a personal call. You pointed out that I was wrong about something. I was making a claim about Bayesianism. It was just factually incorrect. I misunderstood how Bayesianism works. But I just remember that we'd never talked before. You're a stranger but you made to point this thing out to me, show me that I was wrong about it, but in such a nice way that I didn't feel either threatened or belittled or anything, but also without being 'patronizing me' nice. I think if someone had done the same thing, but had just seemed like they were trying to be too kind, I would have also been like, "Ugh! They think they're better than me." I was really, really impressed. I remember after that call, I was like, "Wow, that guy is doing something right." Do you know what that is? Is there some secret to that?

SPENCER: Well, I'm really glad to hear that. I had this challenge that happens to me where people will sometimes contact me and ask for feedback on their project. They'll reach out and say, "I would love to know your thoughts on my project, get your feedback." Sometimes I agree to do this, depending on the situation. Then I'll go through their project and I'll be thinking to myself internally, "This is a disaster. I do not think it's a good project." Then it's like, how do I communicate that in a way that's not so devastating and try to encourage them to make it better? I don't want to demoralize people or make them feel bad. I just want to provide what's useful. I feel like it's fairly difficult riding that line. But I generally try to resort to saying things like, "Well, what do you think about this?" or, "I'm concerned that one of the challenges you might face is that..." doing it in ways that are less likely to create defensiveness. Now, I don't always succeed at that. Sometimes people do get defensive, but I really try to avoid triggering defensiveness because I think it just brings the conversation to a very unuseful place very quickly.

URI: There's something really cool there about empathy, combining with the search for truth. Caring about exploring the truth together but also caring about how people feel, which is clearly such an input into whether they're able to constructively and fruitfully explore the truth together.

SPENCER: The reality is a lot of new ideas are gonna be wrong. A lot of new projects are gonna be dumb. They're gonna have flaws. But through constructive feedback, maybe you can make them better and make them more likely to succeed, and so on. If someone's asking me for feedback, I don't care per se, and the only reason to give feedback is if you can help them with their project. So if you just give it in a way where they're not emotionally able to receive it, then you've defeated the purpose. So how do you give feedback in a way that people are able to receive it? I think the way to do that is to really strongly communicate that you're on their team as much as you can, which I don't always perfectly succeed at. But that's what I try to do. I try to give the vibe 'we're on a team together trying to figure out how to do this thing really well,' and like, "Have you thought about this? Well, what about that? Could this be a problem?" and so on.

URI: On that note, is there anything I could say right now that you would just absolutely call me out on, to which you would say flat out, "No, no. That is wrong"?

SPENCER: About what? Just anything?

URI: Yeah. If I said, "Left-handed people are bad at maths," just in the middle of my podcast conversation with you...

SPENCER: If you said that, I would be really intrigued about why you said that. I would try to flip into a curiosity mode, "I'm surprised by that. What makes you think that?" Sometimes something that happens on an episode is someone says something I know is a factual mistake. A lot of times in that case, it's literally just a mistake. They didn't mean to give the wrong number. They were just misremembering or something. In those cases, I will say, "Oh, hey, are you sure about that number?" Most of the time in that case, we'll just take a pause to check it. If they were wrong, they'll just immediately correct it and we'll just redo that segment because there's no point just putting false information that they didn't actually believe as soon as they looked it up. That's the really clear-cut case. I think where it gets a lot trickier is when it's in the realm of, "Well, you can't just look it up. It's not like there's an agreed-upon answer."

URI: I think you seem to have a really good recall of facts and that always plays in quite interestingly. I think you often manage to bring in a statistic or a piece of evidence about something that the guest is talking about but that most of us don't have that strong recall of. It feels to me, for reasonable people, it's easiest to say, "Oh, yes, I was wrong on a matter of fact," or, "Oh, I misremembered that number." I'm trying to think what are the things that it's hardest to admit you were wrong about. What are the things I would be most defensive about? I guess I feel most defensive when I feel misunderstood. So maybe so long as I feel like the other person understands, I don't feel so defensive even if they disagree with my broader claims.

SPENCER: What if you're making a claim that's related to your identity? You wrote a book on Thinking Statistically. So being challenged on Bayesianism, that can be actually quite difficult because it's maybe tied to your identity, right?

URI: Yeah. I do think I'm quite unpossessive about statistics. I think of myself as not very possessive about any parts of my identity. But honestly, that's probably a lie. That probably just means I'm not noticing the parts of my identity I feel the most defensive about. Have you ever been riled up by something a guest said? Or felt attacked?

SPENCER: No, I don't think so. I tend to be extremely low in anger. Almost nothing makes me angry. One of the most dramatic examples of this is, I really don't relate with the idea of reading Twitter as being a thing to make one angry. I can't relate to that at all on a personal level. I know that it does make a lot of people angry. So I accept that as a fact but I don't think I've ever had that experience. The closest thing I have had as an experience is, sometimes when people use really bad logic to argue against me, it actually makes me pissed off. Someone's telling me I'm wrong but they're making really awful arguments. For some reason, that riles me up.

URI: I get somewhat riled up by Twitter for sure. I think it's partly just not being clear enough about who I am. I tend to think that the clearer you are about your goals and where you're going in life and what you want to do, the less you're offended or threatened by other people succeeding or other people seeming to get a claim for something. But then I do also just get annoyed when other people seem very confident about something. I'm fairly confident that that's not how it works. But I can't convey that to them and they can't convey it to me. Maybe that feeling that both of us think the other person is being somehow unreasonable is like a major stress for me on Twitter.

SPENCER: In my own case, one thing that helps is that I don't view anyone as being on the opposing team. So there is no opposing team to me. Because there's no opposing team, there's no "Oh, that's the bad guys. They're saying this thing."

URI: Yeah, I will try to be more like that. Well, this is a good segue. When we were first prototyping the show, I suggested doing some experiments of different kinds of conversations, including one more jokey inflammatory kind of conversation, and you were not having it at all.

SPENCER: If people are curious to hear that, that's episode ten. You were definitely bringing in some more controversial things you were trying to push, I think, basically, trying to challenge the listener in a way that I don't normally do, in a way that might rile them up or something. How would you describe that?

URI: I think riling up is right.

SPENCER: I think, just to clarify, I want to challenge the listener intellectually, but I think you were trying to rile them up which is a different kind of challenging.

URI: Well, how did you feel about that? It seemed like you really didn't vibe with it. You actually didn't like it as an option for what the show could be like.

SPENCER: Well, I think one way to get attention is to overstate things or say things in an extreme way. But I find it quite unappealing. I do think it's worth trying to craft an idea in such a way that people can consume it easily and that it resonates. But I don't like trying to craft it in such a way that it's exaggerated, so as to be more emotional or so as to make people want to fight against it or share it, or other things like that.

URI: I really respect that. For what it's worth, I dislike that direction in the internet as a whole. It does feel like we've reached this evolutionary process where the most outrageous people and the people who do the most riling seem to get the largest audiences and I actually really love that you've pushed against that. I also want to say that episode we did, it was about universities and about the charity GiveWell. You just did an episode with Elie Hassenfeld who runs GiveWell. One thing I love about GiveWell is that they're so nice. Even when I was trying to rile them, they were just very, very polite about it and very appreciative of the people who were discussing their work. I think that's super admirable.

SPENCER: Yeah, it was really cool. We gave them the opportunity to comment on the episode and I thought they were extremely nice about it. I appreciated that.

URI: In real life, I think of you as really playful. Often in social settings, you'll come up with games or weird social experiments. But I think you do that less on the podcast as well. Not necessarily in a conflict direction but just in general, I don't see you doing bizarre episodes very often. Can you tell us a little about that?

SPENCER: That's a good question. One time, I met a guy who I had talked to a bunch of times online but we'd never met in person. The first thing he said when he saw me, he's like, "I don't understand, Spencer. You're such a weird person, but you look so normal." Well, he was referring to just, I was dressed in very normal clothing and nothing about me indicated an unusual person. I do think that I tend to have unusual ways of thinking about a lot of things. But when it comes to the podcast, I think one of the things that's going on is that it has to be a container for all different kinds of guests. I think that's a challenge because some guests I bring on are essentially just friends of mine who I think have interesting things to say. Some guests are people I've never met, who are in the public eye, who have their spiel that they tend to give and their talking points. And then some guests are maybe interesting people I heard about, but they're not used to doing podcasts, but they're working on an interesting thing and they don't know me. I'm trying to navigate all these different types of people and I think that makes it harder for me to act in a banter-y way, the way I might interact with you one-on-one, where I might show certain quirkiness as aspects of my personality.

URI: Yeah, that's interesting. I often think that I'm quite a high variance person. I meet people and act weird and some people respond to it well, some people respond to it badly. The people who respond to it badly, I'm less likely to see again. But I guess, as a host of a podcast, you have to maybe play a lower variance strategy and do things that will make anyone feel comfortable.

SPENCER: Well, I feel like banter is a two-way interaction. Just to give an example, there are some great podcasts where there are two hosts and the hosts have amazing banter. It is very playful and silly or whatever and that can work amazingly well. But it's tough when you have a different person on every time because you can try to banter with them and then they're just giving you a very serious response and it doesn't work. Now I'm not saying it couldn't work. Maybe someone more skilled than me can make it work. But I think it'd be a big challenge for me. It's a lot easier to bring a more serious tone because everyone can deal with that. Whereas, if I bring a more playful tone, I feel like some percentage of my guests are just gonna not jive with the energy they're bringing.

URI: Yeah. Energy alignment in conversation seems really important to me in personal life. Even more so on shows, even just matching the speed that someone else likes to talk, or their energy, or how much they like to get interrupted or not. I find this really difficult in real life. Then on a podcast, it's even more so. This person you don't necessarily know, trying to figure out in real time how to get in sync is just a really interesting problem.

SPENCER: Yeah. There probably is someone who's more skilled, who has figured out how to banter even with someone who's not a banterer. They still make it work. I would actually like to do more of that. I think actually another aspect of this is that I think I can occasionally be funny or silly in a way that's appealing, but I'm not confident I can do it consistently, certainly not on the expectation of doing it on every episode.

URI: It's a really hard thing to do. I think when you see professional funny people, you don't realize how incredibly hard it is to show up every day and be very funny on the spot.

SPENCER: Yeah. Maybe when people go on The Late Show or something like that, at least there's an expectation of the type of conversation style that's gonna occur, and that probably really helps. So even if the guest is more serious, they know they're coming into a situation where there's gonna be jokes and playful banter.

URI: I think an interesting problem about our world is that norms are really important. We're breaking out of smaller communities. We used to grow up in a community and understand that community's conversational norms, and then know what to expect in all kinds of settings. As communities break into other communities and we come to be part of this larger whole, I think a lot more of those kinds of subtle or implicit norms get broken down and you have to do things that will work with anyone that comes along. A lot of conversation podcasts focus on a piece of media, a new book someone has written, some written corpus that they have. But this isn't the case with the Clearer Thinking Podcast, I never get the sense that I'm missing out on something in an episode because I haven't done the reading previously. How do you think about that and what impact does that have on the show?

SPENCER: I don't really like episodes that are focused on books as a main mechanism. The reason for that is I feel like it's inviting the guest to just talk about the bullet points of their book that they probably talked about 50 times before. And that's bad for two reasons. One, because you can get that content in a lot of places. It's not very unique. The second reason it's not good is because it gets the guests into a mode of regurgitation which I really don't like. I try to keep them out of that mode. Now of course, they're gonna do that to some extent if they're someone who talks about a topic a lot. But I'm really trying to get the guests to be generating things that they haven't said 50 times before. I'm perfectly happy to have a discussion about one of their books. That seems totally fine to me. I just don't want it to be a regurgitation of their talking points about their book.

URI: Right. There is something funny when someone is doing a book tour and they do 30 different podcasts. If you're listening to enough different podcasts, you're like, "Hey, wait a minute, this is just the same person saying the same things." I also noticed that myself when I'm talking about something I've talked about too many times and I fall into the patter, the pre-arranged patter. It's clearly less interesting and more rehearsed. I don't enjoy it as much. I feel like I'm just saying something I've already said so I try to speed through it maybe.

SPENCER: One thing I really struggled with, with the podcast for a while, is not remembering if I'd ever said something on the podcast before. The guest would say something that reminds me of something that I think is interesting, or an idea I had, and I'm like, "Oh, no, did I say this before?" I'm starting to become paranoid, like, "I don't know if I've ever said this." So now sometimes I'll be like, "I'm not sure if I mentioned in the podcast before," to just at least flag that it might be redundant.

URI: I believe in spaced repetition, so maybe repeating things is good for us as listeners. We'll get more reminders, and it'll refresh the anecdote from last time when we heard it.

SPENCER: That's a good point. It's like that conversational norm where, if you're sitting with someone and you've said something to them and then another person joins and then you have to say the same thing. It's like a social thing where you want to acknowledge that you've already said this. You're like, "Oh, I was just telling this to so and so but..." Then you repeat yourself and, by acknowledging that you're repeating it, that person knows they can zone out and they don't have to participate for the next two minutes. It's a very nice social thing. That's what I'm trying to do, "If you feel like you already know this, you can zone out or skip this part because I might have said this before."

URI: One last question on this theme, I think something striking about you is that you're really a generalist. You seem to have interesting things to say about a very wide range of topics. I don't know if that's because you know a lot of different things in general or if you just have a few tools or tool sets that you apply to lots of different topics. But I think that comes through a lot on the show. You're often talking to people about different things every week, and things that you don't necessarily have depth of expertise on, but you still manage to bring an interesting lens. Can you speak to that? Would you agree?

SPENCER: I would say I'm somewhere between a generalist and a specialist. I'm a sort of specialist, but in a handful of areas, if that makes sense. For example, I know very little about chemistry. I'm terrible at geography. I don't know much about biology. I don't know that much about literature. There's tons of topics I don't know very much about. I know people are much more generalist than me. They could have a conversation about almost any topic, and they'll know some stuff about it. I don't feel like I'm like that. On the other hand, there are a bunch of topics that I know a lot about. I know a lot about psychology, math, computer science, and programming. There are a bunch of topics I've studied in quite a lot of depth, like machine learning. With any of those topics, I feel like I can go pretty deep on them. So it's like this in-between thing. But I will add something else which is that a very useful tool I found on the podcast is, if I don't know about what the person's discussing, I just flip into curiosity mode and I just try to understand it in real time. I don't need to come in with an understanding of it because there's a good chance my listeners don't have an understanding of it necessarily. So I can just be like, "Oh, interesting. Okay, tell me about that." And I try to ask the questions till I understand it, or I understand it to the level that I can in a short conversation.

URI: Yeah, absolutely. I think in episode four with super producer Josh Castle, you talked about wanting to focus on ideas that matter. Both the ideas have and the fact that they matter, they have some implementation in people's lives. I think that is important to how the show comes out, that it's about ideas, it's about things that you can just talk about with curiosity, that you don't necessarily need pre-existing experience or a fact base in order to make sense of.

SPENCER: I think that touches on another point, which is that I think of ideas that matter as either ideas that you can use — you can apply them — or ideas that help you understand the structure of things, like how things work. So even if you don't have a direct application right now, the idea of just understanding things better, that can help you down the road in unknown ways. But when you think about the structure of things, that's something that I think one can learn to get better at without being domain specific. For example, something I find in my own life... Let's say I meet someone who's really into fashion. I'm really not into fashion. I just wear simple stuff. I don't really care. But I can have a really interesting conversation with them about the structure of fashion or the generalizations about fashion. For example, I would really enjoy discussing with a fashion expert, well, how exactly is it that a new trend develops? Is it that the magazines that are at the top of the field first identify the trend, and then they start putting stuff out and then certain people copycat them? Or is it that the cool kids start doing something random, and then the magazines notice it? I would really enjoy that conversation about the structure, how it works, but I would really not enjoy a conversation about what exactly is fashionable right at this moment. Because I'm a person who's obsessed with the structure of things, and how things work, and how things operate, that allows me to go into lots of different fields and bring a toolkit of trying to understand how things work even if I know nothing about that specific field.

URI: I think that really comes through.

[promo]

URI: Moving on, I wanted to ask you some questions about what you've learned about conversational podcasting as a whole. After 100 episodes, tell us what's top of your list? What have you learned?

SPENCER: Well, as longtime listeners may have noticed, I suck significantly less than I did when I first started. I still suck in some ways but I suck less. I'm pretty confident in that as a host. One of the most useful things I did was, right when I first created the podcast, I recorded a couple episodes and then I paid (I think it was) 40 people to listen to them and critique them. And that was super valuable. Then I think I recorded another eight episodes and then I invited a bunch more people to listen and critique those. That was super, super valuable. That helped me upgrade my hosting a lot really quickly. The next phase of my learning, I started asking my guests at the end of every episode, what could I have done better and that was useful for a while. But then that stopped being useful because after doing that a bunch, I got to the point where my guests would be like, "Oh, I don't know, I can't think of anything." So then I was in another stagnation point. I was like, "Okay, I need to push to the next level." So then we ran the listener survey where we invited all the listeners to critique and that gave me the next round of feedback to get even better. I view it as, get some data, work on that data for a while to get better, then hit a plateau. Okay, what's the next data source? And then keep going. The listener survey had so many points, I'm still working on a bunch of things from that. But I've already improved on some of them. One thing a listener said in the survey, and I wish I knew who it was because it's such a great quote, they said, "Act as if the guest is the most interesting person in the world and you have the last hour with them. Wring them of every single interesting thing before you let them go." That really resonated with me and that's one of the things I'm really trying to work on, really getting that mindset of, "This person has interesting things to say. Let me make sure that I bring those interesting things out in this very limited amount of time I have." I want that to both come through in the way that I ask questions and come through just in my vibe of enthusiasm talking to them and really getting psyched about the fact that I get to have this fun conversation, which is really a gift and a really nice thing to be able to do in my life. Anyway, that's really what I aspire to. That's one of the big elements I'm working on.

URI: Yeah, I really feel that. I really feel like this gets to one of your core traits to me, which is genuinely trying to improve at stuff. I think a lot of people feel like they want to improve and stuff, myself included, but stop at the first plateau. You seem very good at not letting the guests say, "Oh, yeah, I can't think of anything," and say, "Okay, well, I guess I'm done. I've learned enough now." You keep pushing forward into new ways to improve.

SPENCER: I tend to view self-optimization in particular as a very exciting thing. I know that some people find it demoralizing, it's like critiquing yourself. But I think of it as, "Man, I could be even better than I am now," and that's super motivating. I call it optimistic optimization. So I'm always thinking about how much better I could be and then using that as a form of excitement to drive me to the next level.

URI: Amazing. Earlier, we were talking about this and I referenced interviews and you said you weren't trying to do an interview show, you were trying to do a conversation. Can you tell us about the difference between an interview and a conversation?

SPENCER: What I would say is, in an interview, the host really recedes into the background. You even might forget the properties of the host, if that makes sense. The host is just there as a conduit for the guests. Whereas, in a conversation, you expect that there's a back and forth. I really like to think of the best conversations as one where you're building something together. You're building this thing, this conversation that is then itself, something interesting. I don't expect to speak as much as my guests, but listeners have expressed in the listener survey that they're shocked how much I talk. Apparently, I talk way more than most hosts do on podcasts. At least, that's what my listeners have told me. I believe that that's true. Basically, I think of it as, "Okay, we're building this thing together." And yes, the emphasis is going to be more on them than on me. But still, I'm part of building this structure.

URI: That is really interesting. One of my side project ideas is to build a thing that can map out podcast conversations: show how often the host speaks, how often the guest speaks, how long they speak each time, just to visually represent the flow of those conversations. It will be very interesting to see how you compare with other hosts.

SPENCER: Yeah. The metrics on that would be fun.

URI: I think this ties together a lot of different themes we talked about because the thing about conversation, the thing about building something in real time, is that you have to be vulnerable. And as a guest, you have to say some things that you haven't really thought through. You've just thought of the idea in that moment because of something the other person said. You don't know if it's gonna be good, you don't know if it's gonna be smart, but you say it anyway and put it out there. And that feels very different and more vulnerable and more dangerous than reciting your spiel that you've already practiced.

SPENCER: Something really interesting has happened with a couple of my guests who are really used to giving a spiel, which is that, I think they're used to giving a spiel, they're used to having people interview them and I challenged their spiel. I asked a follow-up question about their spiel that doesn't accept certain aspects of it that maybe they're not used to being challenged on. And I got this very surprised reaction from the guest. I felt like they floundered a little, like they weren't used to defending aspects of the spiel. I thought it was really interesting and surprising to me that sometimes people will go around saying a thing over and over and over again on endless podcasts but they haven't been pushed on like, "Well, is that premise that you're operating on true? And how do you know that?" And then they're thrown off guard. It doesn't mean that they're wrong about it but just, they're not used to getting that question and they don't have a prepared answer. Their spiel is one level deep but not two levels deep.

URI: I think of this as just one of my favorite properties of people I like in a personal sense, as well. It's the people who are most relaxed and most able to go layers deeper. And as you said, just because someone is thrown off by the question, it doesn't mean that they aren't right about the thing. But there is something really delightful when someone can spontaneously go to new levels with you and explore new facets or new angles.

SPENCER: Yeah, I do find it quite frustrating when it feels like people aren't willing to really engage in a deeper conversation about their premises. Because it feels like they just want to say their thing and just implant it in the audience's minds. They don't want to actually have a conversation about whether it's true.

URI: Yeah, I feel that. On a related note, do you have any meta sense of what makes a good conversation topic for this podcast, and which things have most often led to fruitful conversations?

SPENCER: One thing doing the podcast has really emphasized for me is that people struggle without concrete examples. So I try to remind my podcast guests that you really need to bring examples to talk about the ideas. I think a lot of people who are good storytellers already know this. But if I talk about what the sunk cost fallacy is and I define it, for most people, that's just not going to resonate. They're not going to be able to use that information. But then if I were to follow it up by giving you three compelling examples, that will be a lot stronger. People will be able to assimilate the information much better. And if I can make those examples really interesting stories, that's even better. It's even more memorable and more fun, too. So I think that's one thing that I've been really trying to encourage my guests to do. It's not always easy, because it's often actually hard to find just the right example for a concept. You can find examples but like, "That's not quite illustrating exactly what I want."

URI: Yeah, it's one of these problems of transferring stuff from one mind to many minds. In my head, there are examples but firstly, I already know the examples. And secondly, they're instantly accessible to me, the object as a whole is accessible. I don't have to say them in this linear fashion using words. Then often, there are multiple examples that collectively are the reason I believe a thing or that illustrate an idea really well for me but I can't say all the examples on a podcast so you have to pick one and it doesn't quite feel right.

SPENCER: Yeah. It's like any one example is insufficient and doesn't capture the whole concept. That being said, without the examples, people just get lost in abstraction and they can't really hang on to the idea.

URI: It's frustrating to feel like you're only communicating something 70 or 80 percent true but I think this is probably how any of us learn anything. We hear a bunch of things. We combine them in our own minds. None of them are completely right. All of them are partial. I think especially for people who care about saying true things, it can be difficult to just accept that the thing I'm about to say doesn't cover all the cases.

SPENCER: Well, doing a podcast forces you into an interesting situation where you know you're gonna inevitably say some untrue things. I'm somebody who really cares about saying true things and not spreading false information. But it's just impossible to do 100 episodes and not say a bunch of false things. Obviously, you can work hard to try to reduce the number you say but you're going to have some false things you're recorded saying, inevitably. Unless you're just so conservative but if you're so conservative, you're probably not gonna say anything very interesting. So how do you deal with that? I think one of the big reframes for me on this was thinking about nonfiction books in a new way. When I was younger, I used to think of nonfiction books as a packet of information, then you're evaluating what is the information in the book. But I think a more helpful perspective is thinking of a nonfiction book as an intervention. Like, what has changed when someone has read the book? Are they improved in some way? On average today of true beliefs, even if they picked up some false belief from the book, did they pick up more true beliefs that compensated for other falses, so overall, now they actually believe more true stuff than they did before? Rather than, "does this book contain any false information?" I think a better metric is, having read this book, does someone have truer beliefs on average?

URI: I really like that. I like that a lot.

SPENCER: That's what I'm aiming for in the podcast, to write true beliefs on average.

URI: It's a great slogan. You could add it to the list. That leads me to a question I wanted to ask you, which is, I am really interested in what different media are good for. I think books are good for learning and transferring information as well as intervening, as you say. I think Neil Postman has argued really interestingly that TV is built for entertainment, that even if you try to do other things with it, the core use of the media is to entertain. I was wondering what you think podcasts are good for? What are they about?

SPENCER: I think they serve a bunch of different roles. One is that they can give you a hook into something that you then go explore. Maybe I do a podcast on a certain topic, and you're like, "Oh, that's interesting." So now you go read a book on it, or you read a bunch of blog posts. It's just the exposure to a lot of different ideas and a lot of different ways of thinking and I think that's really useful. The second thing is, I think it can serve as a pretty good form of something that... It fills a certain space, like when you're doing the dishes, you probably can't read an article, but you can listen to a podcast. You can learn some interesting things and, even if you don't retain them all, you can still do dishes and feel like that was a learning experience. You got net positive value out of it in a way that you wouldn't if you just did the dishes, or that you wouldn't if you (let's say) just watched dumb TV. It's like you're able to get a benefit in a very passive way, if that makes sense.

URI: Well, so audiobooks also exist. Audiobooks can be listened to in any of the kind of time slots where you can listen to a podcast, I think, but do you think that's serving a different function or on the same spectrum?

SPENCER: I think it's on the same spectrum but my experience is that audiobooks require more attention. Audiobooks are in a format that's very compressed. When people write in a nonfiction book, they're thinking about every sentence, and so on. Whereas, in regular speech, we're very, very used to listening to speech. We're very, very good at processing it in audio format. We do that all the time. And there's a lot of redundancy and tone of voice things that we use to get context and interpret what's being said. I think in audiobooks, it can just feel like, "Okay, I need to really give this my full attention if I'm really going to process what's being said." This is probably not true if audiobooks are mainly stories, but I think when it's an audiobook with a lot of information, then I tend to feel that way. How do you feel about that?

URI: So you can't listen to an audiobook while you're doing the dishes? It's not as good a fit?

SPENCER: I feel like it's not as good a fit when you're doing something else. I think it's a great thing to do if you're just sitting there, audiobooks, awesome. But less good for multitasking with simple activities.

URI: Yeah, this is really interesting. I suspect that audiobooks do lose a bunch of information streams, that only broadcasting in one band, which is the literal information of text, if that makes sense. Whereas podcasts, you have the interaction between the multiple people that are usually on them, you have the different tones of voice and the different speeds and the uncertainty that is sometimes in someone's voice and sometimes isn't, so in some ways, a richer medium, I think.

SPENCER: To the question you asked of, what's the purpose of podcasts, I would also add something else which is, I think they can be really good for challenging our beliefs on a topic. Suppose that you support a certain policy. You support decriminalization of prostitution or something like that. You listen to a podcast episode which interviews sex workers talking about their views on that. Even if you don't take away really specific facts, you might have come away with, "Huh, I'm not so sure about my views anymore." I think it's actually really good for challenging the listener to just reconsider what they're thinking, even if the facts don't stick that well. I think realistically, the facts from podcasts probably don't stick that well. I think in audio format, people are not probably memorizing significant parts of the conversation.

URI: I would agree with that. I think there's a really social aspect to them that I really like, and social in an important way. They feel closest to me to chatting with friends, or I guess overhearing a conversation, or half participating in a conversation. But I feel like people on podcasts are my friends and I think that is unjokingly important because, if I listen to certain people, I want to be more like them. I want to embody the values they do, or it helps shape my identity. Somehow podcasts feel more personal in that sense. I think that can be really useful.

SPENCER: It's funny you said that. It was actually one of the inspirations for creating this podcast. I was thinking to myself, "Okay, my favorite type of conversation is when I'm talking to a really brilliant person and we're exploring ideas together and we're building something together." And I was thinking, there are probably a lot of people that have a desire for that kind of conversation but they don't actually have the people around to have it with. And maybe they'd want to eavesdrop on it and be part of it passively.

URI: Brilliant. Okay, Spencer, we're going to do a quick-fire round. I'm going to ask you lots of questions. As I mentioned before, we put out a call to listeners of the show to send in any questions they had, so some of these questions will be from our wonderful listeners. First up, who is a personal hero of yours or someone you really look up to?

SPENCER: I have a funny story about this. One day, my mom calls me and says, "Hey, Spencer, do you want to come over for dinner? I'm having people over." And I said, "Sure, Mom, who are you having over?" She said, "Daniel Kahneman." I say, "Daniel Kahneman, the Nobel Prize-winning psychologist?" And she's like, "Yeah, I met him at a party." It's just hilarious. I actually had a chance to have dinner with him and then I actually kept bumping into him. I bumped into him at an AI conference, and I bumped into him at another AI thing, and then another event. It's just funny, I kept bumping into him. And so I got to chat with him a handful of times. I have to say, I'm not the sort of person that has heroes. I never really have had people I look up to in that way. I feel like that aspect of my brand... It's not a trait that I have. That being said, if I was forced to choose someone, I think Daniel Kahneman would be quite a good choice. There are a few reasons for that. One is that I really respect the work that he has done. I think that he's just discovered a huge number of different important things about human nature. I'm not just gonna say that nobody had ever figured them out before but they had never been formalized. At least they'd never been described in such detail or studied in such depth, and I think some of them are probably just truly novel. Maybe nobody had even seen them before, and I respect that greatly. The second thing is that, just talking to him in person, I really feel like he reflects the person that I get from his papers. He's just extremely skeptical and humble in person in a way that's just shocking for someone who's so famous and so accomplished and a Nobel Prize winner. Whereas, a lot of people who you think are awesome from a distance then you get to meet them, you're like, "Ugh, God, that was awkward and weird. This person is a total narcissist" or whatever. So I think I would choose him. I was really happy to have him as a guest on my podcast. He had no reason to come on my podcast, it was really nice of him. So that was something that made me really happy.

URI: That is a wonderful episode. If anyone listening has not heard it, go back and have a listen to that. Is the reason you don't usually have heroes that so many people turn out in real life not to be like their public persona?

SPENCER: No, it's not. It's just because it's not the way my brain works. It just doesn't make sense to me to have heroes. I just think of everyone as a person and people are complicated and flawed. We're all just somewhat intelligent apes. It just never really made sense to me to have heroes.

URI: I like that. Another question. What is a math thing that's really surprising, mind-blowing, or counterintuitive that most people probably don't know?

SPENCER: This is not my field of math but it is something that I've always found super fun. Many years ago, I actually brought this topic up on a first date. Afterwards, the person said, "This is the best first date I ever had." We spent the whole time talking about this. It's the idea of computable and uncomputable numbers. Basically, very roughly speaking — again, I'm not an expert on this specific topic — but very roughly speaking, a computable number is a number where there's an algorithm that you can implement, like in computer code, that will give you as many digits as you want of that number. For example, take the number 1000. Obviously, you can write a computer program that outputs the number 1000. That's no problem. Or pi, there's lots and lots of algorithms that you can write in computer code. You can say, "Okay, I want the first N digits of pi," and it will output the first N digits of pi. Or the square root of two. For all these numbers, we know how to... These are computable numbers. Now, here's the really weird thing. Every number you've ever heard of, basically, unless you're really deep in math stuff, is computable. Just think of any number you can think of, it's computable. So what's the deal with the uncomputable numbers? Well, where this gets really weird is that all numbers are uncomputable except some tiny, tiny amount. Yet, despite the fact that essentially all numbers aren't computable, we can only think of computable ones. What I mean by this is, imagine that you randomly pick a number between zero and one. It's selected uniformly at random. There's a 100% chance it will be uncomputable. Yet, nobody knows any of these numbers, except obscure mathlons. They're really weird and hard to construct.

URI: This is great. I see why this was a great first date.

SPENCER: One of the funny things is that some people are really uncomfortable with uncomputable numbers. They've tried to remove them from mathematics and tried to rebuild math without them. Because they're like, "What the hell are these things? These are bullshit. You can't compute them. There's no algorithm that will give you the first N digits for any N of finite length and finite time." But as far as I can tell, those attempts are not very successful. Basically, they just seem so embedded in the structure of math, you just can't get away from them.

URI: That's great. Our listener, Sasha, asks, "What are your blind spots?"

SPENCER: It's hard to know your blind spots. What I'm gonna give you is a blind spot that two different people have told me I have, which I sort of don't agree with. But if it's a blind spot, I would probably not know. Two people have told me that they think I have a naive model of evil. I think what both of those people would say is that there's a lot more evil around than I acknowledge and that there's a lot more bad intentionality than I think. I don't agree with them because I think that I have a model of evil that's pretty accurate and that it's predictively accurate. I can predict a lot of people's behaviors and that it's more that I'm just not interpreting things as being evil that they might interpret as being evil, if that makes sense. There's a lot of activities. Let's say, take someone who is part of an extremist group, and they believe that the way to true salvation is to go do some horrible acts that hurts a bunch of people. I would not necessarily call that evil if the person genuinely believed, for example, that the people they were going to hurt were actually going to be benefited because of this, because of their messed up philosophy that's totally false. But their intentionality is such that they actually think they're helping people. So I wouldn't call that evil. Now, of course, it has really bad outcomes. Of course, we should try to stop that person, but I wouldn't call that evil. But anyway, this is something I've been critiqued on. So maybe it's a blind spot. I don't know.

URI: Well, as you said, there's something meta about blind spots. You wouldn't know where your blind spots are necessarily. You can only see them obliquely somehow, via other people's reactions.

SPENCER: Exactly. So maybe I have a naive view of evil. I tend to think that there aren't that many people with very bad intentions. Maybe that's partly what they're picking up on. Maybe they just think there are a lot more people with bad intentions than I think.

URI: Great. Our listener Sarah asks, "What is your theory with the least factual basis?"

SPENCER: I have a couple that come to mind immediately. One is a theory of yawning. It's been debated why humans yawn. One of the really puzzling things about yawning is that they're transmissible. If you yawn, it's much more likely that I'll then yawn. It also happens over Zoom meetings. I'm not sure if you've ever noticed that. If someone yawns in a Zoom meeting, it can cause people to yawn. It can even happen across species. People will yawn when they see a lion yawn, which is pretty funny. Here's my theory about this, which is probably wrong but it's the best theory I have. It's that the evolutionary purpose of yawning is not just a spandrel. It's not just something that just happens to happen but has no evolutionary purpose. I think that it has something to do with sleep regulation, where imagine you're living in a tribe of 50 people 50,000 years ago. Yawning might be a way to coordinate everyone to go to sleep at the same time. You see someone yawn, you now are more likely to suddenly feel tired and yawn yourself. Also now you know on a cognitive level, "Oh, this person is probably getting tired. Maybe we should all go to bed now."

URI: Great. What was the second one?

SPENCER: Here's another one thing that I think is true. I may well be wrong. I don't have a lot of evidence. But I think it's probably true, which is that, I think that consciousness has an evolutionary role. What I mean by that is, if you accept that humans evolved over millions of years of evolution, natural selection, and that humans are conscious... In other words, there's something that it's like to be a human. There's an internal experience that humans have, which I assume you believe. Most people believe that, although some don't. But you think that, for example, rocks don't have internal experiences, and probably bacteria don't have internal experiences. So if you believe that humans do, and let's say, bacteria don't or rocks don't, then something about the process of evolution created these internal experiences or qualia or consciousness. If that's the case, it is hypothetically possible that consciousness is just a weird coincidence. Evolution wasn't optimizing to create consciousness. It just happened that consciousness occurred. This seems to me like this would be extremely weird, that it just so happens that evolution coincidentally makes conscious beings. So I strongly suspect that it's not a coincidence, that in other words, natural selection of selecting for survivability actually selected for consciousness, because consciousness actually improves survivability. In other words, it has a survival advantage. But this is weird to think about because if consciousness has a survival advantage, it means it has some causal role. It's very hard to define what that causal role is and I think this gets into very weird philosophical problems that are very hard to make progress on.

URI: So since Clearer Thinking podcast is reliant on consciousness, would you say that the Clearer Thinking podcast is, in some way, evolutionarily selected for?

SPENCER: No. [laughs]

URI: [laughs] Great. Another big fan of the show is the artificial intelligence GPT3. So I fed the questions that are lovely.

SPENCER: Lies! GPT3 is not a fan of the show.

URI: Is that not true?

SPENCER: Yeah. GPT3 doesn't have consciousness.

URI: [laughs] Well, I've fed the questions from our lovely human listeners into GPT3 and it came up with a couple of questions for you as well. The first one is, when was the last time you were surprised by someone?

SPENCER: One thing that happened recently is a friend of mine did something that really disappointed me. I assumed that it'd be the end of our friendship. Then that person reached out and apologized and told me they thought that I hadn't done anything wrong. They said they wanted to see me. So that was a really positive surprise. I was especially appreciative because I've felt quite disappointed in a bunch of humans in the last couple of years. So that was really nice to get.

URI: That's really touching. Do you have thoughts about apologies more generally? Like, do you think... I was talking to someone recently about whether it's always just good to reach out no matter how long it's been, or how small the thing was, and whether it's just always ideal if you can tell people you're sorry about things. Do you think that's true?

SPENCER: I don't know about always but I think usually it is good to apologize, especially if you structure an apology in an effective way. I think an effective apology is one that acknowledges the harm that was caused and acknowledges the person's role in that harm, and then also makes it clear how one is going to reduce the harm in the future. That was one of the things that impressed me about this apology that I got. It actually had a plan of action to avoid this kind of harm in the future, which I think is a really, really good guideline when giving an apology.

URI: Nice. Another question from our friend GPT3, who I will insist is a fan regardless of fact. GPT3 asks, what was a pivotal moment in your life?

SPENCER: A pivotal moment for me was when I sought exposure therapy, which is a type of technique for treating certain types of anxiety. It's often used in cognitive behavioral therapy. Exposure therapy is a method where you expose someone in a controlled way to a thing they fear. For example, if someone is really fearful of spiders, a therapist could work with that person and maybe they first have them come to the office and look at photographs of spiders and sit with those photographs. The person might feel really, really anxious at first, but then the anxiety will fade over time. Then they'll maybe show them a plastic spider, and then they'll get really anxious and eventually that anxiety will fade. This could be done over a number of sessions. But basically, the gist is, expose them to something they fear in a controlled way until it becomes easier for them. You can imagine even having spiders crawl over your body, which sounds terrifying. If you just had to do this for ten hours, you probably at some point would be thinking about other things. It's very hard to stay focused. Not to say that exposure therapy lasts that long, but it just illustrates the idea that, if nothing bad happens over a long enough period, you just get used to it and you'll stop thinking about it, you get more bored by it. Anyway, I started exposure therapy for certain kinds of anxiety I would have in social situations. My therapist was really awesome, would give me homework to do, things like, "Okay, go to a bar. Introduce yourself to five strangers." I would be terrified of this, but I would go do it. It was just really, really impactful. It really gave me a lot more ability to navigate social situations with dramatically less fear, not to say that I don't ever get anxious in social situations, but I'm sometimes now cool as a cucumber in situations I think most people will be stressed out in. Sometimes I am still anxious, but a lot of times in social situations, I'm just like, "Yeah, whatever, I'll talk to that person. It's cool." I did this many years ago, but I'm just so glad I did.

URI: This is truly amazing to me. I don't think I realized you could do exposure therapy for social anxiety.

SPENCER: The key is you have to pick situations that you can put yourself in regularly. For example, let's say you only feel anxious at parties but you don't have access to a lot of parties, that can be tough. But if you could go to a party every single week and actually practice in a controlled way, challenging that fear... The key is you have to challenge that fear but in a way where the bad thing you fear is not going to happen. Or if it does happen, it doesn't actually have the negative consequences that your brain thinks it's going to. So when I would go up to strangers at a bar, not everyone wanted to talk to me. But I was polite. I wasn't pushy. I'd just say hello, introduce myself. If they give a negative vibe, then I'd say, "Oh, nice to meet you," and walk away. No big deal. But my brain was like, "Oh, something terrible is gonna happen if this person doesn't want to talk to you." Well, no. You just walk away. You say, "Oh, nice to meet you," walk away, not a big deal, right? But your brain needs to see that again and again to realize that nothing horrible is going to occur.

URI: This is so brilliant because I'm simultaneously thinking, "Oh, that's so silly. What's the harm of going up to a stranger at a bar?" and absolutely petrified when I actually think about myself and imagine trying to go up to strangers at a bar. I don't think I could do it. I don't know how I would pull through. For the last question, several people, including John and Sarah, asked us questions about your intellectual history. Maybe you could tell us what you felt confident about a few years ago, what you've changed your mind about, and what you believe now to be completely wrong.

SPENCER: Definitely a lot of things on the list. One that I'll point to that I think may be especially relevant to the listeners is, when I was younger and as I got better and better at analyzing situations and analyzing systems and learning about biases and building tools of rationality, I — like I think many people in that situation — started noticing more and more suboptimal things throughout the world. So many things just seem to be executed badly. So many human institutions seem to not be doing what they claim or doing a bad job. I think when I was younger, this made it seem to me like there's just low- hanging fruit everywhere to improve everything. As I've grown older, what I've realized is that identifying these kinds of problems is often ten or even 100 times easier than correcting them. So I've just really updated on, yes, you can identify all kinds of inadequacies in human civilization. But how do you actually solve them? That's really the challenge. The identification is the easy part. Not to say it's always easy, but that is the easy part. I'd like to continue trying to get better and better at that second part of don't just identify, but find an actual way to make it better.

URI: That does seem right. Spencer, it's been such a pleasure to talk.

SPENCER: Uri, thanks so much for this. You did a really great job. Maybe you'll start a podcast of your own one day.

URI: Thank you.

[outro]

JOSH: You and I have worked on analyzing a large number of popular tweets. Can you tell us some of your favorite tweets to come out of that analysis?

SPENCER: To give a little more color on that project, Josh and I have been exploring, seeking the best intellectual tweets of all time, analyzing people who are Twitter intellectuals that have reasonably sized followings, then which of their particular tweets seem to have been most popular. At least that's one metric of 'best.' Obviously, there's other metrics of 'best,' but just looking through that, I'll read a few of the tweets that came up as incredibly popular tweets from Twitter intellectuals that I think are quite good. The first one comes from Shenaniglenns, who is, I think, one of the funniest people on Twitter. A lot of the most popular intellectual tweets actually come from him. It's amazing how effective he is. So here's one. "Fool me once, shame on you. Fool me twice, shame on you. You are not very nice." Here's one from Peter Herford. "It amazes me how much of science is powered by a single Kazakhstani woman who does crimes." If you don't know, he's referring there to Sci Hub which is a very popular site for getting academic papers. The funny thing about it is it's used by a huge number of researchers who actually had access to academic journals through the university they work at. But it's so inconvenient to get them through their own libraries that they use this illegal file sharing site that this Kazakhstani woman created. Of course, it's also used by lots of people who don't have academic access so they're able to access academic papers. One thing about that is a lot of these papers that are published are actually funded by government funds, at least in part, and yet, they're locked away behind paywalls which is really sad that people can't access them. Here's another tweet that I thought was quite good. This is by William_Blake. "In 1665, Cambridge University closed because of the plague. Isaac Newton quarantined himself at his childhood home. It was the most productive time of his life. He discovered the calculus on laws of motion, stuck a bodkin in his eye to study optics. How will you spend the next year?" I love it. It's simultaneously inspiring and depressing. It's like, "Oh, God. I'm never gonna do as much as Newton." Obviously, that was written during COVID times. Here's one more that I like a lot. It's by Stephen Kaas. "You are not the king of your brand. You are the creepy guy sitting next to the king going, 'A most judicious choice, sire.'"