Robin Hanson - The Long View

Dwarkesh Patel: Okay, Robin Hanson needs an introduction, so let's begin. So you've written a few essays on the long view, how there's no significant players that are optimizing for it. But isn't the most obvious explanation for that that optimizing for the long term is ineffectual given how unpredictable the future is, or even counterproductive given that you might be ignoring present circumstances?

Robin Hanson: Well, if it's unpredictable, then optimizing should take that into account. But it does look like people are deviating from what they would do optimally trying to take the long term into account and take into account the actual amount of uncertainty. That seems so. For example, there's something called discount rates in economics and interest rates. And so if you have the long view, you would take a very low discount rate. That is, you would think the future was just as important, although with uncertainty, that might make it hard to make some other choices. But, for example, resources are just generally useful. So if you just invested in collecting resources, you don't have to know exactly what's going to happen in the future to know that resources are going to be pretty useful, whatever happens. And so if you are discounting the future and not being interested in collecting resources, that's a pretty clear indication that you're not so interested in the long-term future relative to other people who might.

Dwarkesh Patel: But if it's true that thinking about the long-term future is useful, then wouldn't you expect the most powerful institutions today to be the ones that thought about the long-term future in the past?

Robin Hanson: Well, that's part of the question I raised in my discussions of it. So you might think that in the very long run, there would be selection for thinking more about the long-term future because that would win. If we look at biology, we can think that there's been a very strong competition in biology for influencing the future. And in some sense, that is what evolution is about. Species and individuals compete to influence the future. Having more descendants is the main way that biology influences the future. Now, animals sometimes make plans and you as a human might make a plan to save up resources and pass them on to your kids. And we know that that sort of planning has a discount rate that goes with it because your kids only share half your genes. So your individual genes, they can't save resources for themselves. They can only save resources for this whole creature that you are, which is a bundle of genes. And this whole creature that you are, when it passes on and has descendants, those descendants only have half as many genes. And so that means that your genes have a choice between doing things that promote the genes right now versus doing things that collect resources and have a 50-50 chance of promoting that gene later on because its descendants may only have a 50% chance of sharing that gene. So that produces a discount factor of once per generation a factor of two, and that's in the ballpark of what actual human discount rates are. That is a plausible explanation for human discount rates: we evolved in this context where when we had a choice to save resources in order to pass it on to the next generation, the next generation only shares half our genes.

So the challenge is to come up with a unit of selection that can pass on resources to itself over the long run. And what that might be, perhaps, is a nation or a firm or a religion or some organizational unit. I think we should reasonably expect that in the long run, organizations will find a way to collect resources and then promote themselves in the long run. But they haven't done that so far because most of the organizations don't really have so much of a separate existence in our world. They are a way that our genes or our individuals use—genes use individuals and individuals use organizations to further their ends. But if the organization wants to just take over and says, "I don't care about you genes or individuals, I just want to promote myself," then the individuals and the genes will rebel and bring the organization back in line. That's kind of what's happened so far; organizations haven't really been able to promote themselves as much as they might like.

The closest you might get is religions encouraging proselytizing, where they're promoting the religion, but they found a way of doing that via the individuals proselytizing promoting themselves. They found a win-win deal where the organization can promote itself and the individuals can promote themselves. But that's hard to find, and so organizations haven't been able to just collect resources because what happens is if a church collects resources or a firm or a club, later on some humans are in charge of that organization, then they just take the resources. They just grab them. They say, "The heck with this organization, those are tasty resources. I can promote myself and my genes by just grabbing the resources and taking them away from the organization." So for the long run, the question is how can organizations find a way to collect resources and keep it directed for the purpose of the organization, as opposed to allowing the people who run the organization to grab those resources for themselves.

Dwarkesh Patel: Have you considered doing this yourself, raising some money and then building an institution to shepherd and distribute it in the future?

Robin Hanson: Well, I mean, I've thought it through just in terms of understanding what's going on. I'm not rich. I don't have that many resources. But one of the big things that's happened is this might happen if we used contracts to make it happen. So, for example, I might make an organization, and then I make the constitution of the organization a certain way, and then I say, everybody who wants to join this organization, you have to agree to this constitution, and then we're going to use law to insist that the constitution be obeyed. And if we were allowed to write a constitution that specified saving resources and that constitution would be enforced, and then the organization was just not going to be taxed or destroyed, then this would work. And what's gone wrong in history is that when organizations tried this, people associated went to the courts and said, "No, no, no, we want those resources," and they push the courts to interpret law so that these contracts are not enforced. We do not allow arbitrary organization contracts that allow the organization to save resources into the future. That's specifically something that law has prevented.

Dwarkesh Patel: What you just described sounds like a nation state. So is the nation state the unit that's figured out how to act as an evolutionary agent and grow more powerful?

Robin Hanson: No, because nation states can't just collect resources and save them. The people who run the nation state, they get to spend them. In a democratic nation state, for example, if the country collects a bunch of resources, then the voters go "tasty" and they can vote in a politician who'll choose to spend those resources on the voters and not save it for the future. Similarly, even when there's a dictator, the dictator doesn't set it up forever for the organization. They're setting it up for themselves or perhaps their children. And again, they don't set it up to just keep collecting resources.

But let's walk it through. Today and for a very long time, almost as long as we know, interest rates have been higher than growth rates. That is, rates of return on assets have been higher than the economy grows. So if the economy doubles in 20 years and assets double in 10 years, then any organization that chooses to grow assets will grow as a fraction of the economy until it comes to dominate the whole economy. And then at that point, interest rates would fall until this organization really couldn't get a rate of return above the growth rate of the economy. But that's an opportunity that's been sitting there for a long time and not really successfully taken advantage of. Nobody's really managed to create organizations that just keep lasting and collecting resources, growing faster than the economy, and therefore growing as a fraction. So what happens? Either the people running the organization decide to take it, or outsiders decide to take it. The local city or the local government says "tasty" and they decide to tax it or just grab it, or in war, invaders come and take it. Consistently over the long run, we just haven't been able to do that.

But it's a thing that remains possible, and I predict that eventually there will be some sort of unit of selection that will take the long view, that will be able to collect resources and make choices and promote itself with a long view in mind. And when enough of those units exist, then they will drive interest rates down to not exceed growth rates. And then at that point, the world will be full of creatures who take the long view, and then the long term will not be neglected. So today, you may well fear and worry that there are things that will happen in the long run that could kill us all, that will destroy us, but nobody really takes the long view, so nobody's very interested now in working to prevent it. And that's because we just don't have any units who take the long view. Individual humans don't because we have these discount rates. Organizations don't because they are run by the humans. And other biological species, they really can't even plan and think about the long run. So that's it.

Dwarkesh Patel: But as you described, religion implicitly at least acts as if it's planning for the long term.

Robin Hanson: In some ways, but not others. Organizations are not like agents who can plan everything with autonomy. They're just a collection of strategies that humans use. And in some ways, they can manage to promote themselves and other ways not. But they aren't these conscious agents, planners who have a utility function and can make complicated plans to achieve the organization's ends.

Dwarkesh Patel: But isn't that exactly what you'd expect in an evolutionary landscape? Just like this sort of—

Robin Hanson: Eventually, right. So that's the whole point. Eventually. But organizations are new. Humans have been around for a long time, humans with organizations. Initially for a million years, human organizations were like 50 people, and that was quite a distinct thing from other primates, which took a lot of work to be able to make human organizations with 50 people and have them function. Those were human bands. Then with farming 10,000 years ago, we managed to make organizations on the scale of a thousand people for a village, and then maybe empires. But empires only control a very limited range of things that people do in villages. For the most part, empires were just things that handled the military. And if an empire came by and they wanted your sons for soldiers and they demanded taxes, you gave it to them, but otherwise they didn't really run things.

Dwarkesh Patel: Sorry. I meant, isn't that exactly what you'd expect in the long term? As in, you'd expect these unconscious, basically memetic units to succeed in the long term. Meaning even in the future, we might just have these cultural units succeeding and not an organization. What state of organization will be more successful?

Robin Hanson: You need a unit that can plan in order to predict that a unit will successfully plan to promote itself. If you just talk about a meme, a meme is just one tiny piece and it really can't promote itself except in conjunction with a bunch of others that form an organization. Just like genes. Genes really don't promote themselves by themselves. Genes promote themselves via organisms, and organisms are big complicated collections of genes because individual genes can't plan, they can't store resources, they can't make strategies. They're very small pieces. It's the whole collection of genes that is capable of doing those things. And similarly for organizations, they have to have a lot of parts that all work together to make an organization work. The question is, can you do that?

Dwarkesh Patel: So Tyler Cowen writes in *Stubborn Attachments* that if you really care about the long term, the best you can do is just promote economic growth. Do you agree with that?

Robin Hanson: Well, it depends on what you care about in the long run. If you only care that everybody's happy in the long run, regardless of whether you're allied with them or whether you like them, then you just want the future to be big. And economic growth is the future being big. That's almost by definition. If you say, "I don't really care who's in the future or what they're doing or what's happening. I just care that there be a big future and there'll be a lot of things happening," then you want growth because growth literally is the measure of how much stuff is happening, how big it is. So the question is, what do you care about?

Now, when I predict there will be organizations with a long view, I'm not predicting that kind of preference because that's not very stable to competition. When we think of biological organisms having a long view, they have a long view to promote themselves. They have a long view to create descendants. That's not the same as wanting the entire Earth to have more biomass. That's them wanting to continue. So to me, the easiest thing to predict is that eventually there'll be units of selection—organizations—that take a long view with respect to themselves.

Dwarkesh Patel: But the rivalries that exist today, you wouldn't expect them to exist in the far future. So why would you even consider them when you're planning for the future? Why wouldn't you just try to lift all boats?

Robin Hanson: Well, that isn't selected for. You can if you want. You don't have to choose to have yourself represented in the future. You can just choose to promote the future if you can and will. You can just do that. But what we predict is that the future will end up being dominated by creatures who do promote themselves and who take a long view about how they promote themselves. That's the thing we should be able to predict about the future.

So you have a choice about whether you want yourself reflected in the future, whether you want to be part of a unit that's competing like that and taking the long view. But if you take a long view but you don't promote yourself, whatever it is in yourself that makes you take the long view, that's dying out. You aren't taking a long view on the long view. In that case, you're having a current long view that you're acting on, but you're not acting to preserve the long view that you are and have it last. If you want a long view that lasts, then you need to focus on preserving a long view, on promoting the thing that embodies the long view and having it last.

Dwarkesh Patel: Yeah, that makes sense. Okay, let's talk about *The Elephant in the Brain*. You explained there that we have hidden motives where we actually want one thing, but we pretend we want something else. Our subconscious and conscious intelligence are different—our capacity to actualize our implicit goals and then our capacity to justify them explicitly. So let's back up so we make sure we don't confuse people.

Robin Hanson: We are complicated creatures and there are many levels at which we make choices. We can talk about causes of our choices that are proximate, that are close to the choices, and distal, far away from our choices. And we can have different ways we think about the causes, distal and proximate. That's the key thing here. When we are talking in our book *The Elephant in the Brain* about motives, we're focused more on distal motives. We're focused more on, from a distance, what are the sort of fundamental forces that are pushing choices one way or the other. Those forces may or may not produce different conscious thoughts just before the choice is made.

If we think about proximate choices, you might ask, just before I made the choice, what was going on in my head? What ideas did I have? What was I talking about? What would I have said if you had asked me about it? Those would be proximate causes. Those would be the things closest to the choice. But those are complicated and context dependent. In many ways, I think it's more robust and even easier to focus on distal causes—what the fundamental forces are that are driving choices one way or the other. Those forces can reflect themselves differently in different final choices or contexts. For example, in different times and places in history, people have thought differently about different things and they have different words about it and different thoughts come to mind, but often it's the same fundamental forces behind all of that that push people to do things.

You could take romance, for example. You could say the idea of romance and what appropriate things are for relationships and which kind of relationships we approve of. When you're thinking of entering relationships, what's in your head and what are you saying about it? Those things have changed a lot through history and places. People have thought very differently and they're changing them now. But if we stand back and say, why do relationships exist? Well, it's a simpler theory, a simpler story. We get why relationships exist because that's how biology reproduces. Without relationships, there would be no descendants. And so from that level, it's relatively easy to find this robust explanation for some of the major elements of relationships that are independent of time and space and are relatively simple.

Similarly, in our book *The Elephant in the Brain*, we're talking about the more basic fundamental motives behind many things we do—the motive being just a force, a thing that produces it. We're less interested in what's in your conscious mind at the time and whether you're aware of it and what you would say about it and maybe even how much that varies across nations, across time, across social classes. Those are all interesting, but I think the first priority would be to just get the common elements. If there's a common element behind schools or medicine across time and space and social class, we should probably know that first before we try to figure out how rich people do it differently from poor people or Europeans do it different from Asians, or it was done differently now versus a century ago. Those are all interesting to some extent, but the first cut should just be what's the common element.

Dwarkesh Patel: But if we go along that demarcation, the question then is, is our capacity to realize our basic motives correlated with our ability to explain them later on with our conscious mind?

Robin Hanson: The key idea in our book is to say your conscious mind has a purpose. It's built for a reason, but it's not the being the king or president that you think of or perhaps have been told. That's not what your conscious mind is for. Your conscious mind is more a press secretary. Its job is to watch what's going on and come up with good explanations that make you look good, and in particular, that protect you against accusations of norm violations. Your job isn't to know why you're actually doing things. Your job is to explain why you're doing things.

Dwarkesh Patel: So my question is, is there some sort of correlation between how good the president is and how good the press secretary is? Because people have differing abilities to justify their actions, and people also have differing abilities to actualize their base motives. Is there some sort of correlation there?

Robin Hanson: There probably is, just because there seems to be this overall correlation in abilities of all sorts. IQ and wealth and just good things tend to be correlated. So I would expect they're correlated, roughly, not finely, but roughly, just because people who are more capable of A tend to be more capable of B for almost all A and B. That's just a general thing for people.

Dwarkesh Patel: Okay. Would you expect meditators to signal less, to just be more upfront about what their motives are?

Robin Hanson: No.

Dwarkesh Patel: Could you tell me why you would expect that? They're more self-conscious, so they can't delude. Maybe they can't fool themselves as easily, and so they're less likely to fool others.

Robin Hanson: Well, first of all, I'll ask, how do I know they're more self-conscious or self-aware?

Dwarkesh Patel: I think he means self-aware. Yeah, sorry. They're more aware of what they're doing and why.

Robin Hanson: I don't know that they are. So let's talk about the truth-oriented community. What does that mean? A lot of organizations and groups and people in the world tell you that they are part of the truth-oriented part of the world. And that's part of their identity. They tell you, "Those other people there are deluded, but we are into truth." It's one of the favorite words among Christians use. Christians are telling you, "We care about the truth. We have a strange religion, but it's true, and we are into truth, and those other people are not into truth." So everybody likes to think of themselves as the people who are into truth and who are less deluded than the other people. I just can't take people at their word for saying that they are less deluded. They might be, but merely the fact that they like to say that about themselves isn't enough for me. I want to see some concrete evidence that they are actually more truth-oriented.

Dwarkesh Patel: But you have no reason to suspect that a practice like meditation would reduce the distance between who you present yourself to be and who you actually are.

Robin Hanson: Well, then I'd have to dig into what I know about meditation. Now, you may well know a lot more than me. I could defer, but the last time I did a bunch of reading about meditation was just before my interview with Sam Harris, because he was into meditation. We never ended up talking about it, but reading about it, I saw this focus on a certain sort of meditative state of mind. They seem obsessed with a certain sort of state of mind. They are really trying to achieve that and they see that that's good for themselves and they somehow think it's good for the world. And right there, that looks a little deluded, I gotta say. Whatever the state of mind they seem to get into, I'm not seeing much evidence that it's this thing that helps the world a lot. I'm not even seeing that much evidence that it's the thing that helps them a lot. What I see is that it's very prestigious in their world. That is their ranking pecking order. The pecking order is who has achieved this higher level of consciousness in their meditation. Every world has some sort of pecking order. Why not that one? But I don't see much of a practical use of it compared to all the resources they're investing in it. They put a lot of time and trouble into this thing, which has a lot of opportunity cost. People take a decade off to go meditate. Just so later on, they're going to be better at something. There's almost nothing you can do for a decade which will somehow change you in a way to make you more productive later on.

Dwarkesh Patel: I think Sam Harris did do that.

Robin Hanson: A decade is huge. The opportunity cost is enormous. So right there, some of the specific things I know about meditation don't give me much confidence that these people are less deluded on average than other people. They're into a relatively religious kind of frame of mind where they've got a particular pecking order. They tell themselves, "No, no, I'm not doing this for me. I'm doing this for the world and consciousness and whatever." I've met people who get together and think the way we're going to create world peace and the way we're going to make innovation, the way we're going to make the world better, is you're going to go meditate for a long time. You go, "What? Feed the poor or stop wars?" No, you're going to go meditate. So they might even be right. I could be wrong about this, but you can see where I'm coming from in terms of on the face of it, there's a reason for skepticism.

Dwarkesh Patel: Yeah, no, I don't meditate. But let me try to steel man their position. The claim is you're more aware of the next thought that enters your mind. And because of that awareness, you can choose, you can realize that your own motivations are false to some extent. And so you're less likely to project the false motivation to others. Do you buy that chain?

Robin Hanson: We wrote this book, *The Elephant in the Brain*, that takes 10 areas of life and talks about how the motives are different. If you showed me a book on meditation that went area by area of life and showed me specifically how they had figured out that the motives were different, I would be impressed. If they just say, "I can see that my motives are lying," I say, "Well, what exactly are the lies?" Because a lie is where you say A and B is really true. If you could make a long list of the specifics A's and B's, I think, "I believed A, but B was really true," and that applied to a lot of people, that would be impressive. You show me the meditation handbook that gives me those long lists of A's and B's. And I would then say, "Yep, those people seem to get self-deception and hidden motives." Without that list of A's and B's, lots of people in the world like to say, "I seek truth." And they like to say, "Once upon a time I was deluded and then I saw the truth." And they don't like to get into specifics of what exactly they are deluded about other than the fact that they had yet to join Christianity or yet to join effective altruism or whatever it was. The usual thing they're happy to tell you about being deluded about is not having joined the new group they're part of now. If I want a longer list of other things they were deluded by, that's sort of missing.

Dwarkesh Patel: So Robin Hanson won't be meditating anytime soon.

Robin Hanson: I might, but I'm not going to put this huge glory priority on it. I scratch my back sometimes. It's nice. Meditation can be nice too, but it's like, I don't say scratching my back I'm doing for the world or it's going to create world peace or we're all going to have a unified world together. No, scratching your back is just nice.

Dwarkesh Patel: That sounds like a straw man, but Robin Wright actually wrote a book where he claimed exactly this. Scratching your back makes world peace.

Robin Hanson: Okay.

Dwarkesh Patel: So along with the faculty to fabricate pro-social reasons for what we're doing, why don't we have the capacity to consciously detect other people's false explanations of what they're doing? That seems kind of useful.

Robin Hanson: Well, we do. When we have rivals, we are unusually able to sense when they might be hypocritical and when they might be lying about their motives. Of course, we go even farther than that and attribute falsities and delusions to them that they don't have because we're really eager to trash our rivals. What we have a harder time doing is noticing our own things and things in common. In our book, we talk about many kinds of hidden motives, and some of the hardest ones for us to see are the ones that we all share. For example, you might say that your rival doesn't care about their grandma because they won't take their grandma to the hospital. And the rival doesn't care about their country because they won't pay for more national health care for the rest of their country members. But if all of us are deluded about medicine actually being helpful, that's not going to show up in those sorts of accusations about rivals, because those are appealing to shared senses of what's important, and then saying this particular person doesn't share what the rest of us share about what's important. The more that motives vary across people, the more you're going to be able to notice that a rival might be having a different motive than they say. But when motives are more common, it's harder to see.

Dwarkesh Patel: Why have we even retained social norms? Why has society even evolved to have the norms it has?

Dwarkesh Patel: If it's so easy to skirt them by just pretending to uphold them. It's not that easy. So just to be clear for our listeners: humans are unique compared to other animals in having norms—rules of what people should or shouldn't do that we enforce via collective groups. If I see you breaking a rule, I'm not just supposed to do something about it; I'm supposed to appeal to other people around me, tell them that you violated the norm, and discuss with them what we should do about it so that we collectively punish. And if I see you violate the norm and I fail to do that, and somebody else notices, that's a violation on my part. That's the key idea of norms. Norms are what we have before law to make people behave, and they are still very important. In some sense, norms are more flexible and fluid. We use norms when they seem sufficient, and we resort to law for things that norms don't seem up to the task of in our modern world. Up until, say, 10,000 years ago, there was no law. Norms were everything. That's how we dealt with all problems between people. So norms are still this superpower humans have, and we continue to use them. We've augmented them with law. The context in which we enforce norms has changed in the sense that we have much larger communities of talking. In the old world of a small band, I saw you break a norm. I could tell my friends, they could tell their friends, and pretty soon, all 30 of us have heard. We've all gossiped about it. And if you have a counter-argument, you gossip your counter-argument, and maybe we all get around the campfire tonight and talk it out and make a decision. In our larger world, we have troubles with that because we're in larger communities and we just can't spread the word about everything to everybody. We have to decide who to tell what, and people have to decide what to listen to and what to believe in contexts where they don't know people as well and don't have as much context about the accusation. That makes it harder to handle norms in our world, but they're still pretty important.

Dwarkesh Patel: Shouldn't we be signaling less over time? It seems that social norms are getting less rigid over time. So shouldn't the amount of signaling go down and the distance between our fabricated and real selves go down over time?

Robin Hanson: I'd say the opposite in the early transition. If you think about foragers living in a group of 30 people, these 30 people have been with each other their entire lives. They know each other pretty well. So whenever they hear about something somebody's done or an accusation of what they've done, they can put that into a lot of context. Those 30 people have been spending their entire lives trying to impress each other, but in any one moment, they don't have to make a crude signal that's hard to interpret. You know this person pretty well, and now you are slightly adjusting what you know about them in the context of this thing they did yesterday. That's much more about integrating the information into a solid impression. People don't change that much over a long time, so mostly you just keep reconfirming what you already knew about their abilities and inclinations. Once you moved to a village-sized world in the farming world of a thousand people, now you didn't know everybody that well. Any one thing somebody did is much more of a potent signal about them. I think people paid a lot more attention to crude and potentially misleading signals once they were a lot more ignorant about the person the signal is about. In a group of a thousand people, somebody comes to you and wants to marry your daughter. You haven't been around that person your entire life. You can use your connections to talk to other people who maybe have and try to get information that way. Or if there's an accusation that somebody's done something bad, you have to decide whether to support that, but you don't know that much about this person. You hear what they said and who's accusing them. In some sense, signaling matters more because you know less, but you still have to make key important decisions. In our modern world, you could see that continuing on. There are many people we suddenly read about in the paper we've never heard about before in our life, and we're making judgments about them. We have to make these crude judgments based on where they live, what degree they have, their gender or age. In some sense, signaling becomes very potent in that context. All of a sudden, the world is accusing somebody of something, and you have to decide whether you're against them or for them, and all you have is a small number of clues. They've known their whole life that they need to try to make those clues look good so that at this moment, if it should ever come, you will be favorably inclined toward them.

Dwarkesh Patel: But isn't there another sense in which each person matters less when there are more people? If there's only 30 people in a tribe, some young boy is very likely to marry your daughter. But if there's thousands of people, the next guy you meet is very unlikely to marry your daughter. So you're less likely to be observing the signals or displaying signals to the random person because they're less likely to be important to your life.

Robin Hanson: Right. But a lot of the signals you send don't cost any more to send to any one extra person, so you've sent them everywhere. You build up your muscles, and they show up, and anybody who happens to see can see your muscles. You have a beautiful singing voice, and anybody who hears your voice can hear it. You have wealth, and you've got gold jingling on your wrist. Anybody who sees you can see that. In a lot of ways, your signaling isn't targeted at particular audiences. You are just collecting signals that are seen by a lot of people.

Dwarkesh Patel: But what about signals that are powerful only because they're targeted, like narrowcasting?

Robin Hanson: That's a very different sort of thing. Loyalty signals, I would think, are much more of that. When people think of signaling, they mainly think about signaling ability or capacity. But in fact, loyalty signaling matters about as much as ability signaling. Loyalty signaling is about who you're loyal to in what ways. That's a signal about particular other people. For example, I say that medicine is largely a loyalty signal—you spend money to help somebody else get medical care, and that's a signal to them that you care about them. You'll have to target that at them. The question is how much more do they know about you in order to interpret the signal. The better they know you, the less they're going to misjudge you on the basis of this one signal. But even when people know a lot about people, they still put a lot of weight on things like this. If you are suddenly injured and you are feeling really scared about your life and whether you're going to be taken care of, having somebody you've known all your life quit work for a bit, come in to take care of you, be around you all the time, say nice things—that means a lot to you. Even if you've known this person all your life, you aren't making further inferences about them in that context. If you've never felt at risk of your life, and this is the one moment when that happens, you've never had that signal from them before. This is your chance to get it.

Dwarkesh Patel: Are we signaling more now than ever because of social media?

Robin Hanson: That's not obvious. We're signaling to different people in different ways. Honestly, I would say 80% of what we do is signaling in some sense. That's been true for a long time. Social media just moves it around.

Dwarkesh Patel: But if we're consciously putting up an image on Instagram or Snapchat, doesn't that increase the effort we're putting into signaling?

Robin Hanson: Your ancestors were doing a lot of things consciously too. They bought a house, wore certain clothes, went to certain meetings, made sure the fence was painted. They were polite when they talked to people. Almost all people have always been doing most of the things they do with an eye toward how it will look to other people.

Dwarkesh Patel: Interesting. I wanted to ask you about your chapter on conversation. You explained that conversation is actually... I'll let you explain first what conversation is for.

Robin Hanson: The setup of the book is that the first third makes it plausible that people might not be aware of their motives. The last two-thirds goes over 10 different areas of life. For each one, it says: in this area of life, if you ask people what their motive is, this is what they will say. Here are a bunch of puzzles that don't fit that very well. And here's our better explanation of the motive that's really going on—one that people don't usually admit to. Conversation is one of our chapters. In that structure, the first thing we have to ask is, why are people talking? We want to set aside very practical talk—ordering a pizza or testifying in court, where the purpose is pretty straightforward and mechanical. We're asking about conversation. We're often just talking without much of a purpose in mind. What are you doing when you're shooting the breeze, chatting, talking about a wide range of topics that aren't very directly related to anything you're doing? The most common explanation is that we're sharing information. I know a lot of things you don't. You know a lot of things I don't. We share back and forth, and then both of us will know more. That's not crazy. All of the things we say as motives aren't crazy. They do apply some of the time. That's why they work as an excuse. When we say they aren't our real motives, we mean that's not our main, most common motive. It applies less often than we admit. Yes, we do exchange information, and that is useful. But it's not the main reason we're talking. How can I say that? There are puzzles that don't fit so well. If we were exchanging information, we'd be keeping track of debts. I'd say I've told you three useful things, it's your turn, tell me something useful. We'd be less eager to talk and more eager to listen. We'd be talking about important things. What's the point of talking about random trivial things when you're trying to exchange information? But in fact, we seem to follow this rule where the conversation is supposed to bounce around at random, and wherever it goes, you're supposed to have something to say. You're not supposed to try to make the conversation go where you want it, especially not to important things. We do try to talk more than we listen, and we don't keep track of debts. So what's going on? Our story is: you can think of yourself as having this mental backpack of tools and resources. Every time a new topic comes up, your job is to find something in that backpack that's relevant and interesting and even impressive. The more consistently and often you can do that—play this game, whatever subject comes up, you've got something relevant and interesting—the more impressive that backpack is. That means if I'm your friend and associate, and some problem happens to me, and I tell you about it, you're going to use that magic backpack to bring out something useful, and that'll be helping me. Ordinary conversation is the game where we test each other's backpacks. We check to see how useful it is, how full it is, what you've got in there, so that we can get access to it later.

Dwarkesh Patel: But the fact that we want to talk more than listen—isn't that a problem for this theory that we're just looking for potential mates and allies? Because we'd also want to listen to the signal our potential mates and allies are sending us. So shouldn't we want them to talk as much as we do?

Robin Hanson: We both always want to listen to other people's signals and send our own. But we are usually more interested in ourselves than other people. We think we've roughly got a pretty good gauge of them, and slight variations on them aren't so important to us, but slight variations on us are much more important. We want to edge ourselves up just a little bit more so that we can seem to be better. If you thought of somebody who is a really good judge of other people but didn't look very impressive themselves, versus somebody who looks really impressive themselves but doesn't seem that good a judge of other people—who are people most impressed by?

Dwarkesh Patel: That makes sense. But just as a tool in judging people, in terms of how useful it is, shouldn't we expect people to also care as much about what other people are like? If you're talking one-on-one, it matters just as much.

Robin Hanson: The question is just on the margin, how much extra effort are you going to put into it? When people are talking, they can either listen to what the other person says or plan the next thing they say. It's a matter of relative priorities. If you think you've got a pretty good judge of them, you may just want to focus on moving yourself up another way. Another way to think about it: for each person, there's a threshold of whether they're good enough for you. What's mainly important is whether they're above or below that threshold. It might be relatively rare that they're right near the border of the threshold, where you're really trying to make a judgment about whether they're good enough—on a first date or an interview for hiring. For yourself, though, there's no threshold effect. You just always want to be better. You might be good enough for one person, but they might talk about you to other people, and you just want to go up.

Dwarkesh Patel: Okay, I see. That makes sense. So you wrote in that chapter that conversation is mostly a way to judge people's capabilities. You wrote: "We do not routinely expect mumbling, stuttering, scatterbrained politicians." So you're going to have to explain the nominees of the 2020 election to me.

Robin Hanson: There's no doubt that relative to a median citizen, the two presidential candidates are impressive. You might think there are even better options out there, but you have to look at all the other constraints. Leaders have to meet a lot of different constraints. Politicians—democratic leaders—have even more constraints, and there's a lot of randomness going on. Our book and our analysis is focused much more on the average case than on individual deviations. Both presidential candidates, I understand at a rough level how they got there and why they're there. Often it's a matter of compromise between different groups. If you can separate groups into distinct groups, each group has their favorite person, and that's not necessarily who goes up. It's often a compromise between different groups, and that compromise isn't necessarily the best from any one group's point of view. If you just want to be an impressive speaker, having a simple, absolute disposition that you take to the extremes is much simpler and cleaner to explain. But that's not where the median voter will go. Political processes choose compromises, and compromise positions aren't as easy to explain or rationalize or inspire people with. There's a trade-off between getting a person who sits in the middle of a large group who everybody can at least not hate too much, and having a person with a simple, clear message that resonates and inspires people.

Dwarkesh Patel: Are you saying that being articulate could in some ways harm you as a politician?

Robin Hanson: Oh yes. There's a standard observation about politicians: they are not very specific with their positions. They are often purposely ambiguous, and they often seem to speak different things to different audiences. The ambiguity is part of a way to cover for that—to say, "No, I didn't really say different things because these words cover different purposes." If politicians were very clear about their positions, that would be a big turnoff to a lot of people for whom the clarity is not the position they want. If they can be vague, the vague words can encompass a lot of possibilities. That's only articulate in a certain politician-y sort of way. If you look at politician speeches, there's a certain style in which they're articulate, but it's a very ambiguous style, not clear and precise. They're supposed to throw out words that are emotionally loaded and evoke images in your mind without committing to much specifics. Bill Clinton was somebody who was articulate, but many people thought that he represented them.

Dwarkesh Patel: With the current nominees, it seems that they don't have the same faculty of clarity and conversation.

Robin Hanson: No doubt it varies. The word "articulate" is often a class-loaded word. Trump went out of his way to not project high-class signals. His winning in 2016 was basically saying there was a group of people out there who felt they were being ignored, and he credibly signaled, "I'm one of you, and I'm not with them, so I will represent you." To do that, he had to show an affiliation with them different than the others, so he had to be a contrarian signal in some ways. If he just gave the usual sort of political prestige signals, that would not convince them that he was one of them. They felt they were on the out, being ignored by cultural elites. Trump needed to not be a cultural elite in order to say, "I'm with you, I'm not with them." They were also more rural and working class. Trump explicitly, purposely, successfully said, "I'm one of you" by choosing cultural styles that were like them and specifically different from the others, in ways that made the other people yell at them. What better way to convince you that I'm not one of them than to make them yell at me? Then I'm not with them. I'm with you.

Dwarkesh Patel: This all makes sense, but it seems to contradict your explanation in the chapter on conversation. If conversation is a way to judge whether we want somebody in leadership, is it either just not important at all or...?

Robin Hanson: Everything we do has many motives. There's the one we'll most often mention, and then there's the one that most often matters, but there's always a lot more. In many kinds of conversations, loyalty signaling is just as important as ability signaling. That's not so much in one-on-one conversations, although it's important there. It's important, for example, when your friend complains about something that you show support. That's a loyalty signal. You're not being very impressive with your support; you're just being loyal. You say, "Yes, you know, when you did the interaction with them, they were wrong and you were right." We do those sorts of loyalty signals all the time when people around us complain or indicate some sort of rivalry. In politics, loyalty signals matter all the more. Honestly, ability takes a second place to loyalty in politics and in many other areas of leadership. Even for a manager or a CEO, when they are looking for subordinates to promote, loyalty matters more to them than ability. That's a common truism about organizational promotions and coalitions. You often have to show loyalty before you show ability.

Dwarkesh Patel: That makes sense. Peter Thiel often says that it's interesting how the people who are the most successful startup founders are somewhere on the Asperger's spectrum. It suggests that they're oblivious to the hidden motives of others or don't feel the need to put a pretense on top of their hidden motives. Does that make sense to you? And you've said yourself that you're a nerd. Does that kind of explain your success in coming up with social science ideas?

Robin Hanson: There are several different things going on here. If we're just talking startups, the most fundamental thing about a startup person is they're on the tail of the distribution of optimism. The vast majority of startups fail, and if you're going to start a startup, you have to be weird—not only be weird but have a weird expectation that you can succeed where everybody else is failing. You're being selected from a tail of distributions. I just expect them to be weird in lots of different ways, not just Asperger-y, but lots and lots of ways. That's just because it's being weird. In addition to being weird, it's going to be contrary. A startup is not just another restaurant in an area where there's lots of similar restaurants. A startup is a kind of business where you are being contrarian with the business. You're choosing a different business concept, a different product. It's got to select for contrarians. It's not going to make sense to be part of a startup unless you have a contrarian take on the business. You have to think that you have a different idea for how the business can go. You have to be pretty confident in it and weird enough to actually go for it. The things you're selecting for in a startup firm is somebody who's not comfortably going up an organizational ladder somewhere else. They have decided to leave that and take this big chance on a big contrarian idea. They also need to be more of a generalist. A startup requires you to do everything; you can't just be a specialist. We can make this long list of things startups are selecting for. Autism could correlate with a number of those, but not necessarily the strongest thing. Another thing is that a startup founder has to have analyzed business. You have to have this theory of this industry, even if it doesn't exist, and what the customers want, what the costs are, what the business competition strategy will be. That's an abstract thing. You have to have some system theory of this industry that explains to you and your investors why this isn't crazy. Asperger's people are systems people; they think in terms of systems. That also explains it. It seems pretty overdetermined. What the fundamental strongest cause is, I don't know. But I'm not sure why I should care.

Dwarkesh Patel: Do you think you being a nerd helps you come up with all these astonishing social science ideas?

Robin Hanson: The story I tell about that is simpler. Most people who are socially skilled function well in their social worlds. If they talk social science, they talk about social science theories, and they won't necessarily notice if those two are in conflict. They will talk about how we're all egalitarian here at this company, and then they will be dominant in their activities and make sure they win out over their rivals. They won't notice the contradiction. It won't occur to them to notice the contradiction because they are so smooth and intuitive at doing the right thing and saying something else. Almost all our social activity is intuitive; it just comes subconsciously. We just do it, and we don't know why. We don't even notice it unless it doesn't work. Nerds, in contrast, don't have this social capacity to smoothly glide through the social world, always doing the right thing. We're doing the wrong thing all the time and wondering why it's wrong. We have much more to gain from explicit analysis and thought about our social world. It still doesn't make us as socially skilled as someone who's intuitive with it, but it makes us a little better because we think about it. If you're a social scientist or a social analyst, in addition to just being a nerd, you have more of a chance of noticing the contradiction. You say this one theory, and then you see yourself doing something else, and you notice that these things don't fit because you see yourself doing things more. You're being more conscious about what you're doing, thinking and observing, and therefore having a better chance to notice whether your theory applies. Nerds are better able to notice when their own behavior or the behavior around them doesn't fit the theories they say, because they will have more explicit theories about the world around them and what they're doing, and more of a chance for those explicit theories to come into conflict or be tested against each other.

Dwarkesh Patel: Even still, it's incredible how many original ideas you've had. Is that a sort of indictment of the academia around social science ideas? Because if it was efficient with regards to social science ideas, you wouldn't expect all this free energy to exist that you can gobble up alone. Like in a stock market, a single person couldn't beat the market consistently.

Robin Hanson: Let's talk about that. We did this chapter on conversation. I can say that academia fits a lot of that model of conversation. If you ask what academia is for, people will say intellectual progress, advancement of the frontiers of research. That's like saying conversation is for sharing information. It's not true. What academics are trying to do mainly is to show off, just like individuals are showing off. Most academics are trying to produce their next paper and have it be impressive. When they submit it to a journal, that journal will evaluate it primarily in terms of how impressive it is—how difficult it was, how many people could do this. The referees who are looking at it do not ask how important this is, how original this is, or how useful this is. They ask overwhelmingly: how impressive is this?

Robin Hanson: Does it use difficult methods? Difficult to obtain data? Did you take a lot of time polishing it? Can we find any flaws whatsoever to pick on, or does it seem to have been very picked clean, to be very carefully constructed and elegantly put together? That's what people are looking for. If you can pass those tests, you are impressive, and therefore academia credentials you as impressive.

Robin Hanson: Academia is the collection of people who have been credentialed as being impressive in different areas. That's what the world wants from them. Students want to go to the schools with the impressive professors. Reporters want to interview the impressive professors when they call up and look for a quote. Funders who hand out money for research want to fund the impressive people. The whole system is trying to be impressive. It's not trying to be original or useful.

Robin Hanson: So it's not much of a surprise that it would be possible to be much more original or even more useful than most academics are, because that's not what they're trying to do. If you do do it in contrary to them, you won't be rewarded much. You won't gain a lot of praise or prizes or money or fame. You will just have been more original and useful and still mostly ignored.

Dwarkesh Patel: But your example seems a clear contradiction of that. I mean, you're quite a famous economist, famous enough that I, thousands of miles away, know about you.

Robin Hanson: No, no, no, I'm not a famous economist. I'm on the margin. I'm lucky to have survived in academia. I'm lucky to have found a place and gotten tenure so I can do a lot of things. But in terms of how I'm counted by academics, I'm not very famous at all. I might be more famous in a world of amateurs out there who talk to each other, but academics don't care about that. They don't care if you've somehow managed to become famous among a world of amateurs. It's not what they're competing for. That doesn't get you a job in academia. It doesn't get you a journal editorship. It doesn't get you a grant from the NSF. Having amateurs like you out there is pretty useless for an academic. So go for it if you want, they say. We don't care.

Dwarkesh Patel: But that incentive to have thousands of fans who will buy your books and pay attention to what you're saying, that's not persuasive to anybody?

Robin Hanson: No. Bestsellers are not made of thousands of sales. That's just a little better than an academic book, which is almost nothing. You want to make a bestseller out there, you'll need hundreds of thousands or millions of sales. That's where you might make a career out of selling books.

Dwarkesh Patel: Oh, interesting. Okay. So let's talk about some of the essays you've been publishing recently. You have an essay called "Dominance Explains Paternalism." So it makes sense why we'd want to dominate our rivals, but why would parents want to dominate their children? It should seem like they have a vested interest in their children.

Robin Hanson: Let's go back to basics here. Social animals, all through the history of biology, have organized themselves within the social group via dominance. Dominance is the most common way that social groups of animals organize themselves. It's the chicken's pecking order. Chickens peck at each other. Whoever's better at pecking is the better pecker, and there's a pecking order of chicken. If you think you're mispecked in the order, then you peck up and down and see if you can change your rank. That's true of all sorts of social animals all over the world.

Robin Hanson: It's always this mixture of personal advantage and social advantage. Social groups, the ideal thing they are trying to do evolutionarily is finding a way to organize the group better while having each participant's actions be in their self-interest. You're trying to make it be in their self-interest to help the group. For example, there are these babbler birds we talk about in our book, early on in the book. They are trying to achieve status, but one way they do it is by going into the top of the bushes and looking out for predators and calling for predators if they see them. That helps the group, and it's somewhat at their own expense of being more at risk for the predator, but they gain status by it, and then that helps the group. The group has found a way to get this striving for status to be in the interest of the group as a whole.

Robin Hanson: In addition, often the high status birds are found cramming food down the throat of low status birds who are resisting. So they manage to share food through this mechanism of status, whereas if you give away more food, you're a higher status bird. Dominance is this thing that's all through all animals, and humans have it too. Human dominance has many functional pieces, not just dysfunctional pieces. Individually, we have selfish reasons to be dominant over and beyond how socially useful it might be, but it is useful.

Robin Hanson: Parental dominance is one of the more useful forms of dominance we've all agreed to. When there are parents and children, parents need to make children do some things that the children aren't initially inclined to do. That's important. If the family's moving, the parent needs to drag the kid along. If the kid's at risk of hurting themselves or someone else, the parent may need to step in and limit that. So parental dominance is functional in a great many ways. Therefore, child submission is functional in many ways and acceptable.

Robin Hanson: This is also important to mention. You think of dominance as something that the dominant party is trying to be dominant and everybody else is resisting the domination. That's just not true. We talk about status moves in our chapter on body language. What we say is that humans are constantly negotiating a relative status in their conversations and interactions. It's not an equal status, but they're both agreeing to it. When you see people awkwardly interacting, that's often when they have not agreed on the relative status. Typically, they do come to an agreement, and that's not an equal agreement. One party is dominant and the other is submissive, and the submissive party is okay with it. They are choosing to be relatively submissive. We actually like being submissive in many ways, although we don't like to admit it. BDSM is a thing, and in that world, it's not just the dominant parties who are enjoying it, it's the submissive parties as well.

Robin Hanson: In humans, we have the separate kind of status called prestige. That's the one it's okay to have and to acknowledge and to accommodate, whereas we're not supposed to be acknowledging or seeking or accommodating dominance. A lot of human hypocrisy is about pretending that dominance is prestige. For example, bosses. Most people, when they think about other people having bosses, they don't like the idea. It's a bad idea. It's not right that somebody should have to do what a boss says. Their boss, however, is usually okay. Similarly with politicians, most politicians are corrupt, but not your politician.

Robin Hanson: In modern workplaces, you need somebody in charge who gives orders. That's kind of basic to the setup. There needs to be a boss and they have to be able to give orders, and that's dominance. They are threatening to take away your money, and money is dominance too. They've got a double dominance whammy. The usual human rules are, you're supposed to resist that. You're not supposed to accept that. We're all supposed to get together and defy that and denounce it and make it stop. But we're all going to work and obeying our bosses. So what gives?

Robin Hanson: One of the things that's going on is our boss's main job in many ways is to project prestige. In these ancient human groups, if somebody in the group were to say, "You all have to do what I say because I'm bigger than you and if you don't do it, I'm going to punch you," or "I'm the best hunter and if you don't do what I say, I won't feed you," that's dominance and that's not okay. But it's okay if somebody is just the best hunter and we all just want to watch him and see how he does it and we're impressed. Somebody is the best singer, the best fire maker. It's okay just to be the best at something and have people say you're the best and have people admire you, as long as you don't seem to be using it for dominance purposes. You are self-effacing and you deny it. You say, "No, no, no, I'm just ordinary." But other people say good things about you.

Robin Hanson: That's what a boss needs to do. As you may notice in modern workplaces, bosses very rarely want to give direct orders. They would rather indirectly suggest things because they don't want to take on this explicit dominance role. They want to be prestigious. We pay a lot for bosses who have impressive degrees from impressive schools, who are tall, who are handsome, who are articulate, who have immense stamina, work many hours, because that makes them prestigious. When they are prestigious, it's okay for us to do what they say or suggest, because we can say that that's prestige. We can actually go to work and do what we're told. That's an important thing, because otherwise the thing doesn't happen.

Dwarkesh Patel: There's a lot of fascinating stuff there. But I want to ask you about how much of our domination of children is actually helpful. You've been publishing on Twitter these tweets about how teaching your kids music seems to be ineffectual.

Robin Hanson: Well, you mentioned even previously about the dominance explains paternalism. Paternalism is this way that we all treat each other that's like how parents treat children. The word comes from parents treating children. As we know, parents often tell children what to do and what not to do. By analogy, we often tell each other what to do and not to do, or through the government, through the homeowners association, through all sorts of mechanisms, we treat each other paternalistically.

Robin Hanson: Sometimes we tell each other what to do because what you do might affect other people around you, and we need to limit the damage you can do to them. But often we are justifying our paternalism in terms of how you're hurting yourself, or how we need to make you help yourself. That's often what we do with children too. My argument was that in both cases, we are not admitting to ourselves how much we're just wanting to be dominant and not actually being helpful. Often we are making somebody else do what we think they should do. We say it's for their own good. But what really happens is that we make them do what we say and everybody sees that we are dominant because we made them do what we said. That feels good to us because it raises our status. Everybody sees, "Ah, you're the dominant one. We need to respect you more and associate you with more because you're the one in charge."

Robin Hanson: That's true with parents too, with children. Even though there are some things parents have to do to keep the children from running into the street, parents quite commonly go well beyond that to make kids do things for their own good. The example is teaching a child to play an instrument that came up on Twitter recently. It came from a study that said if you look for academic or cognitive benefits, like they're better at school or more disciplined or more practiced, that's just not there. On average, kids who play an instrument are not getting any of those benefits out of it. You could say, "Well, yeah, but they like to play music." Okay. I was saying, well, if they don't like it and it's not giving these other benefits, what's the point? Maybe that's a big waste.

Robin Hanson: Obviously, one of the things that's going on is a lot of parents have heard, rightly so, that in many social worlds, it's very prestigious to play an instrument. You are seen as cultured and high class if you play an instrument. Maybe those high class colleges will give you a better look if you are seen as a high class sort of person who plays violin. Many parents push their kids to play an instrument so that they can be high class. Not only does it supposedly give the kid a leg up later in competition for class, but it also makes the parents able to brag about the kid. Honestly, I might say this is a case where you're pretending to do something for their benefit and it's more for yours.

Robin Hanson: Some of that might even just be for bedtimes. When we tell a child to go to bed at a bedtime, we say it's for your own good. "You'll be cranky in the morning if I don't make you go to bed now." But more plausibly, it's because you want some time alone after the kids go to bed, and that's important to you. Also, if they're cranky in the morning, that'll affect you, not just them. This is about you. But you're not so much willing to admit that. You tell yourself you're doing things for the kids. A lot of paternalism towards kids is also going farther than necessary for helping them. It may help you, and maybe that's justified in the sense that you're part of the story here and what you want should matter too. But parents often are uncomfortable with admitting that they're just doing it for themselves.

Dwarkesh Patel: But the reason parents might prefer their own preferences over their children is just because of the discounting, because they share 50% of your genes?

Robin Hanson: Of course. From an evolutionary point of view, if we just look for proximate causes, we know that we care about ourselves above our children. Children aren't the only thing we care about in the world.

Dwarkesh Patel: Yeah, yeah. Okay. You had another post recently about remote work, where you said it's going to be a very big deal in the future, among other things. But if it's going to be such a big deal, and we've had video conferencing technology for so long, why hasn't it happened until now? Or why would we expect it to happen in the future?

Robin Hanson: When you're thinking about the long run, there are two kinds of things to predict that will happen. There's the expected and the unexpected. Among the expected, over decades there have been this long list of things that people have been predicting will happen. When you're thinking of the future, those are the easiest things to predict, the things that people have been predicting for a while. The unexpected thing that nobody sees yet will happen too, but it's much harder to predict.

Robin Hanson: The most basic prediction game is to look among all the things that people have been predicting and ask, "Which of those is finally going to actually happen?" Because a lot of things are predicted well before they happen. It's easy to confuse a clear view with a long distance. You're standing on a mountain, you see across the other mountain, and you see it clearly, but that doesn't mean it's close. It just means you can see it clearly. When you think about the future, there are many ways in which people say, "Ah, that's got to happen eventually." And then they often think, "And so it's about to happen now," which is not quite the same thing.

Robin Hanson: Space is a very dramatic example of that. People have been able to see for quite a while that eventually space is going to be a really big deal. That doesn't mean it's happening yet. Or even fusion energy. If you look at the fundamental sources of energy that are possible, you say, "Look, fusion, look how powerful that is. That's got to happen eventually. Eventually we've got to tap into that." Is it going to be right now? Not necessarily.

Robin Hanson: The game for most of these things is to ask, "When will it actually happen?" Not just in principle that it will eventually happen, but to look at the details of what it takes to enable it. With telecommuting, many people have been predicting that way in advance when not much happened. People struggled to say, "We were predicting telecommuting 30 years ago, but it didn't happen much. Why didn't it happen?" One of the things that goes on is you make these predictions based on naive theories about how things work, and then you need to refine them.

Robin Hanson: For telecommuting, many theories were based on very simple-minded theories of what people are doing at work. People have correctly said, "At work, you have to not only do your job, you have to negotiate politics. You have to form alliances. People have to be able to read you and your loyalty. You need to lobby for your side and your allies." Those are all things that are harder to do when you're back at home, especially with very limited communication mechanisms such as we had in the past. The question is, are those obstacles sufficient to block it forever, or do they just block it for a modest period of time?

Robin Hanson: To think about a change like this, you want to think about two things. You want to think about the things in the way, like playing politics and negotiating things online versus in the office. And you want to think about what's the draw, how big a payoff could this be? I think people have correctly understood the political obstacle, but they haven't really fully appreciated what the big draw was. That's the thing I was trying to call attention to in my recent writing. The draw here is much bigger than most people realize. That's why this is going to be a big deal. These obstacles will be overcome because there's such a huge draw.

Robin Hanson: The huge draw is agglomeration. Cities, bigger cities are much more economically productive than smaller cities or no cities. The reason is when you are in a big city, there are just a lot more different interaction partners you can have. There's a lot more room for specialization. In a small town, there might be one Chinese restaurant. In a big city, there can be 100 Chinese restaurants who can then specialize in different areas of Chinese food and different kinds of restaurants and specialize in being low quality or high quality. There's just lots of room for specialization. That gives a lot of payoff, all these different specialized roles.

Robin Hanson: Remote work allows planet-wide specialization. That's the key thing to realize. It can be as if the entire planet was one big city. That's an enormous increase in productivity when you are able to take advantage of specialization. There are many industries today that have specialized internationally as it is. For example, Google or Apple. There isn't one of those in every city or every country. There's one of those in the world. People in those companies can specialize in one particular task, and then the benefit of what they do spreads across the entire world. They make the iPhone slightly better in one way. Everybody in the world gains from that by getting that iPhone.

Robin Hanson: There are many other jobs we do that we haven't been able to specialize so that there's a world involved. Think of a plumber or a hairdresser. Your local hairdresser doesn't have to be the best in the world at what they do, like the person at Apple working on the phone has to be, because you can't go around the world to get a haircut. You have to go near where you are. Your local hairdresser is only competing with a limited number of hairdressers in your area. They can specialize that much. The number of people they would work on is limited. There are only a limited number of kinds of hairdressers that can be in your small area.

Robin Hanson: Similarly with plumbing. Your plumber is not specialized in one particular kind of plumbing job. They have to do most all kinds of plumbing jobs that will show up in your area because you're going to have to have a local plumber who's close enough to drive over and work on your pipes. Those are kinds of jobs that we have not gained this specialization where there's just somebody in the world who's best at each thing and they're doing that one best thing.

Robin Hanson: With remote work, if you could actually have people remotely cut hair or remotely work on pipes, now they could come from a very long way. There'd be this huge region from which you could draw people to do it. They could specialize. There could be the person who knows about just your kind of pipe and just your kind of job. During the job, if there are five different parts of the job, each part could be done by a different person who specializes in that part of the job. That's an enormous potential productivity. That will be the draw that makes us willing to pay these extra costs to figure out how to do things online, which are harder and so far have been the obstacle.

Dwarkesh Patel: Let me offer up two concerns and see how you assuage me. One is Caleb Watney wrote an article in the Atlantic a little while ago where he said remote work might actually decrease innovation and productivity growth because there's no way to replace the physical agglomeration effects of just being in a big city. Two engineers in Silicon Valley run into each other and quit their jobs and start a company. Remote work won't replace that. The other is that specialization will mean that the 10% best plumbers will be doing all the plumbing and 90% of plumbers will be out of a job because they're just not good enough if they can just compete with each other.

Robin Hanson: That second thing is exactly what we want and exactly what we get in other areas. Yes, we want the best people to be doing it. The other people go do something else. That is the nature of the modern specialized economy. Yes, that's what we want. More of that, please.

Robin Hanson: With respect to the first, it's about to what extent can you have more interactions, and to what extent is being physically close important for that? In the past, most of your interactions were physical and direct, and you had this minority of weird interactions online. That minority of weird online interactions just could not fill all the social roles in your life. If you tried to do that, it didn't work very well because most people weren't doing those social roles online. They were doing them in life.

Robin Hanson: If you and your co-workers, you stop by their office and chat, they go out for drinks after work, come over to their house for a birthday party. Those are all important social interactions. You could chat with somebody at the birthday party about your new idea for an engineering product. You might have a speaker who comes into town and you'll go off and listen to the speaker and you chat with other people after the speaker about this talk. When you're in a city, there's just more of those things you can do. The closer you are to someone, the less far you have to drive to get to their birthday party. You just get more opportunities to do things.

Robin Hanson: As we do more things online, then those things can happen online instead. If you have the birthday party online, then you'd be chatting with people at the birthday party online. If you're going out for drinks online, then that's where you would chat with people. If you have a business meeting and you chat afterwards at the business meeting, it happens online, then those things would happen there. The question is just, can all those things happen there? That's the big question.

Robin Hanson: As we slowly move more kinds of interactions online, they can become the basis for innovation. They can become the basis for serendipity and new conversations and new groups that you form. But we have to work to make those things happen because we haven't been doing them online. Initially, when we try to do online, they're awkward and not very satisfying. We have to work to find ways to do them so that they can be more natural, fluid, and satisfying. That'll take a while. That's why I was making this prediction that it's the biggest thing over the next 30 years, not the next five or even 10 years.

Robin Hanson: When people make predictions about the next big thing, they usually talk in the next 10 years. I've learned enough over the years to go, no, stuff doesn't happen that fast. 10 years is too slow. You want to be predicting maybe over the next 30 years. That's enough time where big things can happen. I say maybe a third of work would be done online in 30 years. Maybe even two thirds. That's a challenging thing to do over 30 years.

Dwarkesh Patel: Okay, so I'm studying computer science in university. Is there something I should be doing differently or some way I can give myself a competitive advantage knowing that remote work is going to be a big thing in the next 30 years?

Robin Hanson: Computer programming has been at the vanguard of remote work. That's an area where people are creating more best practices and standard practices that are using remote work. As you may or may not know, most work done outside of school is collaborative. In school, you tend to want to do your own projects and not collaborate with other people. For the purpose of learning what it'll be like to be out of school, it's better if you try to do more collaborative projects in school. If the teacher won't assign them, then assign them yourself. Do projects where you work with people. As long as you're working with people, try some of your projects to be online collaborations. That's how you'll learn how to do it.

Dwarkesh Patel: Okay. And do you have any other advice for a generic 20-year-old?

Robin Hanson: Life is a long time. You will live a long time. You have a long life. Don't tell yourself that if you screw up right now, life is over and it's too late ever to fix it. That's just not true. I started my PhD program at the age of 34 with two kids, age zero and two, and managed to get a PhD and become a professor later in life than most. Whatever you pick as a thing you're going to focus on, if you stick with it for many decades, you can do a lot. Just know you have a lot of options ahead of you. It's not all about right now or it's too late.

Robin Hanson: Early in life, you want generic skills. Learning the latest particular programming language or toolkit, programmers are often really into that. They're very faddish. They're always, "What's the new skill I'm going to learn that I'm going to put on my resume? Look, I'm doing blockchain," or whatever it is. That's a pretty short-term investment. Early in life, you want to be making the long-term investments. Learn how to write, learn how to talk, learn basic analysis. Learn how social groups work, how to coordinate with people. Learn basic skills that you can use all your life. That's a fact of all people, all young people everywhere. This is your time of life where skills and abilities collected now can be used across your entire life, across a changing and varied life. Don't make so many bets about exactly what you'll be doing. Collect a lot of relatively general, long-lasting skills. That includes social skills of having associates and convincing them you're with them and that you like them and that you're listening to them.

Dwarkesh Patel: If you were 20, would you still go into social science or academia, or would you be working on another set of problems?

Dwarkesh Patel: So my path, the way I saw it at the time was focusing on what seemed the most interesting topic to me. And then once I learned key things about it, that changed what I thought were the most interesting topics. And then I switched to what were now the new more interesting topics. I repeatedly did that until I got to a place where nothing else beat that as a more interesting topic, which to me meant it seemed to be very important and have high potential. I don't see an easy way to shortcut that. Unless you happen to stumble across the most important topic right at the beginning, you won't really know what the most important topics are until you think about them and dive into them. It seems like a time and attention expensive algorithm.

Robin Hanson: Right, yes. So basically, it depends on how much you're going to take on the task of finding the most important things. Most people you see just stumble into a local world and say, "This world seems like it'll take me. It seems like a world I might be happy enough in. So I'm just going to be in this world and succeed in this world." But they're not going to make the judgment about the relative importance of this world compared to other worlds. That's a lot of work. They say, "These people say I'm good for them. They seem to like me. They seem to have sufficient prestige and resources to be okay to be with them, so I'm just going to stick with them." Or, of course, they reject you, and then you go look for the next thing, and you keep going until somebody accepts you. When people do it that way, they're trusting the world to know what's important. The world itself is telling you, "Oh, you're good at this, we're willing to pay you," and you say, "Okay, I guess I'll do it." That might be perfectly fine for most people, but it seems to me that some people should be asking, "What's absolutely important? What overall matters? What's missing and neglected such that somebody should do it, but nobody is because it's missing and neglected?" So in some senses, what I specialized in is saying, "I'm going to take an overall evaluation. I'm going to look at as many things as I can. I'm going to look for missing and neglected things, and then I'm going to try to do those."

Dwarkesh Patel: There's no way to look for missing and neglected things without taking on yourself the judgment of looking at a wide range of areas and asking how important they are. It seems like a case of asymmetrical information where if you're already burrowed into a path, you're not aware of what somebody else might know about what the neglected areas are. Is there a way you can kind of fine tune the problem?

Robin Hanson: Honestly, it's not as hard as you might think. If you go to most people and ask them why what they do is important, most people don't have much to say. They have a few memorized phrases, but if you try to probe them on it, they just haven't thought about it. So the very first step is just to push into that level of detail in different subjects. Just ask, "Why is medicine important? Why is advertising important? Why is air transportation important?" Just have a simple enough model of the world where each piece has a simple reason why it would matter and how it matters relative to other things. Most people don't even bother to construct that sort of model.

Dwarkesh Patel: But if I've read *The Elephant in the Brain*, wouldn't I be skeptical of others' claims or even my own claim that tries to justify why what I'm working on is important?

Robin Hanson: Sure. But even when you're broken and biased, actually trying to look at a question beats not doing it. This is just a basic fact. Often in areas, there's this old saying: two people come across a bear in the woods. One guy starts putting on his tennis shoes. The other one says, "You can't outrun a bear." And he says, "I don't need to outrun the bear. I just need to outrun you." In a lot of areas of life, it's about how well other people are doing at it and whether you can do better than what other people are doing. So you don't have to be great at it. You don't have to be spectacular. You can actually be pretty bad. But if everybody else is worse, you'll be the best.

Dwarkesh Patel: That's some very good advice. I think it's the best advice. I've been asking everybody that I interviewed about their advice for young people, and I think that's the best advice I've gotten so far. Let me ask you about idea futures. If even a few firms adopted using prediction markets for their internal decisions, if you're right, you would expect them to become more successful and over the long run become dominant. So would you expect in the future the dominant firms to be the ones that have integrated these sorts of practices?

Robin Hanson: I have a post somewhere whose name is escaping me where I list 20-some odd different facts, things that firms are doing that don't seem to be maximizing profits. Profit maximization of firms is the simple default theory that we economists tend to invoke, and it has a simple justification: firms that don't profit maximize will lose out on average in competition with the ones that do. That's not a crazy theory. But when I can make 20-odd examples of things that don't seem to be profit maximizing, I have to revise that theory. I have to say, "No, that's enough data to say that's not quite right." It doesn't mean firms aren't profit maximizing across many margins. We've talked about many organizations having a long view on some margins, but how many margins, how many areas are they taking a long view? For example, a firm might well be profit maximizing in terms of how many ounces of peas are in the can, or how thick the wall of the can is, or how long they stew it for. There might be a limited number of parameters where they just really carefully tried to maximize profits with respect to those parameters, because those are pretty important parameters and it wasn't really a problem. But if you look at this other list of features that firms don't seem to maximize profits on, that tends to be where politics gets in the way. So the first approximation of firms is as a forum or battlefield in which different political coalitions fight. The rules and procedures that will be adopted will be the ones that favor each coalition in its conflict with the other coalitions, even if that's not at the benefit of the firm as a whole.

Dwarkesh Patel: But even if that's true, over the long term, you'd still expect the firms that have optimized for minimizing the impact of politics to be dominant.

Robin Hanson: That's very hard. The important thing to realize is we are still early in the stage of optimizing organizations. Two centuries ago, there were basically not large organizations. Almost everything was done by small organizations. Even nations were run by smaller organizations. Over the last two centuries, we have gained a lot of experience organizing people into larger organizations. We've learned a lot and they've changed a lot over that period. We are still a long way from optimizing organizations. I don't know how long it'll take, but it could well take several more centuries at the rate we've been going to figure out how to set up organizations so that they actually do maximize profits. The temptation is to say you're maximizing profits and then do whatever helps your coalition. Firms are run by people in coalitions, not by some abstract algorithm. Until you can get those people to do what's in the organization's interest, they will do what's in their interest.

Dwarkesh Patel: So you seem skeptical of the odds that primary players will reform their institutions and how they make policy. When I wrote that post where I imagined the press conference with reporters who care about the betting odds of different policies, as you said, it's not likely because reporters aren't going to care about policy effectiveness. But couldn't I say that the American Revolution was a time when our founders did care about the incentives and constraints of the institutions they created? So maybe it can happen once, and it will keep happening again and again.

Robin Hanson: There are two things to keep track of. One is that things change sometimes. The other is that people on purpose change things to be better. We tend to focus on that second process, on people looking at institutions and deciding what to do because they think that would be better. But honestly, that's probably a smaller process and matters less than the first fact that sometimes things change. The key to innovation is just that sometimes things change. When you don't change, you can't innovate. A lot of these things just don't change that often. You talk about the U.S. Constitution. There are a lot of ways now, 250 years later, to say that we don't have a very good Constitution. Maybe it was state-of-the-art back then, but it's not now. But it doesn't change. Similarly for many business practices, they're just standard business practices and everybody does it that way, so those don't change. The fundamental thing that needs to happen in general for innovation is just for there to be change, for a range of possibilities to be tried. It doesn't much matter why they're tried. It doesn't matter the motives of people for trying them or what they had in their hand or what their goals were. The main thing is that stuff is tried, because sometimes when something is tried, it works really well and people go, "Oh, let's keep that." That's the key process: when people look at something that was tried for whatever reason and notice at least some part of it that's going really well, and then they want to keep that. Even at a finer level, you notice a thing that seems to be going well, except some other part of it is going badly, and now you try another variation. You say, "What if we did this thing instead? Because maybe that'll keep this good part and cut away the bad part." You need places where people are trying and trying and trying by exploring a space of variations for looking at what works. That's the key to innovation. In software, that's how software innovation works. You write some code, you try it differently, and then it breaks in one part, and another part doesn't do what you thought, and then you keep having to retry it and rechange it until you find a thing that all works.

Dwarkesh Patel: The conventional wisdom on our constitution is that it was a good thing that our founders made it very hard to pass new laws, because there are more ways for things to go wrong than to make things better. But you would disagree with that? You would think it should be easier?

Robin Hanson: Both are true. The vast majority of innovations are bad. The vast majority of changes in any area of the world are bad. That's true in medicine and everything. Nevertheless, we're only going to have a net change if we try things out. We both have to try a lot of things and expect most of them will be bad and we'll throw most of them away so that we can accumulate some better things. Unless you're willing to try some things even though most of them are bad, you won't get the better things.

Dwarkesh Patel: So you'd prefer the British parliamentary system where they try socialism and it doesn't work and so they go back to some sort of capitalism?

Robin Hanson: For innovation, you can go too much. It's certainly quite possible to have too much innovation. What you want is enough innovation and, of course, to have people have incentives for choosing the better ones. You do have better innovation if you can give people better incentives for picking things that would work better. That's why a venture capital system of innovation might work better than a government agency that just tells people to try things, because they might have better incentives to pick better things. If you just change everything at random, that's too much change. You need to usually just be changing a few things while you keep most other things unchanged, paying close attention to whether it's working better or not, and being ready to abandon it if it's not working well.

Dwarkesh Patel: What's your favorite political system in the world that does this well?

Robin Hanson: As you know, I have a proposal for something called Futarchy, which is based on betting markets. That would be the thing I'm proposing. But an existing system? Demarchy is interesting, where you put people randomly in office and see how they do. That was something the Greeks had. Random sortition is another name for it. They randomly assigned people to political offices.

Dwarkesh Patel: Did that work well?

Robin Hanson: That's the question of whether Athens worked well. It was time consuming and expensive, but there's a wide range of governance. I think people glamorize governance too much. People are focused on the United Nations or the World Bank or international or national organizations, but honestly, most governance happens at a local level: your church, your homeowners association, your firm. Most innovation should happen at that level too. What matters is innovation at those small levels, and then the larger organizations will copy the things that work at smaller levels. I'm actually not that interested in international governance or national governance because so many people are lobbying to influence it. I won't have much of an influence and it's not really a place to do innovation. You want to innovate in small organizations and then adopt the best practices at the larger ones.

Dwarkesh Patel: So your view of governance is very much evolutionary, just like random mutation and selection?

Robin Hanson: Not so much random, but you'll need a lot of it. Even when you're trying not to be random, you are in fact being pretty random. The basic fact is there's a lot of things you don't know. There's a lot of context you're not taking into account. So when you're making your best guess, you're still being pretty random.

Dwarkesh Patel: I sense a lot of skepticism towards experts who are sure that they know what the next step is. Is this just a high-accusing thing?

Robin Hanson: It's very easy to be overconfident, but of course, you definitely want people who know best when you can, unless they are being so overconfident as to destroy their advantage. That's just generically true. There are many areas where the person who's most known for being best is so overconfident that you might be better off picking random people in that role than somebody who's that overconfident. But if you actually have the best person at it and they have a decent incentive to not be overconfident about it, then of course, yes, you want the best person.

Dwarkesh Patel: Thank you.