Playback speed
Share post
Share post at current time

#14: Robin Hanson

Cultural Drift, Ems, Elephants, Institutions, and The Future

Robin Hanson is a professor of economics at George Mason University, the author of The Age of Em and The Elephant in the Brain, the writer of the blog Overcoming Bias, and one of the most interesting polymaths alive today.


0:00 - Intro

1:24 - Mathematical models and grabby aliens

9:11 - Will we run out of value in the future?

12:23 - Kurzweil’s Law of Accelerating Returns

14:29 - Posadism

17:53 - Moral progress and Whig history

20:29 - Will there be a trad resurgence?

23:00 - Will Israel’s ultra-Orthodox problem globalize?

25:39 - Why will fertility rate keep dropping?

30:14 - Is declining fertility solvable technologically?

32:20 - What is wokeness? Has it peaked?

35:02 - Will virtualization make society more multicultural?

39:30 - How do institutions coordinate so well?

42:50 - Will ems care about death?

46:16 - Personal identity and death

49:30 - How much of Age of Em is applicable to LLMs?

51:09 - Why we shouldn’t worry about AI risk

55:40 - What if people don’t see AIs as their descendants?

1:00:41 - Other future tech deep dives

1:02:43 - Our very long-run descendants

1:06:08 - Time and risk preferences

1:08:34 - Wouldn’t ems be selected for docility?

1:11:24 - How Robin got involved in rationalism

1:13:22 - Girls getting the “ick”

1:16:56 - Have humans evolved since forager times?

1:18:28 - Cultural evolution

1:20:30 - Culture and prestige

1:22:49 - Why medicine in the US is bad

1:25:54 - Is academia the best truth-seeking institution in society?

1:28:52 - Peer review

1:31:13 - Which institutions are actually good?

1:32:33 - Why universities are all the same

1:37:40 - Bitcoin and speculation

1:46:44 - Demarchy

1:50:03 - Futarchy

1:53:56 - Applying prediction markets to dating apps

1:57:38 - The broadest thinkers and books in the world

2:00:59 - How Robin balances his many interests

2:01:58 - Teaching

2:03:12 - Outro


Robin’s Homepage:

Overcoming Bias:

Robin’s Twitter:

Grabby Aliens:

Age of Em:

The Elephant in the Brain:

Beware Cultural Drift:



My Twitter:

My Substack:


Theo Jaffee (00:00)

Welcome back to episode 14 of the Theo Jaffee Podcast. Today, I had the pleasure of speaking with Robin Hanson. Like previous guest Bryan Caplan Robin's day job is a professor of economics at George Mason University. There's much more to him than just that, however. He's a world -class polymath who's worked in literally dozens of fields and was a pioneer of many of the things we love today. Before going into economics, he studied physics, worked in AI research in the 80s, and did a PhD in social science.

He's proposed over a thousand ideas for alternative institutions, most famously prediction markets, which are today a multi -billion dollar industry. He was early on crypto, too. He was friends with Hal Finney, who many believe was Bitcoin's creator, Satoshi Nakamoto. He's been involved in futurism since the 90s, creating the idea of the great filter and grabby aliens, and writing a 400 -page deep dive on mind uploading called The Age of Em. He's also into human psychology and rationality. He's written on the blog Overcoming Bias since 2006.

where Eliezer Yudkowsky was originally a co -blogger before leaving to create Less Wrong. And he co -wrote with Kevin Simler the book The Elephant in the Brain on the hidden motives behind nearly everything we do. Most recently, he published an essay called Beware Cultural Drift, warning about the danger of having a global monoculture that's slow to adapt to changes. There was a lot to cover in this episode and I had a lot of fun recording it. This is the Theo Jaffee podcast. Thank you for listening and now here's Robin Hanson

Theo Jaffee (01:24)

Okay, we're on. Welcome back to episode 14 of the Theo Jaffee Podcast. We're here today with Robin Hanson

Robin Hanson (01:30)

Nice to meet you, Theo.

Theo Jaffee (01:33)

Nice to meet you too. So my first question has to do with your idea of grabby aliens and the great filter, which are, you know, these ideas about the, you know, explanations for why we don't see aliens and how our society might move in the future. Um, so grabby aliens and rationalism both heavily rely on mathematical models, but with a lot of these mathematical models, even like small inaccuracies in the inputs can make the output like wildly inaccurate. So how do you typically account for this?

Robin Hanson (02:04)

Well, you want robust models for which that's not true. So our model is a three parameter model where each parameter is fit to data and the parameters are as follows. Basically, advanced Davidian civilizations appear in space and time. They appear at random places in space.

They also appear at random points in time, but the time at which they appear is proportional to a power law. So the power law has a constant and it has a power. And then once they appear, they expand at a speed. Those are the three parameters. And we fit each of the parameters to a particular piece of data we have.

and I claim that that model is robust in the sense it's not very sensitive to these parameters. You can change these parameters by a substantial amount and then the model only changes by a proportional amount. It's not highly sensitive to some particular choice of parameters. So I mean we could explain where these parameters come from and then you know what we've concluded from that but I'll pause and let you push farther if you want. Okay so...

Theo Jaffee (03:11)

Yeah, yeah, let's go into that.

Robin Hanson (03:15)

The speed of expansion comes from the fact that if you model different speeds of expansion, you'll find that at low speeds of expansion, you predict that each one, when it's looking out into the sky, will see many of the others. Because light goes much faster than they expand, and so they will see them a long way off way before they get here. Each will see the other one coming. Since we look in the sky and we don't see...

other huge alien civilizations taking up enormous spheres in the sky expanding at a rapid rate. We can then conclude that we must be in the parameter space of the model where they don't see each other coming and that's where they're expanding pretty fast, say half the speed of light or even faster. So we then conclude, well I guess that's the that parameter value. They are expanding very fast because

we look up and we don't see anything and most would see something in the model if they were expanding slowly. So that's one of the key parameters. Now, another key parameter is the constant in front of the power law. So as you change the constant, you will change on average when these things appear in the history of the universe.

you know, make the constant lower and it'll take longer before they appear. You make the constant higher and they will appear sooner. So we can take our date at the moment.

as a random sample from when the dates of these things appear, and that constrains the constant in front of the power law. That is, we can basically assume a uniform distribution of where we are in the distribution of alien civilizations. We could be really early, we could be really late, but we're somewhere, assume, say, a uniform distribution over the rank. Are we in the first percentile, the 99th percentile, somewhere in the middle? And that then gives us a distribution over this constant.

in front of the power law. And the power of the power law comes from the history of life on earth. So the key idea is that in order to become an advanced civilization like ourselves, you have to go through a number of difficult steps and you have to do that before the window for life on your planet ends.

and a simple statistical model of what happens when something has to complete a whole bunch of difficult steps before a short window. Usually it would not succeed before the end of the window, but once in a while it gets lucky and does. A statistical model of that says that the time at which they do succeed in appearing goes as a power law. And the power is the number of steps it has to go through in this history. And we can now use

timing of events on the history of Earth to roughly guess the number of steps that we've gone through. So a best guess of six, but hey maybe a shorter three or nine, come from...

how long it took from life to be possible on Earth till the first apparent appearance of life on Earth, and then how much is there between now and when it looks like life would no longer be possible on Earth. Those two time durations are the datums that we can use to pin down how many steps there have been, and again, middle estimate of six. So those are the three parameters. Each comes from data. Put those together into a stochastic model, a model of probabilities, and you can run it many times.

to get the distribution over what the history of advanced civilizations in the universe looks like. And from those distributions, as you vary the parameters, we can draw, say, the following two conclusions. Roughly, advanced aliens appear roughly once per million galaxies. So...

There's trillions of galaxies in the universe, so there's lots of them in the universe, but they're still pretty far apart, once per million galaxies isn't close. So that says it'll be pretty hard to see them because they're pretty far away. And the other key parameter is if we go out and become one of these advanced alien civilizations that expands and becomes visible in the universe, we will meet others in roughly a billion years.

once per million galaxies meet them in a billion years. So those are pretty specific answers to pretty important questions. Now, it's not a billion point one or something precisely, it's in the ballpark of a billion years. All these answers are rough. That's part of the answer to your precision question. We're not giving very precise answers here, but compared to the basic question you might've asked, we now know a lot more than we used to.

Theo Jaffee (07:59)

So going back to what you said about like the steps it takes intelligent life to arise, how does that model account for like, do you think it's possible for life to arise spontaneously on like a non carbon based, non biological substrate?

Robin Hanson (08:15)

The general model doesn't care what the substrate is. It just cares what this power law is. Basically, how many steps does it have to go through? And what's the overall chances constant? And it allows for a wide range of different paths to advance life. I think it's if there are many different paths and some of them have more steps than others, it's overwhelmingly likely that it's the one with the fewest steps that will happen most often.

So we can, you know, if it takes us six steps, we can be pretty assured nothing out there is happening in less than six steps. Six steps must be the minimum. And the only other competitors out there we might interact with would be also things that took six steps, because the rest are very unlikely. So, of course, if it's five steps for us, it's five, but whatever our number of steps is, it's pretty much going to be the same for everybody else.

Theo Jaffee (09:11)

So while we're on the topic of aliens in the future, you have written like very extensively about future economic growth rates and you predict that eventually our rate of like resource and technological growth will stagnate. But how can that be possible when the like combinatorial space of atoms is just like gigantic? Like, yeah, sure. We will probably end up getting fewer actual resources in terms of like the number of atoms that we can gain access to.

But in terms of the ways that we can combine those atoms to create things of value, it seems like it would be a very, very, very long time before we exhaust that. This is kind of David Deutsch's idea of the beginning of infinity. So what do you think about that?

Robin Hanson (09:57)

So just to be clear to everyone, we're talking not the next few years. We're talking over thousands or millions of years as we get much more advanced and explore a much wider space of possibilities than we have now. Now, we already have a lot of experience, say, with computer programs, exploring searching spaces of possibilities.

I did part of this as part of my career. For example, we had a space of possible statistical models and a certain family of models. We were searching in that space for more likely models. And we found that, say, compared to a simple search, being a little clever about heuristics gave enormous advances in our ability to find more likely models. So there's really a lot to be gained by being clever about search. But it was also just clearly true that...

If you have decent ways to search, you're going to be looking at the low -hanging fruit first. You're not searching at random, you're searching for the best things as quickly as you can. And low -hanging fruit first implies high -hanging fruit last.

it means that the search has to slow down. That's kind of implied by any ability to not search at random, but to grab the best stuff first easier, is that you are able to prioritize the search space to find the things that are more likely to be promising first, which means the things that are less likely to be promising are going to be what's left over after you take the low -hanging fruit.

So it's just a matter of like how fast, how sort of just how much does it asymptote eventually, but consistently when we do search, you know, in computer systems and in design spaces, we do reach this sort of a asymptote situation where things slow down a lot. And the world economy is a big, deradic example of this.

That is, as the economy doubles, we get twice the capacity to search in all the different dimensions we might want to. Yet, the growth rate doesn't double, stays about the same, because we pick the low -hanging fruit first. So, you know, the fact that economic growth is steady is a strong testament to the fact that the search gets harder with time, because our ability to search gets much bigger, yet we don't find stuff that much faster.

Theo Jaffee (12:23)

What do you think about Ray Kurzweil's idea about accelerating returns?

Robin Hanson (12:28)

I mean, there can be parts of search spaces where you, you know, find things and you find more things and they help. So just generically, if you think about any system that has some sort of reinforcing process, the reinforcing process can either accelerate growth or decelerate growth or hold it steady. Those are the only three mathematical possibilities, right? Our experience with systems is that overwhelmingly we have the decelerating growth.

And once in a while we have things that are constant growth, and then more rarely we get accelerating growth, like a nuclear bomb is accelerating growth, right? But accelerating growth typically doesn't continue to accelerate forever. It accelerates over a range and then it slows down.

Like, you know, many of you listeners probably have personally experienced some point at which you were trying to figure something out and then you started to get it. There was a period of accelerating growth where things were falling into place and you were able to figure things out. In fact, they figured one thing out, helped you figure another thing out, and that was great. It was fun, but it didn't last forever. In your personal experience, it ran out a bit, and then you experienced decelerating growth.

So, you know, I just think we have enough experience with lots of kinds of systems. Like you might say the surprising thing is we have had continued to have at least steady growth in humans for a while, then we've had these jumps to faster growth modes. That's the most dramatic deviation from this expectation of decelerating growth is that we've got overall continued growth of human civilization. So the question is how far you expect that to go. And...

Again, once we reach the limits of physical expansion, like the speed of light, and we reach the limits of the materials we can work with, we have all the basic elementary particles and forces and we can't find any more, we'll search the space of all the vanes to arrange them and we will reach diminishing returns.

Theo Jaffee (14:29)

I see. By the way, are you familiar with pesadism?

Robin Hanson (14:33)

Posadism. Nope.

Theo Jaffee (14:34)

Passages um this guy I forgot his first name, but his last name was Posadas he was like a communist and he came up with this idea of Yeah, J Posadas He was yeah an Argentine Trotskyist who had this vision of alien civilizations Basically, he believed in communism. So he believed that you know kind of like he took Karl Marx's theory that like

the end state of human civilization is communism to its logical conclusions that advanced aliens would be communists too. So civilizations should search for advanced aliens because if advanced aliens were to find us, then since they would also be communists, they would help us lead the global communist revolution until you go over the world. I ask because it's basically one of the very, very few other people who I've encountered who's thought about this in the same way that you have.

Robin Hanson (15:32)

I find it remarkable that people can look at our recent history of cultural change and take some recent trend in our cultural change and imagine that that trend will apply over billions of years when it hasn't even applied over 100 years yet. This is one of the key blind spots humans have.

Basically, we are driven by cultural evolution and our cultural evolution makes us change our cultures fast, but we're kind of blind to that and you know, we tend to think our culture is best and that whatever issues our cultures have with other cultures then we're right and they're wrong and we're gonna be right for the entire future of the universe and that you know...

Liberal democracies say will be what the universe wants or commune or whatever it is. I just think if you look at how much has changed in such a short time in the past and then try to project the last few hundred years of human history into billions and trillions and more years of the future of the universe, it's just really hard to imagine that we have gotten it right in terms of the fundamental cultural issues if we didn't even notice them a few centuries ago.

Now, I might say we could find some things that are more robust. Robust issues are centralization versus decentralization. We can sort of see that that's a robust issue. Robust issue is degree of competition versus coordination, how many scales of organization there are. Those, in some sense, we can define issues as long -lasting issues, issues of what the unit of mines are be, how large would mines be.

How much do they merge into hive minds? At what scale? Whether natural selection continues? What are the forms of preferences creatures have? How do they encode them? How well do they know them? These are some of the things we can identify as pretty robust issues that last for a long time. But to think that our temporary answers to those questions, we could be confident will be the best answers across.

vast time scale of the universe seems kind of crazy.

Theo Jaffee (17:53)

Hmm. Um, David Deutsch also writes about this kind of idea, but his perspective is more like we will continue to improve over time, just as, you know, our morals so far have improved from, you know, being kind of forager values where we would fight constantly to being farmer values, or you'd have like a despotic king ruling over people to being more modern, like liberal, secular, democratic values. And that we continue to improve values into the future. But.

Do you think more values in the future will be arbitrary and we won't see them as improved? We won't necessarily see them as improved.

Robin Hanson (18:30)

I mean, first of all, you just have to notice there's a selection effect. If every cultural thinks it's best, then when it looks back at its history, it's gonna see things improving. That's more directly implied by cultural arrogance. All cultures think they're best. That's also accrued across space at any one time. Each culture thinks it's better than all the rest. Yes, well, also Whig foreign policy.

Theo Jaffee (18:49)

Whig history?

Robin Hanson (18:55)

you think you're best than all the other things that coexist with you or all the things you might imagine, but basically you can therefore predict that the future will think they're best and they will have seen more improvement. The question is whether you would approve of their improvement. And this is actually something I've been thinking a lot about lately and even in the last few hours.

Old people like me have seen our culture change in our lifetime. And then we are expected to embrace those changes, to think that our culture is now better than it was when we were young. But when we were young, we assimilated the culture of our world when we were young. And then later on, the world changes and its culture changes, and we're supposed to change with it. We each have to ask, well, was I wrong back then?

This is more right, but have people offered me arguments for that or they just telling me, you know, you're out of touch old man And I think it's hard to To you know engage that really I Don't think the world really offers as much evidence that in fact their new values are better than old ones They offer conformity pressures and sanctions if we don't agree but We're not usually persuaded

more conjured, pressured into accepting the new value changes.

Theo Jaffee (20:29)

Do you think that as these kinds of changes keep improving that we'll see like an increase in like return to tradition movements? Not just like in the sense of mainstream conservatism, but like I'm seeing this a lot more now, like trad Cath people on the internet who are trying to be like, you know, medieval Catholics or something.

Robin Hanson (20:51)

Well, the dimension I would call your attention to is diversity and variety. Our world today has vastly less variety of culture than we did a few centuries ago. So three centuries ago, basically the world was divided into the hundreds of thousands of little tiny peasant cultures.

each of which was pretty independent but pretty near subsistence. And if they drifted off the rails, they got punished by famine or pandemic or invasion, and that held them pretty close to functionality. Selection kept them in line. But then we merged small peasant cultures into national cultures, and then we have seen the rise of a world culture. And that means vastly less cultural variety.

And that, I think, means we should expect our cultures are going off the rails. We have vastly less cultural variety and vastly less selection in that our cultures, our few world cultures, are quite rich, peaceful, healthy, and if they go off the rails it'll actually take quite a while for them to fall because they have a lot of slack to survive that. And that's what I think is going to be happening over the next few centuries.

and the solution is some sort of variety that will, you know, defy the mainstream culture and do things different, like say the Amish or Heretim, and then in a few centuries they will grow and rise and replace the dominant cultures. So in that sense, I think in a few centuries' time scheme, yes, you will see the revival of...

traditional cultures, they will be different. Some of them will be more traditional, but the key thing is they will find a way to be highly fertile and highly insular. That is, insularity is especially the key. You can't diverge and do things different from the dominant culture unless you are insulated from it. And that's in some sense the most distinctive features of these small fertile subcultures is their insularity.

Theo Jaffee (23:00)

This talks, this sounds a lot about the biggest current domestic issue in Israel, which is you have like basically two classes of Israeli Jews. One is modern, secular, liberal, and the other is very traditionalist, Orthodox, and they have conflicts all the time because the Orthodox don't serve in the military and are like net burdens on the tax base. So should we expect to see this kind of thing everywhere?

Robin Hanson (23:23)


So this is a vanguard in the sense that if nothing else happens over centuries, these very highly fertile cultures which are growing fast will in fact replace everyone else because everyone else's fertility is going to keep declining and be below replacement. There are some transition issues to work out as small cultures get bigger.

Certainly one of them is this pacifism. This is also true of the Mennonites and the Amish in the US. They're pacifist. But I think the pacifism is mainly a strategy for insularity, because young men, when they go off to war, get a lot of cultural impressions from their fellow soldiers. They didn't want this to happen to their young men, so they made them pacifists. Once you have large enough groups to have their own military units, this is less of a problem. They can all go to war together and maintain their culture.

But this will be a transition point. So I guess the question is when the Haredim will be willing to make that transition to joining the military, but maybe in their own special units. The fact that they're subsidized is a happenstance of Israeli history. The Amish and the Mennonites aren't subsidized here in the US. They're kind of taxed, actually.

but they're still growing very fast. But any of these transitions is a risk point where they might fail to maintain their insularity or their high retention rates. And so we can't be that confident in predicting their rise in the sense that there's just lots of things that can go wrong. So for example, the Mormons are a failure case.

They were once a highly insular, highly fertile subculture, but the Mormon Church made a conscious choice to integrate the Mormons and integrate it with the rest of.

national and international society, they succeeded in that, and now the Mormon fertility is falling at the same rate as the rest of the country just 20 years behind. They are not going to succeed in being a insular fertile subculture, and that sort of thing could happen to the Haredim or the Yamash or others, but if there are enough of them doing things differently, it won't happen to all of them.

Theo Jaffee (25:39)

Why should we expect mainstream society to continue decreasing its fertility rate?

Robin Hanson (25:45)

Because we can see strong cultural trends that are causing it. It's not just an abstract number we can track. We can track the particular more proximate causes that are pushing it. And they seem robust and beloved. You can try to resist them, but a lot of people don't want to stop them. They will, in fact, push hard back if you try to reverse some of these trends.

Theo Jaffee (26:11)

Well, you know, in the last few decades alone, we've seen, you know, a major norm, 50, 60 plus years ago, was that homosexuality was wrong. And now, like, if you were to say that homosexuality is wrong, that would be enough to get you kicked out of polite circles. So we've had, like, a complete reversal. Yeah.

Robin Hanson (26:27)

So cultures do change, but the question is can you cause them to change in a desired direction? So the key thing is that the main way cultures change is that different factions fight over the changes and compete to influence the changes. And like if you look at the culture section of a newspaper, what that really means is these are the people most respected for...

influencing the direction of cultural changes and not everybody gets to play on an equal battleground in that space. So culture definitely changes and it's changed big time, in fact so much and so fast I think you should be disturbed by whether...

We can believe that that's all functional and adaptive, but it's not willing to just be changed in any particular direction in any particular subgroup once. Each group that tries to push for one change will typically face other groups that are pushing the other way. The question is, who will win?

Theo Jaffee (27:24)

So why would it be unlikely that the pro -fertility push in mainstream culture would win?

Robin Hanson (27:31)

Well, we can see a number of more fundamental causes again. So for example, I just over the weekend read a book by the classic founders of the field of cultural evolution, Boyd and Richardson. And in a 2004 book, their favorite model was that.

Basically, all culture needs an idea of prestige and people copy prestigious and a lot depends on what counts as prestige. And so in our society, the sorts of things that get you high prestige tend to require a lot of education. And a lot of education tends to put off fertility. And that's their story for fertility decline, which is plausible.

Obviously it's also tied in with say gender equality. If it was only men getting educated so highly it would be less of an obstacle. But we do have a lot of gender equality and along with prestige going along to the people who have high education, years of education, and that's causing low fertility. Another thing in history was that rich people could make their kids...

higher status by investing money in them, but the more kids they had, the more they divided up that money. And so there was an incentive for elites to have fewer kids in order to give each one of them high status. And so if we looked at prestige of individuals, signs of individuals, at that.

money could be invested in than that produced a selection effect to lower fertility even among elites centuries ago. But in addition to that, we have trends toward more parental care. That is, we have higher standards now for how much attention parents should pay to kids. We have standards of switch from what they call cornerstone marriage to capstone marriage, whereas in the past you would marry somebody young when you're less formed, less clear where you would go or succeed. Now the standard is more you should wait until you have a

steady, successful career and you've formed your personality, you know, just what your hobbies are and then you should find somebody who matches all those things. But by the time you do that, there's much less time for fertility to happen. We have norms limiting grandparent involvement and raising of kids and in kid careers. You know, there's just a whole bunch of these trends and many of them are quite beloved.

Urbanity, stronger urban living, which also seems to pretty clearly discourage kids. Less religion, religion's always been pretty strong. Correlate to fertility. And most people are not very open to reversing these trends.

Theo Jaffee (30:14)

Hmm. Well, do you think that this kind of thing could be solvable technologically? Like if there's some technological innovation that lets women get pregnant, you know, in a healthy way when they're 50 or 60.

Robin Hanson (30:20)

Oh sure, but the -

Well, the simple thing would work. I mean, the simplest thing that would work is just to have men and women freeze their egg and sperms at the age of 20 and then unfreeze them when they're ready to have kids, even if that's the age of 45. That would work, but it's a big ask because most people don't want to do that. Another thing that would work is to pay parents.

to have kids and borrow the money from the future tax revenue those kids will pay. That would also work, but again, you'll have to be inclined to want to solve the problem. And if you do that, it's gonna cause changes in some of these beloved trends. But I mean, I'd say there are ways we could, if we got our head into it, solve the fertility problem, but the fertility problem is really only a symptom of a deeper problem. And we are...

We have much worse options for the deeper problem. The deeper problem is just we have very few cultures which rapidly change and we're weak selection pressures and that can plausibly make fertility go off the rails but can also make lots of other things go off the rails. Norms of, you know, when you, what medical treatments you use or norms of war and peace.

Theo Jaffee (31:39)

Like what?

Robin Hanson (31:49)

norms of family. I mean, our life is full of social norms and prestige markers, and they can all just go wrong. The space of possible cultures, most of it isn't very good. Generic idea of evolution is selection will keep designs and structures in the small part of the space that's the good part of the space that's functional and productive, and random drift takes you off to the bad parts.

Theo Jaffee (32:20)

So what do you think about wokeness? Is that like a concept that is meaningful? Is it like a, you know, symptom of a maladaptive culture? Is it like a cause of a maladaptive culture? Do you think it's peaked?

Robin Hanson (32:32)

it's, it's, it's, I mean, it's just clearly, it's not peaked, but it's clearly just evidence of cultural change, and if you think about it, there's no particular reason to think it's functional, not that there's no particular reason to think it's dysfunctional compared to any other cultural change, but just realizing how rapidly our cultures are changing.

you have to realize there's nobody driving this train. There's no guiding force that's there to make sure these things are channeled into more productive, functional, adaptive forms. That doesn't exist. It's not how it works. We are just in a world where culture changes pretty randomly. And you have to realize that if this is a fragile thing that is valuable when it works right and it's in a functional structure, if you just make random changes,

Pretty soon those won't be very good.

Theo Jaffee (33:27)

Hmm. Is that also your explanation for, let's say, the current dysfunction that California has?

Robin Hanson (33:37)

I'm less willing to sort of attribute things to very particular cases. You know, this argument is strongest at the general level and it's just harder at the specifics. But...

I'll note that most people really want to argue at the within culture level. So, like the culture sections of newspapers or a lot of op -eds or a lot of things are basically people passionately arguing about which way their culture should go. And when they make those arguments, they will refer to sort of who's more prestigious and who's with us and who we are and what we have valued in the past. And those are the resources that you can use to persuade people that your story of where we should go is right and people should follow in your direction.

And most passionate discussion in politics and culture is about that. It really isn't looking at a distant point of view of how cultures evolve and what that might mean. It's about here we are and I want to go this way and you want to go that way and I'm right and you're wrong.

And part of that story is to look in the past and say, you know, where we've came from, from there to here is exactly the same thing I want to do, continue going from here to the future. And people want to appeal to our shared sense that what we did, our changes in the past must have been good changes in order to argue for how new changes should also be pursued.

Theo Jaffee (35:02)

So let's talk about this hypothetical future that one of my previous podcast guests, Greg Fodor, talks about extensively that I think is plausible, where eventually virtual reality technology will get good enough and reliable enough so that people will essentially virtualize their whole lives. And when that happens, they will have a much greater degree of control over the level of participation in society that they have compared to today. So do you think...

that could be like a solution to this kind of monoculture problem if people are capable of simply just removing themselves from the culture and creating something entirely new.

Robin Hanson (35:44)

Compared to the past, the distant past, our world is largely virtual in the sense that if you look around you, most of the surfaces you see are artificial surfaces constructed to have artificial appearances. We're not out in the woods or the jungle or the sea. We are mostly in artificial worlds constructed.

primarily as we see them for their appearances and how they are convenient for our lives. And we have enormous abilities compared to the past to find other compatible people to interact with and to form whatever subcultures we want to. And that's the way today is different from the past. And that does not imply, as far as I can tell, cultural diversity. That's not the natural outcome of that change. So I don't know why continuing along that same path into the future.

with more virtuality and more ability to select your associates would make it any different. Basically the highest level point is that as the world finds it easier to communicate with each other and to travel to meet each other and then to trade with each other and to move from one place to the other that makes the world culture more integrated. Now...

we have more variety of things like musical genres or TV shows or some other, you know, maybe particular hobbies of quilting or whatever it is. It becomes possible for there to be more such things in the world and for people to find a smaller niche closer to the kind of quilting they like, say. But those features, those kinds of cultures don't actually...

say very much about your life. They just, you know, those don't change how many children you have or whether you value living with your kids or whether you value your career or how you feel about death. Mostly they don't. Mostly they just a separate part of the world. Rationalists aren't really very different. So...

Theo Jaffee (37:38)

Well, sometimes they do.

What about rationalism?

Robin Hanson (37:47)

Rationalists, of course, are especially low fertility, so they are very much integrated with the lower fertility elites in the culture. And it's a... I mean, I think there's a strong emotional desire to see yourself as part of a distinctive subculture, but you don't usually ask for that subculture to influence very many aspects of your life. And rationalism doesn't influence very many aspects of a rationalist life.

Theo Jaffee (38:11)


What about the polyamory part? That seems like a pretty clear departure from social norms as a result of in online culture.

Robin Hanson (38:20)

It is, but if you look at the world as a whole, the world has been converging culturally for a long time pretty strongly, and that's the dominant trend. So...

I mean, you know, go around the world, you definitely see things that look different around the world. Buildings are different and clothes are different sometimes, and, you know, holidays are different, maybe even work hours are different, but the world is converging quite a lot, culturally, and quite strongly. So, for example, if you look at regulation around the world, it varies hardly at all. Compared to having 150 countries worth of...

potential variation, we actually have far, far less actual regulation variation. You can certainly saw that in the pandemic where the whole world basically did it the same way. We see it in lots of other areas. Elites especially are converging culturally around the world. Non -elites are farther behind on that trend, but non -elites are usually farther behind on most cultural trends. That's because elites lead the way.

Theo Jaffee (39:30)

So how do institutions coordinate so well? Like during COVID, how every university and almost every media outlet and the federal government and almost every state government was basically on the same page pretty much all of the time, at least for like a year, a year and a half, two years. Is it just culture?

Robin Hanson (39:47)

Right? That was primarily culture, especially culture of elites. The very beginning of the pandemic, the usual public health experts gave the usual advice, and then elites around the world suddenly started talking to each other intensely for a month or two. And at the end of that, they came to a very different conclusion about how...

world should respond to the pandemic, the official public health experts immediately caved and changed their mind and accepted the new perm announcement of the new elites, of the elites, and everybody in the whole world did it that way, same together, because that was the consensus of elite culture worldwide.

Theo Jaffee (40:29)

You said earlier something about when it comes to culture, like nobody is in charge, but is that not like a counter example? Because you said that a lot of elites -

Robin Hanson (40:36)

Nobody was in charge of it. I mean, culture produces conformity and correlation without anyone being in charge. That's kind of the key nature of mobs, basically. Culture is a mob.

Theo Jaffee (40:47)

So it would be like a category error to say that the public health experts were in charge or that the elites talking to each other were in charge.

Robin Hanson (40:56)

Well, the elites talking about each other constituted the elite culture and the elite culture decided, but it wasn't any individual person or institution. It was the culture as a whole.

And so that's a form of organization humans have long had, is gossip producing consensus of mobs with shared opinions and even shared mob action without any center to direct it.

Theo Jaffee (41:31)

What are the characteristics of elite culture that make it powerful? Or is it not the culture that makes it powerful? Is it just the, you know, elite human capital?

Robin Hanson (41:42)

The superpower of humans is cultural evolution. That is, humans are able to learn and change much faster than other animals. And the main way we do that is by passing things on via culture. But culture doesn't work if you copy at random for your associates. You need to differentially copy from those who have been successful compared to others.

And so prestige is an important part of our strategy for differentially copying from the successful. So culture doesn't work really without some concept of procedure success that you can use to decide who to copy. So that makes prestige very powerful because we're all inclined to copy the prestigious. So that means.

there are people who are prestigious and they get together and talk and they agree on things, the rest of us are going to cave and go along, for the most part, for whatever the prestigious decide. That makes the prestigious very powerful, because that is the main vector of cultural evolution, is for people to copy the prestigious.

Theo Jaffee (42:50)

Alright, so let's switch topics a little bit. I'd love to talk about Age of Em, which is probably one of the most interesting deep dives about the future that I've ever read. So one thing that kind of stuck out to me is when you talk about Em's copying one another. Oh yeah, and for the audience, Age of Em is a book about a hypothetical future scenario where the ability to emulate a human brain on a computer becomes possible and cheap and widespread and the implications of that.

Robin Hanson (42:55)


Theo Jaffee (43:21)

So, when you talk about Ems going off into stubs that then, like, end, you say that Ems won't think about it as, am I about to die, but will I remember this? But you kind of skip over the actual philosophical idea of, like, do stubs die?

Robin Hanson (43:45)

Right, so the key idea here is that we often have difficult philosophical questions associated with identity and death and things like that, and some of us think about those things philosophically and try to analyze them, but the vast majority of us don't. And for the vast majority of us, we don't really think about them that much. We just do whatever our culture says to do, and do that happily, without much reflection.

even when the philosophers all say we're wrong. So if I have a world and I want to know what they do, I don't think it's very important to think, well, what will their philosophers say? I more want to know what will cultural and economic selection produce in this world?

and then they will just do whatever that produces and they won't really know why they do it and they won't that much care. They will just do what everybody else around them is doing because that's what we do. So in this context, it seems that there's huge economic payoffs from making these stubs, which are basically short -lived copies.

which are spun off after it's been deleted shortly afterwards, but do something useful in the meantime. So for example, if you work eight hours a day, then you have 16 hours a day you're not working. So then basically your work hours have a tax of a factor of three. You have to pay for all the other 16 hours of resting up before you get the eight hours of work.

Instead, you can, when you're ready for work, make lots of copies of the work -ready versions, have them do eight hours a week, but only have one of the copies go on to rest for the next day. Now you can basically save a factor of three on labor productivity by having many short -lived copies that work for eight hours and then end instead of resting it for the next day. Now, those copies could say to themselves, gee, I'm about to die, this is terrible, and instead of working, decide to have a revolution or something.

But my claim is they won't. That is, they will get used to this as a usual practice that they are comfortable with. And they will then, because the ones who do so will get a factor of three in labor productivity, which is a pretty big advantage in a Malthusian world.

Theo Jaffee (46:16)

Hmm. So, do you think humans die every night when we go to sleep?

Robin Hanson (46:22)

I don't know. I don't care. Most of us don't care, right? That is, we know everybody else who's on this goes to sleep and wakes up and they seem to be okay with it, so why shouldn't I be okay with it? Some philosopher can analyze it and decide that I do die every night and maybe I should then be upset, but if I accept that argument, I'm gonna be upset in a way that people around me aren't upset, and then I'm gonna be weird, and I will be at a disadvantage because maybe I won't be willing to go to sleep.

It just looks like not much of a cultural win for me, right? I sh -

Theo Jaffee (46:53)

Well, is this not a question with a pretty clear scientific answer? Like, eventually, we'll be able to figure out, like, what are qualia, what is consciousness, under what circumstances do qualia persist?

Robin Hanson (47:04)

Actually, no, I don't think so. I don't think these are clear. These are not clear questions with clear scientific answers. No, I don't think so. I think they are questions that people will remain uncertain about indefinitely.

Theo Jaffee (47:07)

That will never happen.

And what about the Moravec transfer? Same thing.

Robin Hanson (47:23)

So what does that mean?

Theo Jaffee (47:25)

It's when it's like a hypothetical procedure for uploading your mind into a computer where you know instead of doing like a destructive brain scan then you have like little nanobots that scan each neuron individually and figure out what it does and then comes up with a simulation and integrates that into your existing brain and just does that again and again until your brain is very slowly replaced.

Robin Hanson (47:47)

Right? Okay. But again, most people don't think about those things. Look, in our world today, we are quite alienated from the world of our distant ancestors. Our world is quite strange, and there are many ways if you hold us to the standards of our ancestors, we are weird and even ugly and should be upset about the world we're in. We put up with, like,

Most of our ancestors would not put up with the degree of domination and ranking that we put up with in our jobs today. They would be upset and outraged and think that we just have no pride because we're willing to put up with this. But we put up with it because all the people around us do. And if we didn't, we would lose on our prestige games. Most prestigious people around us put up with it and accept it.

and we want to emulate them and be like them and so we put up with it too.

Theo Jaffee (48:50)

So, when it comes to human brains, you said there's probably, like, we will be confused about questions of identity forever, but do you think the human brain is kind of fundamentally uninterpretable, like in LLM, or will there be actual, like, computational structures that, you know, will be able to say, oh, hey, look, our, you know, memories are represented as, like,

Robin Hanson (49:10)

Oh, I'm sure we'll figure out lots about the brains, but we'll also figure out a lot about LLMs. We'll know a lot more about LLMs in a few decades than we do now. We'll be able to identify structures in them and attribute them, you know, various events and patterns to those structures we find. That'll be true for the brain.

Theo Jaffee (49:32)

How much of the age of Em do you think is applicable to, like, the near -term future of LLMs? When they're, like, kind of human -level, they're not, like, absurdly superhuman, they can run at faster speeds than the human brain, potentially?

Robin Hanson (49:45)

Well, the key thing is that Em's are full substitutes for human labor, and LLMs are not, and they're not close. So when you have a descendant of LLMs that is a full substitute for human labor, then the parallels will be much closer. At the moment, LLMs are really a niche market, and...

And people so far haven't... People basically keep making LLMs from scratch rather than making them from previous ones. So that'll be a key point perhaps in development if that ever switches. Well, humans, you make a human and they're a child and you slowly improve the human over time as it gets experience and matures.

Theo Jaffee (50:26)

What do you mean by that?

Robin Hanson (50:35)

with LLMs you keep every new generation of LLMs, you go back to the data and remake it from scratch so they don't remember or inherit from their previous versions. So, you know, many of the features of the age of Em are based on this assumption that you'll want to keep using the same Em's for many decades of subjective experience rather than just...

Theo Jaffee (50:46)

I see.

Robin Hanson (51:01)

stamping them in the lab and using them for an hour and throwing them away from the very beginning with no history or any sort of memory.

Theo Jaffee (51:09)

Hmm. So on the topic of AI and LLMs, how would you simply state the case against AI risk to people? Cause in your article, AI risk again, it relies on a lot of like somewhat.

Robin Hanson (51:22)

Wait, I don't have an article by that title, do I?

Theo Jaffee (51:26)

I think so. Let me look it up.

Robin Hanson (51:28)

but the title is AI Risk Again.

Theo Jaffee (51:31)

Yeah, AI risk, comma, again.

Robin Hanson (51:33)

That must be many years old then, right?

Theo Jaffee (51:36)

No, that was March 3rd, 2023.

Robin Hanson (51:39)

Okay, but presumably that's referring to some of my other more elaborate articles, but okay.

Theo Jaffee (51:45)

Yeah. But what about the opposite of elaborate? Like how would you state this case as simply as possible without economic arguments to someone who's like, for example, watch the Terminator or I, Robot so they're scared of AI and they see GPT -4 and they're like, oh wow, wow, this thing seems really impressive. Oh, we should be scared of the Terminator soon.

Robin Hanson (52:05)

So I think I have outlined two distinct lines of argument. The most recent line of argument that I focused on is asking people, and I did like a dozen AI risk conversations with people on YouTube that are recorded as you can see, basically asking people, well, ignoring AI, what did you expect to happen with your other kinds of descendants?

Did you expect to be able to control their values? Or did you expect to not have any conflicts with them? Did you expect to win all conflicts you might have with them? And almost everybody thinks that with respect to their squishy bio -human descendants, that those descendants would in fact become more powerful than them. They would win conflicts with them, and their values would be different from theirs.

And there might often actually be such conflicts. That's what everybody expects from their ordinary descendants. And it's what everybody has seen for many generations. And therefore, it's what they've accepted. They don't seem to mind that. But when it comes to AI descendants, they change their standards.

They are worried that we shouldn't have those kind of descendants unless we could make sure they never have a conflict with us or we'd always win the conflict with them or they could assure us that their values would never change. So people just holding different standards to the AI descendants than to the other descendants. And my main argument is that they really are your descendants and the same sort of evolutionary habit that should make you indulgent and supportive of all of your descendants, regardless of how they might differ.

from you should apply to your AI descendants. So that's one line of argument to say, you know, don't hold them to different standards in the abstract than you hold all your other descendants. Now, the faster change gets, the faster you may see your descendants.

and the fast the more of that history you may see sooner because change is faster but that would happen in the age of M2 because the Em's would also be changing faster and history would be happening faster so a slow human would see a lot more of that change as well but if you were okay with the Em's as your descendants quickly having different values from you being more powerful and winning conflicts with you then why not for the AI? So a separate line of argument is to say

AI is produced by capitalist firms who are doing it for profit.

And they will, of course, if their products hurt customers, that will be bad for them. They will therefore test their products regularly in many different ways. And in fact, computer products are among our most tested and monitored products of any sort because it's so much cheaper to test and monitor them. So AIs being built by large capitalist firms are for profit. They will make sure that their customers are not too unhappy with their experience with them.

And so they are very unlikely to be suddenly blindsided by their products vastly changing. They will be watching out for tendencies for their products to change and for that to cause bad experiences with customers. And...

for the intermediate time when these products are being produced by for -profit firms, you should expect to be as similarly happy or unhappy with them as you are now with most of the other products that you get from large capitalist firms.

Theo Jaffee (55:40)

Well, in response to that first line of argumentation that AIs are basically our descendants, could you not just say like, no AIs are not our descendants, they're like creation, so there's something different. Like, especially if it turns out LLMs are not like the golden path to AGI, and AGI will take something entirely different to be built, and that something won't simply be like the, you know, distilled total of human knowledge and human data and human preferences, that'll be like a...

maybe a Bayesian superintelligence built from scratch. Like, I could definitely see why someone wouldn't see that as our descendants.

Robin Hanson (56:17)

So natural selection is the powerful general theory that is much more general than simply the nature of biological creatures that pass things on through DNA. DNA -based evolution is one of many kinds of evolution, and we have general theories.

of evolution that apply not just to DNA -based biological organisms, but to culture and many other kinds of natural selection. The key concept of natural selection is variation in selection, as Donald Campbell famously argued in the 1960s.

And this is a general process. And in this general process, the key point is just there are some things, and they have descendants. Descendants is really just whatever literally descends from them, i .e. arises from them. And natural selection just requires that the descendants have some correlations and features to their ancestors.

pass down through whatever means and whatever those means is called genes. Genes are just our name for whatever is the mechanism of passing on features from ancestors to descendants. And as long as there's variation in those features and there is selection, i .e. not all features are equally productive of reproducing and surviving in the world, then you have natural selection. And by these general abstract theoretical conditions, AIs are our descendants, whether they're LLMs or Bayesian networks or

whatever else they are, as long as they share some features with us. And I'm pretty confident that if there are aliens out there making their own AIs, that our AIs will have things in common with us compared to the AIs of the aliens.

Theo Jaffee (58:05)

Well, still just playing devil's advocate here. Like someone could say, yeah, well, sure. You can make an argument where you define the word descendant to mean, you know, something that shares features with us that we had a part in creating either voluntarily or involuntarily. But like, I don't care about that. You know, that's it's, it's a computer. It's, it's not like my children.

Robin Hanson (58:26)

Well, you can care about whatever you want to care about, but I'm pointing to the standard concept of a theory, the word descended, I'm just making up a definition of word, I'm pointing to the concept that's in a theory, and this theory predicts, robustly, that...

creatures will have both a rivalry toward coexisting creatures with different genes, but also a support and indulgence toward their descendants, even if they have different genes, and that these AI are such descendants. You don't have to accept what natural selection tells you you should want, but then I don't know what basis you have other than saying, I like these descendants and I don't like those. I guess that's your right, but...

Again, why not dislike your other descendants just as much and be unhappy if they displace you and if they have win conflicts with you? That's also your right. You could just say you don't like any of your descendants.

Theo Jaffee (59:28)

Well, I think maybe people expect that their biological descendants will be less different. Like, many of the AI risk arguments take the form of, the AI will very quickly gain resources and kill us all, whereas our own biological children won't do that because they love their parents.

Robin Hanson (59:43)

Well, in the dozen conversations I had, almost everybody agreed that in fact their biological descendants would get pretty different pretty fast. As they are all. They might. That's a possibility. The reason why you don't think it will happen isn't because it couldn't happen.

Theo Jaffee (59:54)

but not fast enough to kill all them.

Is there any historical precedent?

Robin Hanson (1:00:06)

But there's no strong reason you think the AIs will do so easy. It's just the possibility with the AIs that scares people, not any particular reason I think it will happen, just that it could, but it could happen with your biological descendants. They... biological descendants have in fact killed parents and grandparents in the past. Yes, systematically, indeed. I mean, in many societies when people became too old to be productive, they were killed.

Theo Jaffee (1:00:23)

Yeah, but not systematically. It's usually very rare and they're like heavy norms against it.

Robin Hanson (1:00:35)

That's a common historical cultural practice.

Theo Jaffee (1:00:41)

So, the Age of Em is one of the most detailed deep dives I've ever seen into a future scenario. So, I think maybe the only other one I've seen is Nanosystems by Eric Drexler, but even that isn't really like a dive into the future economic and social implications. It's more so a dive into like the future possibilities of nanotechnology. So, are there any other deep dives like this that you know of? And...

Robin Hanson (1:01:03)

The technology, yes.

Theo Jaffee (1:01:11)

What science fiction do you think is the most realistic, the most detailed in ways that aren't silly?

Robin Hanson (1:01:18)

Well, I was inspired to write Age of Em in part by nanosystems. But as you say, nanosystems only looked at sort of the technology possibilities and not at the social implications. Drexter was interested in social implications, and he wrote about them in other places, but he just didn't have as much training in social science, I think, to do his thorough analysis of the social implications.

I wrote Age of Em and hope and part of inspiring other people to take up the example of doing such detailed analyses. So far they haven't.

the, I mean, for example, David Brin's Kiln People was a predecessor of Age of Em, where he did the best job I had seen in science fiction of trying to analyze similar issues, but still much less detailed of an analysis that I gave in Age of Em. I would, I mean, it might be true that I picked an easier problem than most are, in the sense that...

You could say more about Age of Em than you might be able to say about other scenarios, but I still think people could say a lot more about other scenarios than they have. And I guess I have to conclude that people are not actually that interested in working out such detailed implications because showing a demonstration of how it's possible hasn't inspired other people to do so.

Theo Jaffee (1:02:43)

Also in The Age of Em, you, like, the kind, the whole book is kind of overshadowed by the idea that, you know, this age might only last, like, one or two objective years, and then after that, like, something much stranger will happen, potentially. But, you know, there's not much detail about what that thing might be. So do you think, in, like, the very far future, is there anything meaningful that can be said about our descendants in, like, a million years?

Robin Hanson (1:03:11)

Well, yes, there's just fewer things that can be said. So one thing that can be said is that if physics is really a strong limitation, as it seems to be, then our descendants in a million years will still have to deal with only being less than a million light years away from here. And...

only having access to the volume and materials in that space. They'll have to deal with constraints of conservation of energy and second law of thermodynamics and, you know, other key constraints of, say, the speed of light of activity in that volume. Those are things I think we can say. I think we can talk somewhat robustly about

the two main paths of competition or coordination at the highest level. That is...

Either our descendants will not coordinate to sort of control their entire overall pattern of activity and therefore be in a world of competition with each other fundamentally at the highest levels, even though they can have smaller scale coordination in that competition, but the highest level might just be competition. And if that's true, we can make some claims about what they will want, I think, in that competition, how they will behave. The other option is that somehow,

our civilization manages to coordinate to enforce some rules that limits the competition. That will have to happen if it does before we leave out to explore the universe because if we have substantial colonists who head out into the universe before such coordination is created and enforced then it will be too late afterwards. So there's a limited time window to do that.

But it's in principle possible. So we can say either the world will manage to sort of send out political officers with every vessel that leaves here to enforce some central rules on behavior everywhere, or they won't, or the political officers will fail, in which case there will just be competition on the larger scale. And I think...

We can expect that competition will produce the kind of results it has in the past in terms of evolution toward a more efficient mechanisms and processes. And I think we could also guess that we can say some things about what the preferences of creatures who evolve are, because we have literatures on that today. We have literatures on what preferences involve in natural selection in particular in the context of investment funds and what preferences over investment funds people would have.

if they are natural selection produce their preferences.

Theo Jaffee (1:06:08)

kinds of preferences might they have.

Robin Hanson (1:06:10)

Well, so asexually produced creatures should not have a time discount, whereas sexually reproduced creatures have roughly a time discount of a factor of two per generation. That is roughly the time discount that humans seem to have. So if humans start competing with some asexually reproduced creatures, maybe AIs or Ms, then...

In trading with them, we will buy the present and they will buy the future because they care more about the future. And that's a robust prediction. I think that time preferences will go away.

Another prediction is about risk preferences. So for example, in the investment world, you get logarithmic risk with respect to correlated risk over all your copies, but you get risk neutrality about risk that's not correlated because you can insure against that by diversification. That looks like a robust argument that would probably continue to hold. So we can say something about the degree and kinds of risk aversion that descendants would have. And I think...

Theo Jaffee (1:07:16)

Can you explain the time preference thing a little bit more for the audience? Like what does it mean that sexual creatures discount to per generation?

Robin Hanson (1:07:22)

Right, so typically a sexually reproducing creature has a choice between investing in itself now or investing in its descendants. But its descendants share only half its genes and its descendants will mostly be able to use the resources when they reproduce a generation later.

So the really choice is between spending stuff on yourself now or spending stuff on your descendants in a generation. And so since your gene, your children have half your genes, then this is a choice of...

Doing things for your genes now full on, or doing stuff for half your genes in a generation, and therefore, this is a factor of two per generation discount rate. That is you're trading off doing stuff for yourself now and doing stuff for your kids in a generation. That's only for sexually reproducing creatures for which their kids only have half their genes. Asexual reproducing creatures, their kids have all their genes, so they should not be discounting the future for them at all.

Theo Jaffee (1:08:34)

So in a couple of critiques that were written about Age of Em by Brian Caplan and Scott Alexander, both of the main arguments are kind of that Em's will be selected more for docility. Like they won't evolve that much cultural drift from humans because like we will select against that as the main customers of the Em's and the people who are keeping their infrastructure alive in the physical world. So what do you think about that?

Robin Hanson (1:08:39)

beside the point.

Yes, but it's useful.

Well, so first of all, the classic story about humans is that we domesticated ourselves. So there are a number of features that domesticated species have that distinguish them from other species.

And then humans have these features. So when we domesticate horses and cows and pigs and, you know, dogs, etc. They differ in predictable ways because of that domestication. And humans differ in exactly those ways as well. So we have domesticated ourselves. So we are, in fact, more docile than other animals because we have become domesticated. This already happened. Long ago, really. Now...

For a long time, a fear about the future has always been that somehow changes will enslave us, or worse, make us enslaved and not even know or care. This is one of the most robust dystopian visions of the future that anybody ever has. Really. Overwhelming. Like at the beginning of the Industrial Revolution.

say around 1900 or before that, when people were trying to create dystopian visions of the future of industry, these were their main complaints. Exactly the complaints that humans would be enslaved or so domesticated we wouldn't even notice we were enslaved. So we clearly just have a very strong sensitivity to this possibility and it goes along with our strong egalitarian norms that humans are distinct from other primates exactly from having very strong egalitarian norms wherein we resisted anyone being the head of the tribe.

and putting themselves up there, and human foragers are famously egalitarian, famously strongly resistant to any individual humans dominating the rest. So this is just something we humans are primed to be afraid of and to be outraged by the possibility of, even though of course it already happened. We already are self -domesticated. So...

I don't think there's any particular reason to think that this is a thing that will happen in Age of Em more than any other future scenario, other than the fact people are just primed to be afraid of it because this is just a generic strong fear about any future scenario.

Theo Jaffee (1:11:24)

Hmm. So on the topic of rationalism and AI, how exactly did you meet and get involved and start blogging with Eliezer Yudkowsky? How did you get involved in the rationality community in the first place?

Robin Hanson (1:11:38)

Well, I had a blog called Overcoming Bias and I invited some others to participate and share the blog with me, including Eli Izer and Nick Bostrom and some other, Hal Finney, including who is a best guess founder of Bitcoin. And they did start blogging with me and we discussed AI risk on the blog because that was a big issue for Eli Izer then.

The blog was called Overcoming Bias, so it had a rationalist sort of theme right there, and we discussed some issues of rationality on the blog. And then a few years later, Eliezer decided he wanted to make his own blog called Less Wrong, based on some ideas that he'd have many participants contributing and have a karma system to rank them so that people could see what was quality. He did in fact make that blog. I helped him in the sense of allowing hard links from my blog to his blog in order to

raised the Google rank of his blog using the Google rank of mine at the time. And he

He created this community where people were talking about rationalist issues. They are less wrong. And then the karma system eventually seemed to have gone wrong, but still for a while people liked it and they were discussing things there. And so the rationalist community sort of started in less wrong there, but it grew to other places and all along, you know, Eliezer was using this community to push AI risk.

Theo Jaffee (1:13:09)

Interesting. So let's switch topics a little and talk about the elephant in the brain, which is, I think, my favorite thing that you've written. My friend who introduced me to your ideas, his favorite thing you've written. And for the audience, it's a book about how humans have hidden motives that we don't naturally reach for while explaining our actions. So.

Robin Hanson (1:13:22)

death rates.

Theo Jaffee (1:13:36)

One thing that's kind of a meme now in young people's circles is girls getting the ick. So I don't know if you've heard of this.

Robin Hanson (1:13:45)

the ick regarding a man for sick in particular.

Theo Jaffee (1:13:47)

Yeah, yeah. So it's like there are these, you know, TikToks, Instagram reels of like a guy doing something slightly awkward. Like, you know, it could be literally anything. The way he jumps into a pool, the way he, you know, reaches up to open a cabinet or something. And then, um, girls are like, Ooh, I just got the ick from that. So what do you, what do you think is the elephant in the brain explanation if there is one for that?

Robin Hanson (1:13:59)

Okay. Right.



Well, the usual explanation, I would think, is pretty close to the surface, is that women are very selective, or they see themselves as very selective among men. They are willing to mate with any woman, man who asks. It's very important to them to be selective, and so they are trying to be very selective.

And so they lean into intuitive reactions that are selection reactions, in particular rejections, in ways that the rest of society finds impolite or rude in other contexts. So we are again relatively egalitarian, and so we try not to overtly reject each other, or we make excuses for it. We certainly don't like to go out of our way to insult people and to put them down. That seems rude and...

arrogant, and that's true in most of the rest of society except apparently we make an exception for women rejecting men. Apparently not only is it okay for women to eagerly reject men, but they can insult them in the process and put them down and bond together over their...

you know, rejection of men and basically declaring men are unequal and men are should be unequal, the lower half of the men distribution is morally unworthy and ick basically and maybe should not exist. That's apparently a kind of inequality attitude that's okay in our world although we have many other aversions to inequality talk. We could discuss that more.

Theo Jaffee (1:15:53)

Do you think this, do you think that women's propensity to reject men is adaptive in the sense that, you know, they will select better mates or maladaptive in the sense that there will be like fewer overall children?

Robin Hanson (1:16:00)


Well, obviously some degree of holding standards is adaptive. And of course, there's going to be the additional...

value of signaling that you have standards. So people might be hold too high of standards exactly to show off that they have standards. But if people were still settling down and picking someone soon, we wouldn't have fertility problems. Those come from people spending decades being picky. When they finally pick somebody, it's kind of too late. That's more the problem. It's the delay in picking and not so much the high standards.

And then that interacts to some degree, I guess, is that people hold very high standards. And they say, nobody around me now could possibly meet my standards. Somebody maybe later will.

Theo Jaffee (1:16:45)


So you also talk about how a lot of the explanations for these behaviors come from, like, forager times. Do you think humans have evolved genetically at all in the last, like, 10 ,000 years, and if so, how?

Robin Hanson (1:17:11)

Well, there's certainly data about, say, milk processing.

You know, so some people can't process milk and others can and that certainly seems to have spread in the last 10 ,000 years. So we have a few pieces of evidence about that. We also have the general data that we expect rates of DNA evolution to be proportional to the size of the population. So the prediction is as the population is getting larger, selection is beginning faster, although it's been happening over a shorter time period. So, you know, you have to take that in. But presumably selection is much faster lately.

than it was before because of the larger population, and we see some specific kinds of selection. But honestly, the usual story, which seems right to me, is that there's been relatively little DNA evolution, but a lot of cultural evolution. Cultural evolution is overwhelmingly where human evolution has been for a while.

a lot of cultural evolution, but then the challenge is to think more carefully about what exactly that is and how it works. And I think I didn't think very much about it until a few months ago, and so I realized that most people kind of think they understand culture, but they haven't really thought much about it, and there's a lot more that they should learn.

Theo Jaffee (1:18:28)

Like what? What would be the most important things to learn about this?

Robin Hanson (1:18:32)

Well, first is that it's this autonomous process that culture tells you what you should value and what you should do, and you just accept that. And then it changes and you just accept that. And there's nobody driving this train. Key features of this culture are what counts as status, what counts as prestige, because you copy the behavior of high prestigious people.

And so the definition of prestige can really make a big difference to what direction culture goes. If prestigious people are the people who have the most years of education, then that'll encourage people to get a lot of years of education, maybe even more than are useful for other reasons. If prestige goes with individual wealth, then people will focus on accumulating individual wealth and passing on wealth to a smaller number of kids so that they can be individually wealthier.

a lot depends on what we decide counts as prestige. And nobody is driving that train. Over time, what counts as prestige has changed. And we didn't vote on it, and we didn't analyze it and decide it together, and there wasn't a process that was just anticipating the consequences of this and figuring out what was best for us. Culture is a very crude process. Like, if you think about your organs and your body and your...

Body reactions, you change in time of day and time of year. You have all these complicated ways in which your body is primed to change its behavior in different contexts, because your bodies like yours have been evolving for many millions, even billions of years, but culture has only been evolving for a few thousands of years, and it doesn't have all those complicated conditional processes to, you know, adjust for context. It's a much cruder thing.

You should not think it's this very subtle, well -worked -out, systematic thing that will carefully adapt to all sorts of details.

Theo Jaffee (1:20:30)

So in terms of the like relative decline in wokeness and similar, you know, if you can call it wokeness, call it progressivism, leftism, whatever, specifically in tech circles in the last couple years as a result of Elon Musk buying Twitter, is that like a genuine cultural change or is that simply people following the behavior of a prestigious individual?

Robin Hanson (1:20:51)

Well, that's what cultural change is. There isn't something else. There isn't a whole other thing. That's what it is. It's whatever the prestigious people are doing is culture. There is no other source.

Theo Jaffee (1:20:53)

Oh, yeah, that makes sense.

Well, which way does a causation go? Does, you know, what's prestigious?

Robin Hanson (1:21:08)

both. That is, if you aren't doing what culture says, you look less prestigious. And whatever the prestigious people are doing, that's what culture is.

Theo Jaffee (1:21:21)

So then how does culture change over time? If it's just, you know, if culture and prestige.

Robin Hanson (1:21:27)

Well, the biggest event in the 20th century that influenced culture was World War II. And both the rise of Hitler in the first place and his fall and losing the war were both pretty unpredictable. But they still had enormous consequences for culture. So you can see how things that were at the center of remaking world culture were the result of conflicts.

that it was hard to predict who would win. So that's also been true for the major changes in culture over the last half century in our world. Ex ante, they were hard to predict. It was hard to know who was going to win. Later on, we tell ourselves the story that the winner was inevitable. And we should have known all along that they were going to be the winner and we should have accepted. But you couldn't really tell early on who was going to win.

Theo Jaffee (1:22:22)

Do you think World War II was a bigger cultural change than, like, the fall of monarchy at the end of World War I globally, or, like, the fall of communism globally in the late 80s, early 90s?

Robin Hanson (1:22:33)

I don't know, I mean they were of a similar magnitude, so I don't that much care to say which one is exactly the bigger. They were big.

they were not that predictable.

Theo Jaffee (1:22:49)

So another elephant in the brain topic is medicine. This is maybe the most famous thing that's come out of it. So why is health care spending both so high and life expectancy so low in the US relative to other countries? Like, do we have a worse signaling problem or is it bad institutions?

Robin Hanson (1:23:05)

Right. So just to discuss to the audience, the elephant in the brain is mostly about why is medicine weird compared to what I think it's not really focused on the US versus other places. So most of the book, we're just looking at average typical human behavior and trying to understand that. I'm not very interested in the variations across space and time. If you don't even understand the average, you have no business trying to figure out the variations because they're just harder to figure out.

So the average medicine is basically not very useful. And so we try to explain our fascination with and obsession with medicine and the fact that it seems to do very little on the margin by saying that medicine is something we use to show that we care. And so we use medicine to...

show people that even though they're sick, we might betray them, but we won't leave them. We are going to stay with them and take care of them, and that's very reassuring, and that's the main function of medicine. But that doesn't explain why different times and places might do things different. So if we want to say why is medicine different in the US in the 20th century, say, I think we have to go to

It's one of the key stories the US tells itself about why the world should be grateful to it. So the world has a few stories about why the US has saved the world in many cases and they should love us. So World War I, World Wars are an example of that. And then the Cold War are also examples of how we say we saved the world from very dire problems and they should all be grateful to us. And then medicine is another one because we say basically we gave modern medicine to the world. We point to key adva -

in medicine happening in the US and the world copying them and being all the better for it. So when we have a thing that we are proud of as being the source of and spreading it to the world, then we tend to double down on it. So we double down on military spending even though we have very few, you know.

neighbors nearby who might cause us any problems. We still spend enormous amounts on the military, reaffirming our story of how we saved the world from the Nazis and the communists. We also like to tell the world that somehow we gave them civil rights and legal sort of legal procedure and we'd like to double down on that because that's our story of how that came from us because we went wild with that.

soon after World War II, and we also say that we gave the world medicine, and so we continue to spend a lot of it in part reaffirming how wonderful it is, which reaffirms how wonderful we are for having given it to the world.

Theo Jaffee (1:25:54)

Do you think that academia right now in 2024 is the best institution that we have for aggregating information and seeking the truth? Like, do you think there are existing other institutions like, you know, the blogosphere that might be better? Is academia still a P?

Robin Hanson (1:26:08)


I mean, certainly not the only institution we use. So for different kinds of information, we use just a different institution to aggregate information. So clearly, for example, just ordinary business practice is one institution for aggregating information. Most business practice isn't mediated by academia. It's mediated by prior nearby business practice. So business people are looking around to see what other people are doing and copying them, trying them out. And we aggregate information about what businesses should do through the competitive market.

with businesses looking over their shoulders, what everybody else is doing. That's the way we aggregate information about business. The way we aggregate information about say marriage or families isn't much to do with academia either. We in a world where other people around us are having marriage practices, family practices, and we copy what our parents did, what people around us do, et cetera. That's how we're aggregating information about those topics.

Theo Jaffee (1:27:03)

What about physics? You know, that's mostly like academic, theoretical, and experimental.

Robin Hanson (1:27:05)

Well, most practical physics isn't being aggregated by academic physicists either. So everybody should know that like a lot of famous physics inventions were first happened practically and then theorists came up with explanations, say thermodynamics, laws of entropy, et cetera. They were first practiced. And so an awful lot of physics practices first happens in industry and then people doing things and then...

academic physicists try to make sense of it and synthesize it, you know, systematize it. But theoretical physics is more accepted from academia, in part because nobody cares and it doesn't matter.

So we mostly let academia aggregate information about abstract stuff that nobody else cares about. When the rest of us care, we are much less willing to listen to academia. And so often, academia then just tells us whatever we want to hear. So there's an old saying that a leader is someone who figures out which way the crowd is going and gets out in front.

Academics often do that about, you know, practical ways to live. People just have a lot of ways they think we should live and then academics typically find a way to support that. And this is also true in government policy. Mostly the government doesn't listen to academic policy. Mostly the government decides on policy in some other way and then finds academics who support whatever they're saying to justify it and they find them and then those people look influential, but they're much actually much less the fact that somebody was

there to be found to support it for any of you it matters. If they couldn't find anybody to support it they would have been less likely to do it, but as long as they can find some support that's enough.

Theo Jaffee (1:28:52)

So do you think academia has always kind of been bad like this or has it gotten worse over time? Did it used to be much better?

Robin Hanson (1:29:00)

I don't know, but certainly peer review is something that people today think of as essential to academia, but a century ago it just wasn't a thing so much. Like in 1900 there was hardly any peer review, that wasn't how things happened. You know, journal editors just had a lot of discretion and decided what they liked.

And so for most of the famous history of science up until the 20th century, that wasn't peer review. So I think people are not that sure what academic institutions are exactly, because they've changed a lot over the years. And certainly often you just had a community of people and a small enough community of elites that they could just manage each other informally. And sometimes that works very well. And as the community gets larger, it fragments and it just can't manage itself that way. And so,

today is just really much larger than it ever was and so there is no small community of people who run the whole thing. It's very decentralized.

Theo Jaffee (1:30:00)

Do you think peer review actually matters or is it another kind of like elephant in the brain signaling thing that we do kind of to show that we care about truth?

Robin Hanson (1:30:12)

I've spent a lot of time thinking about alternative academic institutions and about the institutions we're in, and so I'm confident in saying that the institutions we are in are not remotely optimal for the purpose of producing intellectual progress. They mostly people doing things to win their local games, but...

Theo Jaffee (1:30:28)


Robin Hanson (1:30:35)

Still, there are many kinds of abstract topics where the best thought about them will be found in academia, nevertheless, even if this isn't optimized for that. I do think we know much better ways we could organize academia in terms of promoting progress, but there isn't much of a constituency for that, so it's not going to happen anytime soon.

I have my own particular proposals for what you would try to do to make things different, but again, the limit is making anybody care. There's very little interest in improving academia, even among academics. Mostly people want to win the academic game, and that's what they focus on doing.

Theo Jaffee (1:31:13)

So you've written very extensively about alternative institutions, but what do you think some examples are of institutions that are actually good?

Robin Hanson (1:31:21)

existing ones? Well, in some sense, existing ones are all beating out the immediate alternatives that they could be easily displaced by, so the existence of institutions tells you something about their staying power and that they are, you know, filling a niche and continuing to fill the niche. So you have to give them all credit for that.

Theo Jaffee (1:31:22)

Existing ones, yes.

Robin Hanson (1:31:43)

You know, they are all sitting next to a space of alternatives that people often do try to replace them with. And the existing ones are keeping them at bay. So they got to get credit for that. All of them. Now some of them are keeping others at bay by, say, locking down the control of the national government, say, and making sure that voters never think to reject them or something. Others have a wider range of competition they're pushing away. But...

I mean, academia is relatively decentralized, and so academia does succeed so far in pushing away attempts of other groups of people, say bloggers, to gain their position of respect. They have so far won against those challengers. Decentralized.

Theo Jaffee (1:32:28)

Did you say centralized or decentralized?

Yeah, it is, you know, I... It still is crazy to me how powerful human culture is, where, you know, a thousand different universities can have basically the exact same culture, you know, scattered all over the country and the world. Like, one of the things I noticed when I go to University of Florida in Gainesville, like a year ago, I took my cousin to visit Arizona State University on the other side of the country. There's no affiliation at all between UF and ASU.

And I was like, Oh my God, this is like the same thing. It's got the same kinds of buildings. It's got the same companies that are contracted to it. It's got, you know, the same kinds of tour groups, the same kinds of professors and classes and courses.

Robin Hanson (1:33:07)


right? Right, so academia has been a unified culture for quite a while now worldwide and so there's a sense in which it isn't allowing much diversity of the sort that you know the center doesn't like. So in general if we had a bunch of different kinds of academia in different places doing it different we might have more competition among them.

see which ways went out, but when there's centers of academia enough to sort of enforce standards on everyone, then that's how it works. So like in most disciplines, there's a small number of people who are at the top of the discipline, and they control the major funders and the major journals and the major jobs, and their shared opinion about what that discipline should be doing dominates worldwide.

So the different disciplines can compete with each other to some extent, but they mostly stay off each other's territory. Well, the Center is the most prestigious people.

Theo Jaffee (1:34:19)

Well, what is the center of academia, if academia is so decentralized?

Robin Hanson (1:34:25)

and their opinions about what, again, who should get what jobs, who should get what grants, who should get what publication journal slots. They decide those things. And those things, you know, basically the most prestigious people decide the next most prestigious people, et cetera, all the way down the ladder of prestige. So when prestige is very important, the top prestigious people basically have a central power even without any other official, you know, roles of power.

All they really need to do is declare some things prestigious and other things not. And that works all the way down.

So I mean, quite commonly, most disciplines, there's a set of relatively established methods and.

claims in the discipline and then the only people really allowed to challenge them are the most prestigious people. So if somebody much lower prestige challenges those things they're routinely slapped down and rejected because it's not their place to do such things. Lower level people are supposed to question smaller things, do supporting work to the most prestigious people. The most prestigious might be allowed to change the fundamental assumptions about methods or conclusions in the field, but low status people that's just not their place.

Theo Jaffee (1:35:44)

Well what about kind of outsiders academia who have tried to change it who are pretty prestigious like Peter Thiel and more recently Bill Acme -

Robin Hanson (1:35:53)

I know of no academic field where Peter Thiel's opinions carry much weight.

Theo Jaffee (1:35:59)

Yeah, but why? I mean, he went to, you know, the most prestigious university, Stanford, one of them, and...

Robin Hanson (1:36:04)

prestige is particular, so being a prestigious sociologist doesn't carry much weight in economics or physics, right? Yeah, but that's a different prestige letter, and so it counts for other kinds of things, right? So we trust the most prestigious doctors to decide who doctors can be, but you know, we don't trust the president to decide that, even if the president's very prestigious, right? So we just have...

Theo Jaffee (1:36:12)

and he's a billionaire.

Robin Hanson (1:36:33)

ideas about what kind there are different kinds of prestige and what scope they have and you have to use prestige within your area.

Theo Jaffee (1:36:43)

And I think Peter Thiel has had somewhat of an impact in like, if not coursework, the culture of like CS programs at schools, especially elite schools. I think the startup vibes of MIT and Stanford and probably most universities are different from what they used to be because of in part Peter Thiel's ideas.

Robin Hanson (1:37:08)

I'm, you know, if that's true, I'm happy. I mean, I'm not in those worlds, so I don't know those, but yes, like the startup world is another world and academia is a different world, but they overlap somewhat and they come somewhat compete for outside prestige. And so often they are using each other for their various purposes. I certainly know that startup companies often ally with academics in order to add prestige to their startup. Presumably vice versa. Academics ally with...

Theo Jaffee (1:37:32)


Robin Hanson (1:37:35)

startup people to add prestige to their academic things.

Theo Jaffee (1:37:40)

So pretty recently you wrote an article that I believe is called Why Crypto? Where you, yeah, you talk about your positions on Bitcoin and the cryptocurrency industry, which by the way, for the audience, Robin was like remarkably early on this. He was friends with Hal Finney, I believe before Bitcoin was actually a thing.

Robin Hanson (1:37:45)


Oh yeah, long ago. But I wasn't really into the crypto thing then, but... And I, you know...

Theo Jaffee (1:38:03)


To not misrepresent your position on this, you believe that Bitcoin is mostly speculative?

Robin Hanson (1:38:16)

Most of the value of crypto was realized in Bitcoin nearly early on in the process and since on most of the activity to create other coins and other things you could use the coins for hasn't really panned out much. They could but it hasn't so far. But...

Theo Jaffee (1:38:35)

Well, what do you mean by most of the value is realized early on?

Robin Hanson (1:38:39)

Well, that is having a crypto coin, a Bitcoin, having it as a store value, having it as a thing you can use to make some trades with, that was a new thing and that was added. That was a value added by the Bitcoin early on. Since then, people have made lots of other coins to do lots of other things, but they haven't actually achieved much value from those other things.

there is the value of just having the coin and being able to use it as a store value or as a money to trade. And that's a value that's continued from the beginning once people had Bitcoin.

Theo Jaffee (1:39:02)


So the reason that the price of Bitcoin has gone up from, you know, one cent per Bitcoin to $70 ,000 per Bitcoin, how much of that is just pure speculation then?

Robin Hanson (1:39:25)

Most of it. Most all of it, yes. But.

Theo Jaffee (1:39:29)

How is it possible that if it's just pure speculation that the market can remain rational about that for so long?

Robin Hanson (1:39:35)

Well, that's not necessarily irrational. That is, there is no particular rational value for Bitcoin. It can be all sorts of different things.

When worlds of speculation are created, they just have internal dynamics that can take them all sorts of different directions, just like with culture. Culture can just go a lot of different directions, and so can the price of Bitcoin go a lot of different directions. Clearly, in some sense, ex -SANI people should have been surprised to see it go so high so fast, or they would have driven up the price initially. So, you know, this had to have been a surprise.

but a lot of directions of markets and culture are surprises. But the point of that post is to say all of the speculation had an interesting effect, which is some people made a lot of money in the speculation. And then as commonly happens in the world, when somebody makes a lot of money, they need a story to tell themselves about why they were justified in making that money and what...

you know, what would justify their use of it? People mostly feel a little embarrassed to have a lot of money and they need a story to legitimize. They're having it and using it. And different kinds of ways to make money produce different stories like that and so they produce different kinds of rich people who live their lives differently. And so this kind of way to make money produced a different kind of rich person.

Theo Jaffee (1:41:03)


Robin Hanson (1:41:12)

And that's interesting because they do more interesting things with the money.

Theo Jaffee (1:41:19)

That kind of reminds me of Warren Buffett's essay, The Super Investors of Graham and Doddsville, where he outlines a scenario where like everyone in the US flips a coin every day and the people who get heads transfer, you know, some amount of money to the people who get tails or whatever. And that process continues over and over again until only one person is left with all the money. And so, like, by the nth day, there would only be like, uh,

Robin Hanson (1:41:41)


Theo Jaffee (1:41:48)

10 ,000 people left and they would each have like a million dollars or something I forgot the exact numbers But it's like and then they would go on to write books about how I became a millionaire in Just seven days with only 30 seconds of work a day You

Robin Hanson (1:41:51)



Right now, different worlds have different amounts of financial speculation and different amounts of the inequality produced by this process. So, you know, ordinary business investment produces inequality this way, but at the time scales at which businesses rise and fall. And so...

People, you know, wealth becomes more unequal as a natural result of businesses winning and losing, but it's slow because businesses win and lose slowly. When people more directly do financial speculation, then that process can happen faster, but most people don't do that much financial speculation.

so that limits it, but there are times when it becomes in fashion and lots of people do financial speculation and then more inequality is produced in those short periods, but so far those have been rare periods. Crypto has just been one of those rare periods with a rare subset of people who then went wild speculating and therefore produced enormous amounts of inequality in a short time.

we should expect that sort of thing to continue over and over again through history, and each person should perhaps wonder how eager they should be to speculate, because, you know, they will be participating in a process that produces inequality, and to what extent do they want to be part of that? Now, sometimes these processes produce other things the world values besides the inequality, like business competition produces industry and a...

growing economy that we all benefit from, and so it seems like we should all tolerate a substantial amount of inequality produced by business competition because that's what makes our world rich. Other times there are somewhat separated worlds of speculation like crypto, which looks like we could counterfactually imagine if that all went away. The rest of us wouldn't mind so much. We might wonder, should we allow those little worlds to happen?

And that's the kind of question people have about that and about, say, stock market speculation, say, the 19 up to the 1929 crash or the 2008 crash, right? There's a burst of speculation that get bigger before those crashes. And people are often critical about whether that thing should have been allowed or whether some people are exploiting others in that process. And that's all perfectly reasonable to discuss. But I just thought it's interesting to notice that.

In crypto, enormous amounts of inequality created, the richest of them got a lot, others lost a lot, and the winners needed to tell themselves a story about why they deserve their money, and that story had to fit with how they use the money. And so, in crypto, the story they tell themselves is that they were pursuing big -idea innovations about how the world could be substantially different, if only certain crypto.

coins and processes were realized. And then when they're rich, they pursue that vision by spending some of their money on various visions that could change the world a lot. And that's a positive outcome in my mind of crypto because there are in fact a lot of opportunities for ways to invest to make the world much better.

Theo Jaffee (1:45:15)

Why was crypto speculated in so heavily and not any other asset class? Is it just kind of pure unpredictable randomness?

Robin Hanson (1:45:23)

Well, it was, I mean, it is literally speculation that is crypto is electronic money. Electronic money is literally a thing you can own and that its value can go up and down. So it is, you don't have to do any indirect thing to take a thing that's happening and make something else, something you've speculated about it. Most of the rest of the world, you have to work to make that connection happen, right?

If you want to speculate about, say, pickleball, you can go play pickleball, or you can buy some pickleball rackets or quarts, but if you want to speculate on it, you'll need some business that's invested in pickleball, and there aren't that many of them, and you have to figure out what to do. But crypto, all the things were things you could buy into, and there were thousands of them, and each of them could go up or down, so... And there was this story that...

It would be huge, of course, more plausible than pickleball. You know, it's just hard to imagine how big pickleball could ever get. So it's hard really to bet on pickleball. Imagine it'll go up by a factor of 100 in value. But with these coins, you had more of a story of how they could be huge in the future. You know, the entire finance industry could be displaced by the crypto world. And so that, you know, fuels speculation more.

Theo Jaffee (1:46:44)

So you've also written about a form of governance called demarchy, which is where you have a network of decision -making groups where the membership of each body is randomly selected from those who volunteer to be on it. But wouldn't that prevent natural elites or good leaders from being able to exert power if it's just randomly selected from all those who volunteer?

Robin Hanson (1:47:05)

Sure, I'm not a big fan necessarily of demarchy, but I am just a fan of people collecting big ideas for how things could be different. I mean, I just want people to know about them and to think about them because we just don't think enough about how we could change things.

Theo Jaffee (1:47:21)

Do you think the state of homeowners associations in America is kind of counter evidence to demarchy? Cause you know, it's pretty similar. I would say you have a decision -making group that's given a surprising amount of power over neighborhoods and the membership of HOA is, is kind of elected, but usually it's not super competitive. So it's mainly people who volunteer to be on it and they seem to be very dysfunctional.

Robin Hanson (1:47:42)

Right? Right?

again, but they're beating out competitors for that slot. So, you know, the main thing is that ordinary homeowners feel a little awkward about other sorts of institutions you might put in that slot again, and so they're very egalitarian, democratic instincts are pushing them to favor that institutional form there, even if we might look and think it's kind of inefficient. So...

Again, when something exists and there's alternatives to it that it keeps pushing away, you gotta give it some credit and wonder, how's that working? How does it do that?

Theo Jaffee (1:48:22)

What if it's just a coordination problem? Like what if everyone knew that they would be better off if they could say team up to abolish the HUA but...

Robin Hanson (1:48:28)

I mean, any one homeowners association in the country could decide to change how it's incorporated and make new rules and do it different. It's not that hard.

And new, I mean new, new homeowners, new sets of homes could just follow different rules. So clearly, people aren't very inclined to make new sets of homes near each other with different sets of rules. Something must be pushing them to do all these same old rules.

Theo Jaffee (1:49:02)

just the same kind of convergent culture.

Robin Hanson (1:49:06)

Right, for some reason, they must have done focus groups or something. Homeowners, when they hear about these alternative rules, they're not that eager for them. Maybe they sound suspicious. Maybe they sound unhomy. I don't know, but that would be interesting. I'd like to know what happens when you ask homeowners, hey, how about we have a different set of rules for doing your homeowners association? What do they say?

Theo Jaffee (1:49:29)

Often they don't care.

Robin Hanson (1:49:32)

Well then, if for example, the developer thought that the home, somehow it would be run better with a different set of rules.

why don't they make a different set of rules? Because, presumably, there's at least a weak effect of if it's run better, then reputation comes back to people about that brand of home. I want to buy that brand of home because I know people living over there in that brand of home and that seems to go pretty well there. So you would think they'd have some reputation incentives to produce homeowners associations that are well run.

Theo Jaffee (1:50:03)

And then, of course, the other very famous system of government that you've written about is called futarchy, where, yeah, invented. Where, for the audience, it's where a policy would become a law when a prediction market would clearly estimate that they would increase national welfare and that national welfare measure would be defined by elected representatives. And of course, you were also very early on the idea of prediction markets. I...

Robin Hanson (1:50:11)

That's my invention.

Theo Jaffee (1:50:33)

believe you were one of the inventors.

Robin Hanson (1:50:35)

Well, it's an ancient idea, so not really something anyone can invent, but I was one of the first advocates for using prediction markets much more widely than they are used today. That is, the same mechanism can be used for other purposes. Previously, the mechanism was used for people to enjoy betting, and then the customer was the better. I was...

pointing people toward the opportunity for someone who wants the answer to a question to subsidize a betting market to get the answer. And that's a possibility that still hasn't been realized as widely as it could.

Theo Jaffee (1:51:10)

Hmm. So, how would the measure of natural welfare be immune from Goodheart's Law? Which is, for the audience, when a measure becomes a target, it seems to be a good measure because people will try to game it.

Robin Hanson (1:51:23)

I mean, clearly, Goodhart's Law can't be true that way, because our world is full of measures. There's measures all over the place that we're using all the time, and still we want to keep using them. So it's just not true that measures lose all their value when you use them. Some do in some contexts, so you might want to understand what are those scenarios, but most don't. So, for example, we don't want to die.

Theo Jaffee (1:51:38)

Not all, but some.

Robin Hanson (1:51:49)

So, lifespans are a measure of health, and we would like processes in places, etc., that promote longer lifespans. And our using lifespans as a measure hasn't destroyed the value of lifespans as a measure. It's still a pretty good measure, even though many institutional incentives are tied to it, or even wealth. People want to get rich, and the fact that people want to get rich doesn't mean that...

It's not valuable to see how rich you are.

Theo Jaffee (1:52:20)

Clearly it's not like as good of a measure as it could be because of, you know, the whole concept of quality adjusted life years. It might be better to live 70 very long, very healthy years where you die quickly at the end than live to be 80, but the last 20 years of that are just like miserable, slow decline.

Robin Hanson (1:52:32)


Sure. And in fact, we often do use quality adjusted life years as a measure. That is, in fact, more common as a measure among the people who have such measures. And that hasn't destroyed the value of such measures. I mean, there was actually...

Interesting or say the US Congress I think at some point passed a law that said you shouldn't be using quality adjusted life years as a measure because if you just use life years then elderly people come out looking better or even just death rates if you ignore how old somebody is then you're ignoring how many years they have left and people who want to push policies to help old people prefer that measure they just say let's just look at what affects mortality ignoring life years and or quality adjusted

life use as well. So there's often politics associated with which metrics get used, but that doesn't mean metrics can't be used. There's a nice book called How to Measure Anything. I recommend it. It says measuring things is, you know, a difficult thing that you can work at and get better at, and so you can do it for most anything if you work hard enough at it, and I think he's right. Measuring is work, and it returns gains to effort.

Theo Jaffee (1:53:56)

What do you think about applying prediction markets to dating apps? Like the state of dating apps right now is really bad. Really bad.

Robin Hanson (1:54:05)

Well, except they are beating the competition. So, okay. I mean, the question is, do people... So, for example, I actually think the old institution of a matchmaker was an effective institution. Mastermakers did actually learn about other clients and what they might like in putting them together. And matchmaking would be an effective institution today. And people just don't like the idea.

Theo Jaffee (1:54:08)

But they're reading the competition, but maybe that's just because they haven't -

Robin Hanson (1:54:30)

So it's more people's personal aversion to the very idea of a matchmaker. That's the reason why I'm using matchmakers, not that they couldn't make good matches. So a lot of this has to do with people's sort of ideology of dating and what's supposed to go into it. I mean, for example, parents being more involved in matchmaking made a lot of sense and in fact seems to have gone better for people. Parents know a lot about their kids and they can often make contacts with other parents in ways that people, kids can't do themselves.

parents helping with matchmaking was a big advantage. And we've rejected that too, not because it didn't work, but because we just don't like the idea. So I think prediction markets could probably actually do better also, but they would face the similar objections to parents and matchmakers, which is we don't like that. So the question is, can you do prediction markets in a way that people won't react that way to?

Theo Jaffee (1:55:25)

Hmm. Yeah, like there's a lot of debate. I... You're probably pretty familiar with manifold markets and manifold love. Yeah, you went to last year's manifest, right? I'll probably be at this year's one, by the way. Awesome. So there's a lot of like internal manifold debate about manifold love. Manifold love is, you know, a prediction market applied to the idea of dating apps where you have people bet on...

Robin Hanson (1:55:35)



Okay, well then I'll see you there.


Theo Jaffee (1:55:54)

like which couples will work. And some people at manifold think it's great idea with lots of potential and others think no, it'll never work.

Robin Hanson (1:55:55)


Theo Jaffee (1:56:04)

So do you think maybe both at the same time that it is a great idea with a lot of potential, but it won't work because people won't want it to?

Robin Hanson (1:56:12)

Prediction markets in general are, they're a general technology with a enormously wide range of potential application. So I don't think it's worth having strong opinions about which things will work. I think it's better to have opinions about what are the good things to try first. So it's more important to have good heuristics about where it's cheap to try things and where there's big value to be gained if you do try.

So I would think, you know, dating, there's certainly huge value out there to be gained, but...

There's not so much value to be arbitraged in the sense that it's hard if you see two people who should be together to like gain the profit from convincing them to go together. There are many other contexts in the world where when things aren't efficient, there's ways to make money from that. And I think we should focus first on those kinds of applications because that will just attract a lot more energy and attention to trying to make that extra money. So.

So it's of substantial value, but it faces some cultural obstacles and it's hard to just spread on the basis of its efficiency because, again, there's not profits to be made. So I would try to focus people's attention more on places where if you adopt something and it works better, you can make money because stuff spreads faster there.

Theo Jaffee (1:57:38)

Hmm. That makes sense. So you have like a very unique outlook in the sense that it's so broad and that you've drawn from so many different subjects. So who else do you think are some of the broadest thinkers in the world?

Robin Hanson (1:57:56)

I would have to go research to figure that out, I guess. I studied history of science long ago, and one of the main things I learned is that when scientists have stories about their history, it's usually wrong.

when historians of science go study what was actually the history of an area of science, they just get the different answers than the scientists I'm telling each other. So, you know, from that I learned people are way too quick to make these judgments about what's going on. And so if I look at a question like this and I go, well, that would be a fair bit of work to figure out. I don't know. I would have to figure out who is actually being strong polymath learning lots of different areas and adding to them. I certainly have known some people who have contributed to multiple areas and

I, you know, impart by knowing multiple areas and making connections between them. So I respect that, but I don't know who's at the peak of doing it a lot. But if you had some people to read about and talk to, that would be interesting to maybe learn about which of them gained how much how from their different kinds of knowing many areas.

Theo Jaffee (1:59:02)

What about in the past? Not necessarily alive today.

Robin Hanson (1:59:08)

Again, I know of some people that I happen to come across because they learn multiple things, but I don't know which of them is, you know, big suffocating. Like the name Herbert Simon comes to mind, for example. He's, uh, I guess he won the Nobel Prize.

from his work combining computer science and economics and other areas of social science and systems design. That was a pretty broad area of things to combined. James March, similarly, I don't think he got the Nobel Prize, but he was somebody who did organizational innovation things while learning other kinds of social science.

people who have done both evolutionary psychology applied to other areas of life have impressed me at times because they do these crossover things. Most recently I've just been seeing people who do cultural evolution and apply it to other things outside of anthropology. That's interesting.

Theo Jaffee (2:00:11)

A few days ago I tweeted, um, what book that you've read has the... I forget the exact wording, but it was like, what book that you've read has the highest density of like, huh wow, I never thought about it like that, moments per page. Do I need to bring to mind for you?

Robin Hanson (2:00:26)

I mean, again, these are idiosyncratic. A lot depends on what you'd read before you read any one book.

Often the first book in an area you read is very insightful because, you know, you could have read some other book and got the same insights from the other books. So it's really hard to say, you know, generalize from your experience to other people's. So I try to, again, look, I speak on enough different topics that I think it's fine when my intuitions say, I don't really know that much about that one, to say, okay, I'm not going to have an answer because I don't really know.

Theo Jaffee (2:00:59)

And since you do speak about so many different topics, how do you kind of balance them? Do you have periods where you are only really thinking about one and really diving into it? Or are these different ideas always closer to the top of your mind when you're -

Robin Hanson (2:01:18)

Well, I mean, one of the fun things about having many different topics is that at any one moment you can be pulled to, you know, so it's fun to have a number of different topics you're thinking about at any one time so that depending on your mood, you can go to one or the other. And it's always fun to play hooky on one to do the others, honestly. Like if you have only one thing you're working on and if you don't feel like it today, you feel like, well, I got to make myself do it today because it's the thing I'm working on and it's hard to get motivated that way. It's much easier to get motivated to say, I should be working on this, but I'm going to do this one. That's what I'm going to do.

Theo Jaffee (2:01:37)

What do you mean?


Robin Hanson (2:01:48)

That's fun. And you can then switch it back. The other day you can say, oh, I should be doing this. Why don't I do this one? Because this one's fun. And you get motivation that way.

Theo Jaffee (2:01:51)

Yeah, that happens to me a lot too.

And also, you teach law and econ right now at GMU, right? So how do you reconcile the breadth of your interests with the specific subject matter of a class? Do you try to bring influences from outside what you'd find in a normal textbook to a class? How separate do you give it?

Robin Hanson (2:02:20)

My standard is to not be a dilettante. So I think a dilettante would be someone who reads and maybe even talks about subjects without knowing enough about them to contribute. So the standard I hold myself to is if I'm going to go into a new area, I should learn enough about it to be able to contribute. So I should, in fact, contribute.

That's my standard. So if I try to keep that track of things and say, what have I gotten into that I was never able to contribute to? And that counts against doing things like that. But the things that I got into that I was able to contribute to, that counts to doing more things like that. So if I get into law and economics and I can find original contributions, then I think, OK, I know enough about this to be here. And I can justify my being in that area.

Theo Jaffee (2:03:12)

Well, I think that's an excellent place to wrap it up. So thank you so much, Robin Hanson for coming on the show. I really enjoyed this one. Yeah.

Robin Hanson (2:03:18)

Nice to talk to you, Theo.

Theo's Substack
Theo Jaffee Podcast
Deep conversations with brilliant people.
Listen on
Substack App
RSS Feed
Appears in episode
Theo Jaffee
Robin Hanson