back to indexRobin Hanson: Alien Civilizations, UFOs, and the Future of Humanity | Lex Fridman Podcast #292
link |
we can actually figure out where are the aliens out there in space time by being clever about the
link |
few things we can see, one of which is our current date. And so now that you have this living
link |
cosmology, we can tell the story that the universe starts out empty. And then at some point, things
link |
like us appear very primitive, and then some of those stop being quiet and expand. And then for
link |
a few billion years, they expand, and then they meet each other. And then for the next hundred
link |
billion years, they commune with each other. That is, the usual models of cosmology say that in
link |
roughly 150 billion years, the expansion of the universe will happen so much that all you'll have
link |
left is some galaxy clusters that are sort of disconnected from each other. But before then,
link |
they will interact. There will be this community of all the grabby alien civilizations, and each
link |
one of them will hear about and even meet thousands of others. And we might hope to join
link |
them someday and become part of that community. The following is a conversation with Robin Hansen,
link |
an economist at George Mason University, and one of the most fascinating, wild, fearless,
link |
and fun minds I've ever gotten the chance to accompany for a time in exploring questions
link |
of human nature, human civilization, and alien life out there in our impossibly big universe.
link |
He is the coauthor of a book titled The Elephant in the Brain, Hidden Motives in Everyday Life,
link |
The Age of M, Work, Love, and Life When Robots Rule the Earth, and a fascinating recent paper
link |
I recommend on quote, Grabby Aliens, titled If Loud Aliens Explain Human Earliness,
link |
Quiet Aliens Are Also Rare. This is the Lex Friedman podcast. To support it, please check
link |
out our sponsors in the description. And now, dear friends, here's Robin Hansen.
link |
You are working on a book about quote, grabby aliens. This is a technical term, like the Big
link |
Bang. So what are grabby aliens? Grabby aliens expand fast into the universe and they change
link |
stuff. That's the key concept. So if they were out there, we would notice. That's the key idea. So
link |
the question is, where are the grabby aliens? So Fermi's question is, where are the aliens? And we
link |
could vary that in two terms, right? Where are the quiet, hard to see aliens? And where are the
link |
big, loud, grabby aliens? So it's actually hard to say where all the quiet ones are, right?
link |
There could be a lot of them out there because they're not doing much. They're not making a big
link |
difference in the world. But the grabby aliens, by definition, are the ones you would see.
link |
We don't know exactly what they do with where they went, but the idea is they're in some sort
link |
of competitive world where each part of them is trying to grab more stuff and do something with
link |
it. And almost surely, whatever is the most competitive thing to do with all the stuff they
link |
grab isn't to leave it alone the way it started, right? So we humans, when we go around the Earth
link |
and use stuff, we change it. We would turn a forest into a farmland, turn a harbor into a city.
link |
So the idea is aliens would do something with it. And so we're not exactly sure what it would look
link |
like, but it would look different. So somewhere in the sky, we would see big spheres of different
link |
activity where things had been changed because they had been there. Expanding spheres. Right.
link |
So as you expand, you aggressively interact and change the environment.
link |
So the word grabby versus loud, you're using them sometimes synonymously, sometimes not.
link |
Grabby to me is a little bit more aggressive. What does it mean to be loud? What does it mean
link |
to be grabby? What's the difference? And loud in what way? Is it visual? Is it sound? Is it some
link |
other physical phenomena like gravitational waves? Are you using this kind of in a broad
link |
philosophical sense or there's a specific thing that it means to be loud in this universe of ours?
link |
My coauthors and I put together a paper with a particular mathematical model. And so we use the
link |
term grabby aliens to describe that more particular model. And the idea is it's a
link |
more particular model of the general concept of loud. So loud would just be the general idea that
link |
they would be really obvious. So grabby is the technical term,
link |
is it in the title of the paper? It's in the body. The title is actually about loud and quiet.
link |
Right. So the idea is you want to distinguish your particular model of things from the general
link |
category of things everybody else might talk about. So that's how we distinguish.
link |
The paper title is If Loud Aliens Explain Human Earliness, Quiet Aliens Are Also Rare.
link |
If life on earth, God, this is such a good abstract. If life on earth had to achieve
link |
N hard steps to reach humanity's level, then the chance of this event rose as time to the nth
link |
power. So we'll talk about power, we'll talk about linear increase. So what is the technical definition
link |
of grabby? How do you envision grabbiness? And why are in contrast with humans, why aren't humans
link |
grabby? So like, where's that line? Is it well definable? What is grabbing what is not grabby?
link |
We have a mathematical model of the distribution of advanced civilizations, i.e. aliens in space
link |
and time. That model has three parameters. And we can set each one of those parameters from data.
link |
And therefore, we claim this is actually what we know about where they are in space time.
link |
So the key idea is they appear at some point in space time. And then after some short delay,
link |
they start expanding. And they expand at some speed. And the speed is one of those parameters.
link |
That's one of the three. And the other two parameters are about how they appear in time.
link |
That is they appear at random places. And they appear in time according to a power law.
link |
And that power law has two parameters. And we can fit each of those parameters to data. And so then
link |
we can say, now we know, we know the distribution of advanced civilizations in space and time. So
link |
we are right now a new civilization, and we have not yet started to expand. But plausibly,
link |
we would start to do that within say, 10 million years of the current moment. That's plenty of time.
link |
And 10 million years is a really short duration in the history of the universe. So we are at the
link |
moment, a sort of random sample of the kind of times at which an advanced civilization might
link |
appear. Because we may or may not become grabby. But if we do, we'll do it soon. And so our current
link |
date is a sample. And that gives us one of the other parameters. The second parameter is the
link |
constant in front of the power law. And that's arrived from our current date.
link |
So power law, what is the N in the power law?
link |
That's the more complicated thing to explain.
link |
Right. Advanced life appeared by going through a sequence of hard steps. So starting with very
link |
simple life, and here we are at the end of this process at pretty advanced life. And so we had
link |
to go through some intermediate steps such as sexual selection, photosynthesis, multicellular
link |
animals. And the idea is that each of those steps was hard. Evolution just took a long time searching
link |
in a big space of possibilities to find each of those steps. And the challenge was to achieve
link |
all of those steps by a deadline of when the planets would no longer host simple life. And so
link |
Earth has been really lucky compared to all the other billions of planets out there,
link |
and that we managed to achieve all these steps in the short time of the five billion years that
link |
Earth can support simple life. So not all steps, but a lot of them, because we don't know how many
link |
steps there are before you start the expansion. So these are all the steps from the birth of life
link |
to the initiation of major expansion. Right. So we're pretty sure that it would happen really
link |
soon so that it couldn't be the same sort of a hard step as the last one. So in terms of taking
link |
a long time. So when we look at the history of Earth, we look at the durations of the major
link |
things that have happened. That suggests that there's roughly say six hard steps that happened,
link |
say between three and 12, and that we have just achieved the last one that would take a long time.
link |
We don't know. But whatever it is, we've just achieved the last one.
link |
We're talking about humans or aliens here. So let's talk about some of these steps. So
link |
Earth is really special in some way. We don't exactly know the level of specialness. We don't
link |
really know which steps were the hardest or not because we just have a sample of one. But you're
link |
saying that there's three to 12 steps that we have to go through to get to where we are that are hard
link |
steps, hard to find by something that took a long time and is unlikely. There's a lot of ways to fail.
link |
There's a lot more ways to fail than to succeed. The first step would be sort of the very simplest
link |
form of life of any sort. And then we don't know whether that first sort is the first sort that we
link |
see in the historical record or not. But then some other steps are, say, the development of
link |
photosynthesis, the development of sexual reproduction. There's the development of
link |
eukaryote cells, which are certain kind of complicated cells that seems to have only
link |
appeared once. And then there's multicellularity, that is multiple cells coming together to large
link |
organisms like us. And in this statistical model of trying to fit all these steps into a finite
link |
window, the model actually predicts that these steps could be of varying difficulties. That is,
link |
they could each take different amounts of time on average. But if you're lucky enough that they all
link |
appear in a very short time, then the durations between them will be roughly equal. And the time
link |
remaining leftover in the rest of the window will also be the same length. So we at the moment have
link |
roughly a billion years left on Earth until simple life like us would no longer be possible.
link |
Life appeared roughly 400 million years after the very first time when life was possible at the very
link |
beginning. So those two numbers right there give you the rough estimate of six hard steps.
link |
Just to build up an intuition here. So we're trying to create a simple mathematical model
link |
of how life emerges and expands in the universe. And there's a section in this paper, how many
link |
hard steps? Question mark. Right. The two most plausibly diagnostic Earth durations seem to be
link |
the one remaining after now before Earth becomes uninhabitable for complex life. So you estimate
link |
how long Earth lasts, how many hard steps. There's windows for doing different hard steps,
link |
and you can sort of like cueing theory, mathematically estimate of like the solution
link |
or the passing of the hard steps or the taking of the hard steps. Sort of like coldly mathematical
link |
look. If life, pre expansionary life, requires n number of steps, what is the probability of taking
link |
those steps on an Earth that lasts a billion years or two billion years or five billion years
link |
or 10 billion years? And you say solving for E using the observed durations of 1.1 and 0.4
link |
then gives E values of 3.9 and 12.5 range 5.7 to 26 suggesting a middle estimate of at least six.
link |
That's where you said six hard steps. Right. Just to get to where we are. Right. We started at the
link |
bottom. Now we're here. That took six steps on average. The key point is on average, these things
link |
on any one random planet would take trillions or trillions of years, just a really long time.
link |
And so we're really lucky that they all happened really fast in a short time before our window
link |
closed. And the chance of that happening in that short window goes as that time period to the power
link |
of the number of steps. And so that was where the power we talked about before came from. And so
link |
that means in the history of the universe, we should overall roughly expect advanced life to
link |
appear as a power law in time. So that very early on, there was very little chance of anything
link |
appearing. And then later on as things appear, other things are appearing somewhat closer to
link |
them in time because they're all going as this power law. What is the power law? Can we, for
link |
people who are not math inclined, can you describe what a power law is? So say the function X is
link |
linear and X squared is quadratic. So it's the power of two. If we make X to the three, that's
link |
cubic or the power of three. And so X to the sixth is the power of six. And so we'd say
link |
life appears in the universe on a planet like Earth in that proportion to the time that it's
link |
been ready for life to appear. And that over the universe in general, it'll appear at roughly a
link |
power law like that. What is the X, what is N? Is it the number of hard steps?
link |
Yes, the number of hard steps. So that's the idea.
link |
It's like if you're gambling and you're doubling up every time, this is the probability you just
link |
keep winning. So it gets very unlikely very quickly. And so we're the result of this unlikely
link |
chain of successes. It's actually a lot like cancer. So the dominant model of cancer in an
link |
organism like each of us is that we have all these cells and in order to become cancerous,
link |
a single cell has to go through a number of mutations and these very unlikely mutations.
link |
And so any one cell is very unlikely to have any, have all these mutations happen by the time
link |
your lifespan's over. But we have enough cells in our body that the chance of any one cell
link |
producing cancer by the end of your life is actually pretty high, more like 40%.
link |
And so the chance of cancer appearing in your lifetime also goes as a power law,
link |
this power of the number of mutations that's required for any one cell in your body to become
link |
The longer you live, the likely you are to have cancer cells.
link |
And the power is also roughly six. That is the chance of you getting cancer is
link |
roughly the power of six of the time you've been since you were born.
link |
It is perhaps not lost on people that you're comparing power laws of the survival or the
link |
arrival of the human species to cancerous cells.
link |
The same mathematical model, but of course, we might have a different value assumption
link |
about the two outcomes. But of course, from the point of view of cancer, it's more similar.
link |
For the point of view of cancer, it's a win win. We both get to thrive, I suppose.
link |
It is interesting to take the point of view of all kinds of life forms on earth,
link |
of viruses, of bacteria. They have a very different view.
link |
It's like the Instagram channel, Nature is Metal.
link |
The ethic under which nature operates doesn't often coincide, correlate with human
link |
morals. It seems cold and machine like in the selection process that it performs.
link |
I am an analyst, I'm a scholar, an intellectual, and I feel I should carefully distinguish
link |
predicting what's likely to happen and then evaluating or judging what I think would be
link |
better to happen. And it's a little dangerous to mix those up too closely because then we can
link |
then we can have wishful thinking. And so I try typically to just analyze what seems likely to
link |
happen regardless of whether I like it or that we do anything about it. And then once you see
link |
a rough picture of what's likely to happen if we do nothing, then we can ask, well, what might we
link |
prefer? And ask, where could the levers be to move it at least a little toward what we might prefer?
link |
But often doing that just analysis of what's likely to happen if we do nothing offends many
link |
people. They find that dehumanizing or cold or metal, as you say, to just say, well, this is
link |
what's likely to happen and it's not your favorite, sorry, but maybe we can do something, but maybe
link |
we can't do that much. This is very interesting that the cold analysis, whether it's geopolitics,
link |
whether it's medicine, whether it's economics, sometimes misses some very specific aspect of
link |
human condition. Like for example, when you look at a doctor and the act of a doctor helping a
link |
single patient, if you do the analysis of that doctor's time and cost of the medicine or the
link |
surgery or the transportation of the patient, this is the Paul Farmer question, you know, is it worth
link |
spending ten, twenty, thirty thousand dollars on this one patient? When you look at all the people
link |
that are suffering in the world, that money could be spent so much better. And yet there's something
link |
about human nature that wants to help the person in front of you, and that is actually the right
link |
thing to do, despite the analysis. And sometimes when you do the analysis, there's something
link |
about the human mind that allows you to not take that leap, that irrational leap to act in this way,
link |
that the analysis explains it away. Well it's like, for example, the U.S. government, you know, the
link |
DOT, Department of Transportation, puts a value of I think like nine million dollars on a human life.
link |
And the moment you put that number on a human life, you can start thinking, well okay, I can start
link |
making decisions about this or that, and with a sort of cold economic perspective, and then you
link |
might lose, you might deviate from a deeper truth of what it means to be human somehow. You have to
link |
dance, because then if you put too much weight on the anecdotal evidence on these kinds of human
link |
emotions, then you're going to lose, you could also probably more likely deviate from truth.
link |
But there's something about that cold analysis. Like I've been listening to a lot of people
link |
coldly analyze wars. War in Yemen, war in Syria, Israel, Palestine, war in Ukraine, and there's
link |
something lost when you do a cold analysis of why something happened. When you talk about energy,
link |
talking about sort of conflict, competition over resources, when you talk about geopolitics,
link |
sort of models of geopolitics, and why a certain war happened, you lose something about the suffering
link |
that happens. I don't know. It's an interesting thing, because you're both, you're exceptionally good
link |
at models in all domains, literally, but also there's a humanity to you. So it's an interesting
link |
dance. I don't know if you can comment on that dance. Sure. It's definitely true, as you say,
link |
that for many people, if you are accurate in your judgment of, say, for a medical patient,
link |
what's the chance that this treatment might help? And what's the cost? And compare those
link |
to each other, and you might say, this looks like a lot of cost for a small medical gain.
link |
And at that point, knowing that fact, that might take the air out of your sails. You might
link |
not be willing to do the thing that maybe you feel is right anyway, which is still to pay for it.
link |
And then somebody knowing that might want to keep that news from you and not tell you about
link |
the low chance of success or the high cost in order to save you this
link |
tension, this awkward moment where you might fail to do what they and you think is right.
link |
But I think the higher calling, the higher standard to hold you to, which many people
link |
can be held to, is to say, I will look at things accurately, I will know the truth,
link |
and then I will also do the right thing with it. I will be at peace with my judgment about what
link |
the right thing is in terms of the truth. I don't need to be lied to in order to figure out what the
link |
right thing to do is. And I think if you do think you need to be lied to in order to figure out
link |
what the right thing to do is, you're at a great disadvantage because then people will be lying
link |
to, you will be lying to yourself, and you won't be as effective at achieving whatever good you
link |
were trying to achieve. But getting the data, getting the facts is step one, not the final
link |
step. So I would say having a good model, getting the good data is step one, and it's a burden.
link |
Because you can't just use that data to arrive at sort of the easy convenient thing. You have
link |
to really deeply think about what is the right thing. So the dark aspect of data, of models,
link |
is you can use it to excuse away actions that aren't ethical. You can use data to basically
link |
excuse away anything. But not looking at data lets you excuse yourself to pretend and think
link |
that you're doing good when you're not. Exactly. But it is a burden. It doesn't excuse you from
link |
still being human and deeply thinking about what is right. That very kind of gray area,
link |
that very subjective area, that's part of the human condition. But let us return for a time
link |
to aliens. So you started to define sort of the model, the parameters of grabbiness.
link |
As we approach grabbiness. So what happens? So again, there was three parameters. There's the
link |
speed at which they expand, there's the rate at which they appear in time, and that rate has a
link |
constant and a power. So we've talked about the history of life on Earth suggests that power is
link |
around 6, but maybe 3 to 12. We can say that constant comes from our current date, sort of
link |
sets the overall rate. And the speed, which is the last parameter, comes from the fact that when we
link |
look in the sky, we don't see them. So the model predicts very strongly that if they were expanding
link |
slowly, say 1% of the speed of light, our sky would be full of vast spheres that were full
link |
of activity. That is, at a random time when a civilization is first appearing, if it looks out
link |
into its sky, it would see many other grabby alien civilizations in the sky, and they would be much
link |
bigger than the full moon. There'd be huge spheres in the sky, and they would be visibly different.
link |
We don't see them. Can we pause for a second? Okay. There's a bunch of hard steps that Earth had to
link |
pass to arrive at this place we are currently, which we're starting to launch rockets onto space.
link |
We're kind of starting to expand a bit, very slowly. Okay. But this is like the birth. If you
link |
look at the entirety of the history of Earth, we're now at this precipice of like expansion.
link |
We could, we might not choose to, but if we do, we will do it in the next 10 million years.
link |
10 million. Wow. Time flies when you're having fun.
link |
10 million is a short time on the cosmological scale. So that is, it might be only a thousand,
link |
but the point is if it's, even if it's up to 10 million, that hardly makes any difference to the
link |
model. So I might as well give you 10 million. This, this makes me feel, I was, I was so stressed
link |
about planning what I'm going to do today. And now you got plenty of time, plenty of time.
link |
Uh, just need to be generating some offspring quickly here. Okay. Um, so, and there's this moment
link |
this 10 million, uh, year gap, uh, or window when we start expanding and you're saying, okay,
link |
so this is an interesting moment where there's a bunch of other alien civilizations that might at
link |
some history of the universe arrived at this moment we're here, they passed all the hard steps.
link |
There's a, there's a model for how likely it is that that happens. And then they start expanding
link |
and you think of an expansion is almost like a sphere. Right. That's when you say speed,
link |
we're talking about the speed of the radius growth. Exactly. Like the surface, how fast the
link |
surface. Okay. And so you're saying that there is some speed for that expansion, average speed,
link |
and then we can play with that parameter. And if that speed is super slow, then maybe that
link |
explains why we haven't seen anything. If it's super fast, the slow would create the puzzle.
link |
It's slow predicts, we would see them, but we don't see them as a way to explain that is that
link |
they're fast. So the idea is if they're moving really fast, then we don't see them until they're
link |
almost here. Okay, this is counterintuitive. All right, hold on a second. So I think this
link |
works best when I say a bunch of dumb things. Okay. And then you elucidate the full complexity
link |
and the beauty of the dumbness. Okay. So there's these spheres out there in the universe that are
link |
made visible because they're sort of using a lot of energy. So they're generating a lot of light
link |
stuff. They're changing things. They're changing things. And change would be visible a long way
link |
off. Yes. They would take apart stars, rearrange them, restructure galaxies. They would do all
link |
kinds of big, huge stuff. Okay. If they're expanding slowly, we would see a lot of them
link |
because the universe is old, is old enough to where we would see them. That is we're assuming
link |
we're just typical, you know, maybe at the 50th percentile of them. So like half of them have
link |
appeared so far. The other half will still appear later. And the math of our best estimate is that
link |
they appear roughly once per million galaxies. And we would meet them in roughly a billion years
link |
if we expanded out to meet them. So we're looking at a grabby aliens model
link |
3D sim. That's the actual name of the video. By the time we get to 13.8 billion years, the fun
link |
begins. Okay. So this is, we're watching a three dimensional sphere rotating. I presume that's the
link |
universe. And then grabby aliens are expanding and filling that universe with all kinds of fun.
link |
Pretty soon it's all full. It's full. So that's how the grabby aliens come in contact. First of all,
link |
with other aliens and then with us humans. The following is a simulation of the grabby aliens
link |
model of alien civilizations. Civilizations are born that expand outwards at constant speed.
link |
A spherical region of space is shown. By the time we get to 13.8 billion years,
link |
this sphere will be about 3000 times as wide as the distance from the Milky Way to Andromeda.
link |
Okay. This is fun.
link |
Okay. It's huge. All right. So why don't we see, we're one little tiny, tiny, tiny, tiny dot in
link |
that giant, giant sphere. Why don't we see any of the grabby aliens?
link |
It depends on how fast they expand. So you could see that if they expanded at the speed of light,
link |
you wouldn't see them until they were here. So like out there, if somebody is destroying the
link |
universe with a vacuum decay, there's this, there's this doomsday scenario where somebody
link |
somewhere could change the vacuum of the universe and that would expand at the speed of light and
link |
basically destroy everything it hit. But you'd never see that until I got here because it's
link |
expanding at the speed of light. If you're expanding really slow, then you see it from
link |
a long way off. So the fact we don't see anything in the sky tells us they're expanding fast,
link |
say over a third the speed of light. And that's really, really fast. But that's what you have to
link |
believe if we look out and you don't see anything. Now you might say, well, how, maybe I just don't
link |
want to believe this whole model. Why should I believe this whole model at all? And our best
link |
evidence why you should believe this model is our early date. We are right now almost 14 billion
link |
years into the universe on a planet around a star that's roughly 5 billion years old.
link |
But the average star out there will last roughly 5 trillion years. That is a thousand times longer.
link |
And remember that power law, it says that the chance of advanced life appearing on a planet
link |
goes as the power of sixth of the time. So if a planet lasts a thousand times longer,
link |
then the chance of it appearing on that planet, if everything would stay empty at least, is a
link |
thousand to the sixth power or 10 to the 18. So enormous, overwhelming chance that if the universe
link |
would just stay sit and empty and waiting for advanced life to appear, when it would appear
link |
would be way at the end of all these planet lifetimes. That is the long planets near the end
link |
of the lifetime, trillions of years into the future. So, but we're really early compared to that. And
link |
our explanation is at the moment, as you saw in the video, the universe is filling up in roughly a
link |
billion years, it'll all be full. And at that point, it's too late for advanced life to show up.
link |
So you had to show up now before that deadline. Okay. Can we break that apart a little bit? Okay.
link |
Or linger on some of the things you said. So with the power law, the things we've done on earth,
link |
the model you have says that it's very unlikely, like we're lucky SOBs. Is that mathematically
link |
correct to say? We're crazy early. That is when early means like in the history of the universe.
link |
In the history. Okay. So given this model, how do we make sense of that? If we're super,
link |
can we just be the lucky ones? Well, 10 to the 18 lucky, you know,
link |
how lucky do you feel? So, you know, that's a pretty lucky, right? 10 to the 18 is a billion,
link |
billion. So then if you were just being honest and humble, that that means, what does that mean?
link |
It means one of the assumptions that calculated this crazy early must be wrong. That's what it
link |
means. So the key assumption we suggest is that the universe would stay empty. So most life would
link |
stay empty. So most life would appear like a thousand times longer later than now if everything
link |
would stay empty, waiting for it to appear. So what is not empty?
link |
So the grabby aliens are filling the universe right now. Roughly at the moment, they filled
link |
half of the universe and they've changed it. And when they fill everything, it's too late for stuff
link |
like us to appear. But wait, hold on a second. Did anyone help us get lucky? If it's so difficult,
link |
what, how do like, so it's like cancer, right? There's all these cells, each of which randomly
link |
does or doesn't get cancer. And eventually some cell gets cancer and you know, we were one of
link |
those, but hold on a second. Okay. But we got it early. We got early compared to the prediction
link |
with an assumption that's wrong. That's so that's how we do a lot of, you know, theoretical
link |
analysis. You have a model that makes a prediction that's wrong. Then that helps you reject that
link |
model. Okay. Let's try to understand exactly where the wrong is. So the assumption is that the
link |
universe is empty, stays empty, stays empty and waits until this advanced life appears in trillions
link |
of years. That is if the universe would just stay empty, if there was just, you know, nobody else
link |
out there, then when you should expect advanced life to appear, if you're the only one in the
link |
universe, when should you expect to appear? You should expect to appear trillions of years in the
link |
future. I see. Right, right. So this is a very sort of nuanced mathematical assumption. I don't
link |
think we can intuit it cleanly with words. But if you assume that you're just wait, the universe
link |
stays empty and you're waiting for one life civilization to pop up, then it's gonna, it
link |
should happen very late, much later than now. And if you look at Earth, the way things happen on
link |
Earth, it happened much, much, much, much, much earlier than it was supposed to according to this
link |
model. If you take the initial assumption, therefore you can say, well, the initial assumption of the
link |
universe staying empty is very unlikely. Right. And the other alternative theory is the universe
link |
is filling up and will fill up soon. And so we are typical for the origin data of things that
link |
can appear before the deadline. Before the deadline. Okay, it's filling up. So why don't we see anything
link |
if it's filling up? Because they're expanding really fast. Close to the speed of light. Exactly.
link |
So we will only see it when it's here. Almost here. Okay. What are the ways in which we might see
link |
a quickly expanding? This is both exciting and terrifying. It is terrifying. It's like watching
link |
a truck, like driving at you at 100 miles an hour. And so we would see spheres in the sky,
link |
at least one sphere in the sky, growing very rapidly. And like very rapidly, right? Yes,
link |
very rapidly. So we're not, so there's, there's, you know, different def because we were just
link |
talking about 10 million years. This would be, you might see it 10 million years in advance coming.
link |
I mean, you still might have a long warning. Again, the universe is 14 billion years old.
link |
The typical origin times of these things are spread over several billion years. So the chance
link |
of one originating at a, you know, very close to you in time is very low. So they still might take
link |
millions of years from the time you see it, from the time it gets here. You'll have a million years
link |
of your years to be terrified of this mass sphere coming at you. But coming at you very fast. So if
link |
they're traveling close to the speed of light, but they're coming from a long way away. So remember,
link |
the rate at which they appear is one per million galaxies, right? So they're roughly a hundred
link |
galaxies away. I see. So the Delta between the speed of light and their actual travel speed is
link |
very important, right? So even if they're going at say half the speed of light, we'll have a long
link |
time then. Yeah. But what if they're traveling exactly at a speed of light? Then we see them,
link |
like then we wouldn't have much warning, but that's less likely. Well, we can't exclude it.
link |
And they could also be somehow traveling faster than the speed of light.
link |
But I think we can't exclude because if they could go faster than the speed of light, then
link |
they would just already be everywhere. So in a universe where you can travel faster than the
link |
speed of light, you can go backwards in space time. So any time you appeared anywhere in space
link |
time, you could just fill up everything. Yeah. And so anybody in the future, whoever appeared,
link |
they would have been here by now. Can you exclude the possibility that those kinds of aliens aren't
link |
already here? Well, you have, we should have a different discussion of that. Okay. So let's
link |
actually lead that. Let's leave that discussion aside just to linger and understand the grabby
link |
alien expansion, which is beautiful and fascinating. Okay. So there's these giant expanding
link |
spheres of alien civilizations. Now, when those spheres collide, mathematically,
link |
it's very likely that we're not the first collision of grabby alien civilizations,
link |
I suppose is one way to say it. So there's like the first time the spheres touch each other,
link |
recognize each other. They meet. They recognize each other first before they meet.
link |
They see each other coming. They see each other coming. And then, so there's a bunch of them.
link |
There's a combinatorial thing where they start seeing each other coming. And then there's a
link |
third neighbor. It's like, what the hell? And then there's a fourth one. Okay. So what does that,
link |
you think, look like? What lessons from human nature, that's the only data we have,
link |
well, can you draw the story of the history of the universe here is what I would call a living
link |
cosmology. So what I'm excited about in part by this model is that it lets us tell a story of
link |
cosmology where there are actors who have agendas. So most ancient peoples, they had cosmologies,
link |
stories they told about where the universe came from and where it's going and what's happening
link |
out there. And their stories, they like to have agents and actors, gods or something out there
link |
doing things. And lately our favorite cosmology is dead, kind of boring. We're the only activity
link |
we know about or see and everything else just looks dead and empty. But this is now telling us,
link |
no, that's not quite right. At the moment, the universe is filling up and in a few billion years,
link |
it'll be all full. And from then on, the history of the universe will be the universe full of aliens.
link |
LW. Yeah. So that's a really good reminder, a really good way to think about cosmology is we're
link |
surrounded by a vast darkness and we don't know what's going on in that darkness until the light
link |
from whatever generate lights arrives here. So we kind of, yeah, we look up at the sky,
link |
okay, there's stars, oh, they're pretty, but you don't think about the giant expanding spheres of
link |
aliens because you don't see them. But now our date, looking at the clock, if you're clever,
link |
the clock tells you. So I like the analogy with the ancient Greeks. So you might think that an
link |
ancient Greek staring at the universe couldn't possibly tell how far away the sun was or how
link |
far away the moon is or how big the earth is. All you can see is just big things in the sky,
link |
you can't tell. But they were clever enough actually to be able to figure out the size of
link |
the earth and the distance to the moon and the sun and the size of the moon and sun. That is,
link |
they could figure those things out actually by being clever enough. And so similarly,
link |
we can actually figure out where are the aliens out there in space time by being clever about the
link |
few things we can see, one of which is our current date. And so now that you have this living
link |
cosmology, we can tell the story that the universe starts out empty and then at some point, things
link |
like us appear very primitive and then some of those stop being quiet and expand. And then for
link |
a few billion years, they expand and then they meet each other. And then for the next hundred
link |
billion years, they commune with each other. That is, the usual models of cosmology say that in
link |
roughly 150 billion years, the expansion of the universe will happen so much that all you'll have
link |
left is some galaxy clusters that are sort of disconnected from each other. But before then,
link |
for the next hundred billion years, they will interact. There will be this community of all the
link |
grabby alien civilizations and each one of them will hear about and even meet thousands of others.
link |
And we might hope to join them someday and become part of that community. That's an interesting
link |
thing to aspire to. Yes, interesting is an interesting word. Is the universe of alien
link |
civilizations defined by war as much or more than war defined human history?
link |
I would say it's defined by competition and then the question is how much competition implies war.
link |
So up until recently, competition defined life on Earth. Competition between species and organisms
link |
and among humans, competitions among individuals and communities and that competition often took
link |
the form of war in the last 10,000 years. Many people now are hoping or even expecting to sort
link |
of suppress and end competition in human affairs. They regulate business competition, they prevent
link |
military competition and that's a future I think a lot of people will like to continue and
link |
strengthen. People will like to have something close to world government or world governance or
link |
at least a world community and they will like to suppress war and many forms of business and
link |
personal competition over the coming centuries. And they may like that so much that they prevent
link |
interstellar colonization which would become the end of that era. That is interstellar colonization
link |
would just return severe competition to human or our descendant affairs and many civilizations may
link |
prefer that and ours may prefer that. But if they choose to allow interstellar colonization,
link |
they will have chosen to allow competition to return with great force. That is, there's really
link |
not much of a way to centrally govern a rapidly expanding sphere of civilization. And so I think
link |
one of the most solid things we can predict about Graviolians is they have accepted competition
link |
and they have internal competition and therefore they have the potential for competition when they
link |
meet each other at the borders. But whether that's military competition is more of an open question.
link |
LW. So military meaning physically destructive, right.
link |
So there's a lot to say there. So one idea that you kind of proposed is progress might be maximized
link |
through competition, through some kind of healthy competition, some definition of healthy. So like
link |
constructive not destructive competition. So like we would likely grab the alien civilizations would
link |
be likely defined by competition because they can expand faster because competition allows
link |
innovation and sort of the battle of ideas.
link |
LW. The way I would take the logic is to say competition just happens if you can't coordinate
link |
to stop it and you probably can't coordinate to stop it in an expanding interstellar way.
link |
LW. So competition is a fundamental force in the universe.
link |
LW. It has been so far and it would be within an expanding Graviolian civilization. But we today
link |
have the chance, many people think and hope, of greatly controlling and limiting competition
link |
within our civilization for a while. And that's an interesting choice whether to allow competition
link |
to sort of regain its full force or whether to suppress and manage it.
link |
LW. Well one of the open questions that has been raised in the past less than 100 years
link |
is whether our desire to lessen the destructive nature of competition or the destructive kind
link |
of competition will be outpaced by the destructive power of our weapons. Sort of if nuclear weapons
link |
and weapons of that kind become more destructive than our desire for peace then all it takes is
link |
one asshole at the party to ruin the party.
link |
LW. It takes one asshole to make a delay, but not that much of a delay on the cosmological
link |
scales we're talking about. So even a vast nuclear war, if it happened here right now on Earth,
link |
it would not kill all humans and it certainly wouldn't kill all life.
link |
And so human civilization would return within 100,000 years.
link |
LW. So all the history of atrocities and if you look at the Black Plague,
link |
which is not human caused atrocities or whatever.
link |
LW. There are a lot of military atrocities in history, absolutely.
link |
LW. In the 20th century. Those are, those challenge us to think about human nature,
link |
but the cosmic scale of time and space, they do not stop the human spirit, essentially.
link |
The humanity goes on through all the atrocities, it goes on.
link |
LW. So even a nuclear war isn't enough to destroy us or to stop our potential from expanding,
link |
but we could institute a regime of global governance that limited competition,
link |
including military and business competition of sorts, and that could prevent our expansion.
link |
LW. Of course, to play devil's advocate, global governance is centralized power,
link |
power corrupts, and absolute power corrupts absolutely. One of the aspects of competition
link |
that's been very productive is not letting any one person, any one country, any one center of power
link |
become absolutely powerful, because that's another lesson, is it seems to corrupt.
link |
There's something about ego in the human mind that seems to be corrupted by power,
link |
so when you say global governance, that terrifies me more than the possibility of war,
link |
LW. I think people will be less terrified than you are right now,
link |
and let me try to paint the picture from their point of view. This isn't my point of view,
link |
but I think it's going to be a widely shared point of view.
link |
LW. Yes. This is two devil's advocates arguing.
link |
LW. Okay. So for the last half century and into the continuing future, we actually have had
link |
a strong elite global community that shares a lot of values and beliefs and has created a lot
link |
of convergence in global policy. So if you look at electromagnetic spectrum or medical experiments
link |
or pandemic policy or nuclear power energy or regulating airplanes or just in a wide range
link |
of area, in fact, the world has very similar regulations and rules everywhere, and it's not
link |
a coincidence because they are part of a world community where people get together at places
link |
like Davos, et cetera, where world elites want to be respected by other world elites, and they
link |
have a convergence of opinion, and that produces something like global governance,
link |
but without a global center. This is what human mobs or communities have done for a long time,
link |
that is, humans can coordinate together on shared behavior without a center by having
link |
gossip and reputation within a community of elites. And that is what we have been doing and
link |
are likely to do a lot more of. So for example, one of the things that's happening, say, with the
link |
war in Ukraine is that this world community of elites has decided that they disapprove of the
link |
Russian invasion and they are coordinating to pull resources together from all around the world in
link |
order to oppose it, and they are proud of that, sharing that opinion in there, and they feel that
link |
they are morally justified in their stance there. And that's the kind of event that actually brings
link |
world elite communities together, where they come together and they push a particular policy and
link |
position that they share and that they achieve successes. And the same sort of passion animates
link |
global elites with respect to, say, global warming or global poverty and other sorts of things. And
link |
they are, in fact, making progress on those sorts of things through shared global community of
link |
elites. And in some sense, they are slowly walking toward global governance, slowly strengthening
link |
various world institutions of governance, but cautiously, carefully watching out for the
link |
possibility of a single power that might corrupt it. I think a lot of people over the coming
link |
centuries will look at that history and like it. It's an interesting thought. And thank you for
link |
playing that devil's advocate there. But I think the elites too easily lose touch of the morals
link |
that the best of human nature and power corrupts. Sure, but their view is the one that determines
link |
what happens. Their view may still end up there, even if you or I might criticize it from that
link |
point of view. So from a perspective of minimizing human suffering, elites can use topics of the war
link |
in Ukraine and climate change and all of those things to sell an idea to the world. And with
link |
disregard to the amount of suffering it causes, their actual actions. So like you can tell all
link |
kinds of narratives. That's the way propaganda works. Hitler really sold the idea that everything
link |
Germany is doing is either it's the victim is defending itself against the cruelty of the world,
link |
and it's actually trying to bring out about a better world. So every power center thinks they're
link |
doing good. And so this is the positive of competition, of having multiple power centers.
link |
This kind of gathering of elites makes me very, very, very nervous. The dinners, the meetings
link |
and the closed rooms. I don't know. But remember we talked about separating our cold analysis of
link |
what's likely or possible from what we prefer. And so this isn't exactly enough time for that.
link |
We might say, I would recommend we don't go this route of a strong world governance. And because
link |
I would say it'll preclude this possibility of becoming rabid aliens, of filling the next
link |
nearest million galaxies for the next billion years with vast amounts of activity and interest
link |
and value of life out there. That's the thing we would lose by deciding that we wouldn't expand,
link |
that we would stay here and keep our comfortable shared governance.
link |
So you wait, you think that global governance is, makes it more likely or less likely that
link |
we expand out into the universe?
link |
This is the key, this is the key point.
link |
Right. Right. So screw the elites.
link |
We want to, wait, do we want to expand?
link |
So again, I want to separate my neutral analysis from my evaluation and say,
link |
first of all, I have an analysis that tells us this is a key choice that we will face and that
link |
it's key choice other aliens have faced out there. And it could be that only one in 10 or one in 100
link |
civilizations chooses to expand and the rest of them stay quiet. And that's how it goes out there.
link |
And we face that choice too. And it'll happen sometime in the next 10 million years,
link |
maybe the next thousand. But the key thing to notice from our point of view is that
link |
even though you might like our global governance, you might like the fact that we've come together,
link |
we no longer have massive wars and we no longer have destructive competition.
link |
And that we could continue that, the cost of continuing that would be to prevent
link |
interstellar colonization. That is once you allow interstellar colonization, then you've lost
link |
control of those colonies and whatever they change into, they could come back here and compete with
link |
you back here as a result of having lost control. And I think if people value that global governance
link |
and global community and regulation and all the things it can do enough, they would then
link |
want to prevent interstellar colonization.
link |
I want to have a conversation with those people. I believe that both for humanity,
link |
for the good of humanity, for what I believe is good in humanity and for expansion, exploration,
link |
innovation, distributing the centers of power is very beneficial. So this whole meeting of elites
link |
and I've been very fortunate to meet quite a large number of elites. They make me nervous
link |
because it's easy to lose touch of reality. I'm nervous about that in myself to make sure that
link |
you never lose touch as you get sort of older, wiser, you know, how you generally get like
link |
disrespectful of kids, kids these days. No, the kids are okay. But I think you should hear
link |
a stronger case for their position. So I'm going to play for the elites. Yes. Well, for the limiting
link |
of expansion and for the regulation of behavior. Okay. Can I link on that? So you're saying those
link |
two are connected. So the human civilization and alien civilizations come to a crossroads.
link |
They have to decide, do we want to expand or not? And connected to that, do we want to give a lot
link |
of power to a central elite? Or do we want to distribute the power centers, which is naturally
link |
connected to the expansion? When you expand, you distribute the power. If say over the next thousand
link |
years, we fill up the solar system, right? We go out from earth and we colonize Mars and we change
link |
a lot of things. Within a solar system, still everything is within reach. That is, if there's
link |
a rebellious colony around Neptune, you can throw rocks at it and smash it and then teach them
link |
discipline. Okay. A central control over the solar system is feasible. But once you let it escape the
link |
solar system, it's no longer feasible. But if you have a solar system that doesn't have a central
link |
control, maybe broken into a thousand different political units in the solar system, then any one
link |
part of that that allows interstellar colonization and it happens. That is interstellar colonization
link |
happens when only one party chooses to do it and is able to do it. And that's what it is there for.
link |
So we can just say in a world of competition, if interstellar colonization is possible, it will
link |
happen and then competition will continue. And that will sort of ensure the continuation of
link |
competition into the indefinite future. And competition, we don't know, but competition
link |
can take violent forms and many forms. And the case I was going to make is that I think one of
link |
the things that most scares people about competition is not just that it creates holocausts and death
link |
on massive scales, is that it's likely to change who we are and what we value.
link |
Yes. So this is the other thing with power. As we grow, as human civilization grows,
link |
becomes multi planetary, multi solar system potentially, how does that change us, do you think?
link |
I think the more you think about it, the more you realize it can change us a lot.
link |
So first of all, this is pretty dark, by the way. Well, it's just honest.
link |
Right. Well, I'm trying to get there. But I think the first thing you should say,
link |
if you look at history, just human history over the last 10,000 years,
link |
if you really understood what people were like a long time ago, you'd realize they were really
link |
quite different. Ancient cultures created people who were really quite different. Most historical
link |
fiction lies to you about that. It often offers you modern characters in an ancient world.
link |
But if you actually study history, you will see just how different they were and how differently
link |
they thought. And they've changed a lot many times, and they've changed a lot across time.
link |
So I think the most obvious prediction about the future is, even if you only have the mechanisms
link |
of change we've seen in the past, you should still expect a lot of change in the future.
link |
But we have a lot bigger mechanisms for change in the future than we had in the past.
link |
So I have this book called The Age of M, Work, Love, and Life, and Robots Rule the Earth. And
link |
it's about what happens if brain emulations become possible. So a brain emulation is where you take
link |
a actual human brain, and you scan it and find spatial and chemical detail to create
link |
a computer simulation of that brain. And then those computer simulations of brains
link |
are basically citizens in a new world. They work, and they vote, and they fall in love,
link |
and they get mad, and they lie to each other. And this is a whole new world. And my book is
link |
about analyzing how that world is different than our world, basically using competition as my key
link |
lever of analysis. That is, if that world remains competitive, then I can figure out how they change
link |
in that world, what they do differently than we do. And it's very different. And it's different in
link |
ways that are shocking sometimes to many people and ways some people don't like. I think it's an
link |
okay world, but I have to admit, it's quite different. And that's just one technology.
link |
If we add dozens more technologies, changes into the future, we should just expect it's possible
link |
to become very different than who we are. I mean, in the space of all possible minds,
link |
our minds are a particular architecture, a particular structure, a particular set of habits,
link |
and they are only one piece in a vast base of possibilities. The space of possible minds is
link |
really huge. So yeah, let's linger on the space of possible minds for a moment, just to sort of
link |
humble ourselves. How peculiar our peculiarities are, like the fact that we like a particular kind
link |
of sex, and the fact that we eat food through one hole and poop through another hole. And that seems
link |
to be a fundamental aspect of life, is very important to us. And that life is finite in a
link |
certain kind of way, we have a meat vehicle. So death is very important to us. I wonder which
link |
aspects are fundamental, or would be common throughout human history and also throughout,
link |
sorry, throughout history of life on Earth, and throughout other kinds of lives. Like what is
link |
really useful? You mentioned competition seems to be a one fundamental thing.
link |
I've tried to do analysis of where our distant descendants might go in terms of what are robust
link |
features we could predict about our descendants. So again, I have this analysis of sort of the
link |
next generation, so the next era after ours. If you think of human history as having three eras
link |
so far, right? There was the forager era, the farmer era, and the industry era. Then my attempt
link |
in age of M is to analyze the next era after that. And it's very different, but of course,
link |
there could be more and more errors after that. So analyzing a particular scenario and thinking
link |
it through is one way to try to see how different the future could be, but that doesn't give you
link |
some sort of sense of what's typical. But I have tried to analyze what's typical.
link |
And so I have two predictions I think I can make pretty solidly. One thing is that we know at the
link |
moment that humans discount the future rapidly. So we discount the future in terms of caring
link |
about consequences, roughly a factor of two per generation. And there's a solid evolutionary
link |
analysis why sexual creatures would do that. Because basically your descendants only share
link |
half of your genes and your descendants are a generation away. So we only care about our
link |
grandchildren. Basically that's a factor of four later because it's later. So this actually
link |
explains typical interest rates in the economy. That is interest rates are greatly influenced by
link |
our discount rates. And we basically discount the future by a factor of two per generation.
link |
But that's a side effect of the way our preferences evolved as sexually selected
link |
creatures. We should expect that in the longer run creatures will evolve who don't discount the
link |
future. They will care about the long run and they will therefore not neglect the long run.
link |
So for example, for things like global warming or things like that, at the moment, many commenters
link |
are sad that basically ordinary people don't seem to care much, market prices don't seem to care
link |
much and more ordinary people, it doesn't really impact them much because humans don't care much
link |
about the longterm future. And futurists find it hard to motivate people and to engage people about
link |
the longterm future because they just don't care that much. But that's a side effect of this
link |
particular way that our preferences evolved about the future. And so in the future, they will neglect
link |
the future less. And that's an interesting thing that we can predict robustly. Eventually,
link |
you know, maybe a few centuries, maybe longer, eventually our descendants will
link |
care about the future. Can you speak to the intuition behind that? Is it
link |
useful to think more about the future? Right. If evolution rewards creatures for having many
link |
descendants, then if you have decisions that influence how many descendants you have,
link |
then that would be good if you made those decisions. But in order to do that, you'll have to
link |
care about them. You have to care about that future. So to push back, that's if you're trying
link |
to maximize the number of descendants. But the nice thing about not caring too much about the
link |
longterm future is you're more likely to take big risks or you're less risk averse. And it's possible
link |
that both evolution and just life in the universe rewards the risk takers. Well, we actually have
link |
analysis of the ideal risk preferences too. So there's a literature on ideal preferences that
link |
evolution should promote. And for example, there's literature on competing investment funds and what
link |
the managers of those funds should care about in terms of risk, various kinds of risks, and in terms
link |
of discounting. And so managers of investment funds should basically have logarithmic risk, i.e. in
link |
shared risk, in correlated risk, but be very risk neutral with respect to uncorrelated risk. So
link |
that's a feature that's predicted to happen about individual personal choices in biology and also
link |
for investment funds. So that's other things. That's also something we can say about the long
link |
run. What's correlated and uncorrelated risk? If there's something that would affect all of your
link |
descendants, then if you take that risk, you might have more descendants, but you might have zero.
link |
And that's just really bad to have zero descendants. But an uncorrelated risk would be a
link |
risk that some of your descendants would suffer, but others wouldn't. And then you have a portfolio
link |
of descendants. And so that portfolio ensures you against problems with any one of them.
link |
I like the idea of portfolio descendants. And we'll talk about portfolios with your idea of
link |
you briefly mentioned, we'll return there with M, EM, the age of EM, work, love, and life when
link |
robots rule the earth. EM, by the way, is emulated minds. So this one of the...
link |
M is short for emulations.
link |
M is short for emulations. And it's kind of an idea of how we might create artificial minds,
link |
artificial copies of minds, or human like intelligences.
link |
I have another dramatic prediction I can make about long term preferences.
link |
Which is, at the moment, we reproduce as the result of a hodgepodge of preferences that
link |
aren't very well integrated, but sort of in our ancestral environment induced us to reproduce.
link |
So we have preferences over being sleepy and hungry and thirsty and wanting to have sex and
link |
wanting to be excitement, et cetera, right? And so in our ancestral environment, the packages
link |
of preferences that we evolved to have did induce us to have more descendants. That's why we're here.
link |
But those packages of preferences are not a robust way to promote having more descendants.
link |
They were tied to our ancestral environment, which is no longer true. So that's one of the
link |
reasons we are now having a big fertility decline because in our current environment,
link |
our ancestral preferences are not inducing us to have a lot of kids,
link |
which is, from evolution's point of view, a big mistake.
link |
We can predict that in the longer run, there will arise creatures who
link |
just abstractly know that what they want is more descendants.
link |
That's a very robust way to have more descendants is to have that as your direct preference.
link |
First of all, your thinking is so clear. I love it. So mathematical. And thank you
link |
for thinking so clear with me and bearing with my interruptions and going on the tangents when we go
link |
there. So you're just clearly saying that successful long term civilizations will prefer to have
link |
descendants, more descendants.
link |
Not just prefer, consciously and abstractly prefer. That is, it won't be the indirect
link |
consequence of other preference. It will just be the thing they know they want.
link |
There'll be a president in the future that says, we must have more sex.
link |
We must have more descendants and do whatever it takes to do that.
link |
Whatever. We must go to the moon and do the other things. Not because they're easy,
link |
but because they're hard. But instead of the moon, let's have lots of sex. Okay.
link |
But there's a lot of ways to have descendants, right?
link |
Right. So that's the whole point. When the world gets more complicated and there are many possible
link |
strategies, it's having that as your abstract preference that will force you to think through
link |
those possibilities and pick the one that's most effective.
link |
So just to clarify, descendants doesn't necessarily mean the narrow definition of
link |
descendants, meaning humans having sex and then having babies.
link |
You can have artificial intelligence systems in whom you instill some capability of cognition
link |
and perhaps even consciousness. You can also create through genetics and biology clones of yourself
link |
or slightly modified clones, thousands of them. So all kinds of descendants. It could be descendants
link |
in the space of ideas too, for somehow we no longer exist in this meat vehicle. It's now just
link |
like whatever the definition of a life form is, you have descendants of those life forms.
link |
Yes. And they will be thoughtful about that. They will have thought about what counts as a
link |
descendant and that'll be important to them to have the right concept.
link |
So the they there is very interesting, who the they are.
link |
But the key thing is we're making predictions that I think are somewhat robust about what
link |
our distant descendants will be like. Another thing I think you would automatically accept is
link |
they will almost entirely be artificial. And I think that would be the obvious prediction
link |
about any aliens we would meet. That is they would long since have given up reproducing
link |
Well, it's like organic or something. It's all real.
link |
It might be squishy and made out of hydrocarbons, but it would be artificial in the sense of made
link |
in factories with designs on CAD things, right? Factories with scale economy. So the factories
link |
we have made on earth today have much larger scale economies than the factories in ourselves.
link |
So the factories in ourselves are, there are marvels, but they don't achieve very many scale
link |
economies. They're tiny little factories.
link |
But they're all factories.
link |
Factories on top of factories. So everything, the factories that are designed is different
link |
than sort of the factories that have evolved.
link |
Yeah. I think the nature of the word design is very interesting to uncover there. But
link |
let me, in terms of aliens, let me go, let me analyze your Twitter like it's Shakespeare.
link |
There's a tweet says, define hello, in quotes, alien civilizations as one that might the
link |
next million years identify humans as intelligent and civilized, travel to earth and say hello
link |
by making their presence and advanced abilities known to us. The next 15 polls, this is a
link |
Twitter thread, the next 15 polls ask about such hello aliens. And what these polls ask
link |
is your Twitter followers, what they think those aliens will be like certain particular
link |
qualities. So poll number one is what percent of hello aliens evolved from biological species
link |
with two main genders? And you know, the popular vote is above 80%. So most of them have two
link |
genders. What do you think about that? I'll ask you about some of these because they're
link |
so interesting. It's such an interesting question.
link |
It is a fun set of questions.
link |
Yes, it's a fun set of questions. So the genders as we look through evolutionary history, what's
link |
the usefulness of that as opposed to having just one or like millions?
link |
So there's a question in evolution of life on earth, there are very few species that
link |
have more than two genders. There are some, but they aren't very many. But there's an
link |
enormous number of species that do have two genders, much more than one. And so there's
link |
a literature on why did multiple genders evolve, and that's sort of what's the point of having
link |
males and females versus hermaphrodites. So most plants are hermaphrodites, that is they
link |
would mate male female, but each plant can be either role. And then most animals have
link |
chosen to split into males and females. And then they're differentiating the two genders.
link |
And there's an interesting set of questions about why that happens.
link |
Because you can do selection, you basically have like one gender competes for the affection
link |
of other and there's sexual partnership that creates the offspring. So there's sexual
link |
selection. It's nice to have like to a party, it's nice to have dance partners. And then
link |
then each one get to choose based on certain characteristics. And that's an efficient
link |
mechanism for adapting to the environment, being successfully adapted to the environment.
link |
It does look like there's an advantage. If you have males, then the males can take higher
link |
variants. And so there can be stronger selection among the males in terms of weeding out genetic
link |
mutations because the males have a higher variance in their mating success.
link |
Yes. Sure. Okay. Question number two, what percent of hello aliens evolved from land
link |
animals as opposed to plants or ocean slash air organisms? By the way, I did recently
link |
see that there's only 10% of species on earth are in the ocean. So there's a lot more variety
link |
on land. There is. It's interesting. So why is that? I can't even intuit exactly why that would
link |
be. Maybe survival on land is harder and so you get a lot more. The story that I understand is
link |
it's about small niches. So speciation can be promoted by having multiple different species.
link |
So in the ocean, species are larger. That is there are more creatures in each species because the
link |
ocean environments don't vary as much. So if you're good in one place, you're good in many
link |
other places. But on land, and especially in rivers, rivers contain an enormous percentage of
link |
the kinds of species on land, you see, because they vary so much from place to place. And so
link |
a species can be good in one place and then other species can't really compete because they came
link |
from a different place where things are different. So it's a remarkable fact actually that speciation
link |
promotes evolution in the long run. That is more evolution has happened on land because there have
link |
been more species on land because each species has been smaller. And that's actually a warning
link |
about something called rot that I've thought a lot about, which is one of the problems with
link |
even a world government, which is large systems of software today just consistently rot and decay
link |
with time and have to be replaced. And that plausibly also is a problem for other large
link |
systems, including biological systems, legal systems, regulatory systems. And it seems like
link |
large species actually don't evolve as effectively as small ones do. And that's an important thing
link |
to notice about that. And that's different from ordinary sort of evolution in economies on Earth
link |
in the last few centuries, say. On Earth, the more technical evolution and economic growth happens in
link |
larger integrated cities and nations. But in biology, it's the other way around. More evolution
link |
happened in the fragmented species. Yeah, it's such a nuanced discussion because you can also
link |
push back in terms of nations and at least companies. It's like large companies seems to evolve
link |
less effectively. There is something that they have more resources, they don't even have better
link |
resilience. And when you look at the scale of decades and centuries, it seems like a lot of
link |
large companies die. But still large economies do better, like large cities grow better than small
link |
cities. Large integrated economies like the United States or the European Union do better than small
link |
fragmented ones. So, yeah, sure. That's a very interesting, long discussion. But so most of the
link |
people, and obviously votes on Twitter represent the absolute objective truth of things.
link |
But an interesting question about oceans is that, okay, remember I told you about how most
link |
planets would last for trillions of years and be later, right? So people have tried to explain why
link |
life appeared on Earth by saying, oh, all those planets are going to be unqualified for life
link |
because of various problems. That is, they're around smaller stars, which last longer, and
link |
smaller stars have some things like more solar flares, maybe more tidal locking. But almost
link |
all of these problems with longer lived planets aren't problems for ocean worlds. And a large
link |
fraction of planets out there are ocean worlds. So if life can appear on an ocean world, then
link |
that pretty much ensures that these planets that last a very long time could have advanced life
link |
because there's a huge fraction of ocean worlds. So that's actually an open question.
link |
So when you say, sorry, when you say life appear, you're kind of saying life and intelligent life.
link |
So that's an open question. Is land, and that's I suppose the question behind
link |
the Twitter poll, which is a grabby alien civilization that comes to say hello,
link |
what's the chance that they first began their early steps, the difficult steps they took on
link |
land? What do you think? 80%, most people on Twitter think it's very likely on land.
link |
I think people are discounting ocean worlds too much. That is, I think people tend to assume that
link |
whatever we did must be the only way it's possible. And I think people aren't giving
link |
enough credit for other possible paths. Dolphins, Waterworld, by the way,
link |
people criticize that movie. I love that movie. Kevin Costner can do me no wrong.
link |
Okay, next question. What percent of hello aliens once had a nuclear war with greater
link |
than 10 nukes fired in anger? So not in the incompetence and as an accident,
link |
intentional firing of nukes and less than 20% was the most popular vote.
link |
And that just seems wrong to me.
link |
So like, I wonder what, so most people think once you get nukes, we're not going to fire them.
link |
They believe in the power.
link |
I think they're assuming that if you had a nuclear war, then that would just end
link |
civilization for good. I think that's the thinking.
link |
That's the main thing.
link |
And I think that's just wrong. I think you could rise again after a nuclear war.
link |
It might take 10,000 years or 100,000 years, but it could rise again.
link |
So what do you think about mutual assured destruction
link |
as a force to prevent people from firing nuclear weapons? That's a question that I knew
link |
to a terrifying degree has been raised now and what's going on.
link |
Well, I mean, clearly it has had an effect. The question is just how strong an effect for how
link |
long. I mean, clearly we have not gone wild with nuclear war and clearly the devastation that you
link |
would get if you initiated a nuclear war is part of the reasons people have been reluctant to start
link |
a war. The question is just how reliably will that ensure the absence of a war?
link |
Yeah. The night is still young.
link |
This has been 70 years or whatever it's been.
link |
I mean, but what do you think? Do you think we'll see nuclear war in the century?
link |
I don't know if in the century, but it's the sort of thing that's likely to happen eventually.
link |
That's a very loose statement. Okay. I understand. Now this is where I pull you out of your
link |
mathematical model and ask a human question. Do you think this particular human question...
link |
I think we've been lucky that it hasn't happened so far.
link |
But what is the nature of nuclear war? Let's think about this. There's dictators, there's democracies,
link |
miscommunication. How do wars start? World War I, World War II.
link |
So the biggest datum here is that we've had an enormous decline in major war over the last
link |
century. So that has to be taken into account now. So the problem is war is a process that has a very
link |
long tail. That is, there are rare, very large wars. So the average war is much worse than the
link |
median war because of this long tail. And that makes it hard to identify trends over time. So
link |
the median war has clearly gone way down in the last century at a medium rate of war. But it could
link |
be that's because the tail has gotten thicker. And in fact, the average war is just as bad,
link |
but most wars are gonna be big wars. So that's the thing we're not so sure about.
link |
There's no strong data on wars with one, because of the destructive nature of the weapons,
link |
kill hundreds of millions of people. There's no data on this.
link |
So, but we can start intuiting.
link |
But we can see that the power law, we can do a power law fit to the rate of wars and it's a
link |
power law with a thick tail. So it's one of those things that you should expect most of the damage
link |
to be in the few biggest ones. So that's also true for pandemics and a few other things. For
link |
pandemics, most of the damages in the few biggest ones. So the median pandemics of ours, less than
link |
the average that you should expect in the future. But those, that fitting of data is very questionable
link |
because everything you said is correct. The question is like, what can we infer about the
link |
future of civilization, threatening pandemics or nuclear war from studying the history of the
link |
20th century? So like, you can't just fit it to the data, the rate of wars and the destructive
link |
nature. Like that's not, that's not how nuclear war will happen. Nuclear war happens with two
link |
assholes or idiots that have access to a button.
link |
Small wars happen that way too.
link |
No, I understand that, but that's, it's very important. Small wars aside, it's very important
link |
to understand the dynamics, the human dynamics and the geopolitics of the way nuclear war happens
link |
in order to predict how we can minimize the chance of a...
link |
But it is a common and useful intellectual strategy to take something that could be really
link |
big or, but is often very small and fit the distribution of the data, small things, which
link |
you have a lot of them and then ask, do I believe the big things are really that different? Right?
link |
So sometimes it's reasonable to say like, say with tornadoes or even pandemics or something,
link |
the underlying process might not be that different for the big and small ones.
link |
It might not be. The fact that mutual sure destruction seems to work to some degree
link |
shows you that to some degree it's different than the small wars.
link |
So it's a really important question to understand is, are humans capable, one human, like how many
link |
humans on earth, if I give them a button now, say you pressing this button will kill everyone on
link |
earth, everyone, right? How many humans will press that button? I want to know those numbers,
link |
like day to day, minute to minute, how many people have that much irresponsibility, evil,
link |
incompetence, ignorance, whatever word you want to assign, there's a lot of dynamics of the
link |
psychology that leads you to press that button, but how many? My intuition is the number, the more
link |
destructive that press of a button, the fewer humans you find. And that number gets very close
link |
to zero very quickly, especially people have access to such a button, but that's perhaps
link |
a hope than a reality. And unfortunately we don't have good data on this,
link |
which is like how destructive are humans willing to be?
link |
So I think part of this just has to think about, ask what your time scales you're looking at,
link |
right? So if you say, if you look at the history of war, you know, we've had a lot of wars pretty
link |
consistently over many centuries. So if I ask, if you ask, will we have a nuclear war in the
link |
next 50 years? I might say, well, probably not. If I say 500 or 5,000 years, like if the same sort
link |
of risks are underlying and they just continue, then you have to add that up over time and think
link |
the risk is getting a lot larger the longer a timescale we're looking at.
link |
But okay, let's generalize nuclear war because what I was more referring to is something that
link |
kills more than 20% of humans on earth and injures or makes the other 80%
link |
suffer horribly, survive, but suffer. That's what I was referring to. So when you look at 500 years
link |
from now, that might not be nuclear war. That might be something else. That's that kind of,
link |
has that destructive effect. And I don't know, these feel like novel questions in the history
link |
of humanity. I just don't know. I think since nuclear weapons, this has been, you know,
link |
engineering pandemics, for example, robotics, so nanobots. It just seems like a real new
link |
possibility that we have to contend with it. We don't have good models or from my perspective.
link |
So if you look on say the last thousand years or 10,000 years, we could say we've seen a certain
link |
rate at which people are willing to make big destruction in terms of war. Okay. And if you're
link |
willing to project that data forward, that I think like if you want to ask over periods of
link |
thousands or tens of thousands of years, you would have a reasonable data set. So the key
link |
question is what's changed lately? Okay. And so a big question of which I've given a lot of thought
link |
to what are the major changes that seem to have happened in culture and human attitudes over the
link |
last few centuries and what's our best explanation for those so that we can project them forward into
link |
the future. And I have a story about that, which is the story that we have been drifting back toward
link |
forager attitudes in the last few centuries as we get rich. So the idea is we spent a million years
link |
being a forager and that was a very sort of standard lifestyle that we know a lot about.
link |
Foragers sort of live in small bands. They make decisions cooperatively. They share food. They,
link |
you know, they don't have much property, et cetera. And humans liked that. And then 10,000 years ago,
link |
farming became possible, but it was only possible because we were plastic enough to really change
link |
our culture. Farming styles and cultures are very different. They have slavery, they have war,
link |
they have property, they have inequality, they have kings. They stay in one place instead of
link |
wandering. They don't have as much diversity of experience or food. They have more disease.
link |
This farming life is just very different. But humans were able to sort of introduce conformity
link |
and religion and all sorts of things to become just a very different kind of creature as farmers.
link |
Farmers are just really different than foragers in terms of their values and their lives.
link |
But the pressures that made foragers into farmers were part mediated by poverty.
link |
Farmers are poor. And if they deviated from the farming norms that people around them supported,
link |
they were quite at risk of starving to death. And then in the last few centuries,
link |
we've gotten rich. And as we've gotten rich, the social pressures that turned foragers into farmers
link |
have become less persuasive to us. So, for example, a farming young woman who was told,
link |
if you have a child out of wedlock, you and your child may starve, that was a credible threat.
link |
She would see actual examples around her to make that a believable threat. Today,
link |
if you say to a young woman, you shouldn't have a child out of wedlock, she will see other young
link |
woman around her doing okay that way. We're all rich enough to be able to afford that sort of a
link |
thing. And therefore, she's more inclined often to go with her inclinations, her sort of more
link |
natural inclinations about such things rather than to be pressured to follow the official
link |
farming norms of that you shouldn't do that sort of thing. And all through our lives, we have been
link |
drifting back toward forager attitudes because we've been getting rich. And so, aside from at
link |
work, which is an exception, but elsewhere, I think this explains trends toward less slavery,
link |
more democracy, less religion, less fertility, more promiscuity, more travel, more art, more leisure,
link |
fewer work hours. All of these trends are basically explained by becoming more forager like.
link |
And much science fiction celebrates this, Star Trek or the culture novels, people
link |
like this image that we are moving toward this world. We're basically like foragers, we're peaceful,
link |
we share, we make decisions collectively, we have a lot of free time, we are into art.
link |
So forger, you know, forger is a word and it has, it's a loaded word because it's connected to
link |
the actual, what life was actually like at that time. As you mentioned, we sometimes don't do a
link |
good job of telling accurately what life was like back then. But you're saying if it's not exactly
link |
like foragers, it rhymes in some fundamental way. You also said peaceful. Is it obvious that a
link |
forager with a nuclear weapon would be peaceful? I don't know if that's 100% obvious. So we know,
link |
again, we know a fair bit about what foragers lives were like. The main sort of violence they
link |
had would be sexual jealousy. They were relatively promiscuous and so there'd be a lot of jealousy.
link |
But they did not have organized wars with each other. That is, they were at peace with their
link |
neighboring forager bands. They didn't have property in land or even in people. They didn't
link |
really have marriage. And so they were, in fact, peaceful.
link |
When you think about large scale wars, they don't start large scale wars.
link |
They didn't have coordinated large scale wars like the way chimpanzees do. Our chimpanzees do
link |
have wars between one tribe of chimpanzees and others, but human foragers do not. Farmers return
link |
to that, of course, the more chimpanzee like styles. Well, that's a hopeful message. If we
link |
could return real quick to the Hello Aliens Twitter thread. One of them is really interesting
link |
about language. What percent of Hello Aliens would be able to talk to us in our language?
link |
This is the question of communication. It actually gets to the nature of language.
link |
It also gets to the nature of how advanced you expect them to be.
link |
So I think some people see that we have advanced over the last thousands of years,
link |
and we aren't reaching any sort of limit. And so they tend to assume it could go on forever.
link |
And I actually tend to think that within, say, 10 million years, we will sort of max out on
link |
technology. We will sort of learn everything that's feasible to know for the most part. And then
link |
obstacles to understanding would more be about sort of cultural differences, like ways in which
link |
different places had just chosen to do things differently. And so then the question is, is it
link |
even possible to communicate across some cultural distances? And I could imagine some maybe advanced
link |
aliens who just become so weird and different from each other, they can't communicate with each other.
link |
But we're probably pretty simple compared to them. So I would think, sure, if they wanted to,
link |
they could communicate with us. So it's the simplicity of the recipient. I tend to,
link |
just to push back, let's explore the possibility where that's not the case. Can we communicate
link |
with ants? I find that this idea that... We're not very good at communicating in general.
link |
Oh, you're saying... All right, I see. You're saying once you get orders of magnitude better
link |
at communicating... Once they had maxed out on all communication technology in general,
link |
and they just understood in general how to communicate with lots of things, and had done
link |
that for millions of years. But you have to be able to... This is so interesting. As somebody
link |
who cares a lot about empathy and imagining how other people feel, communication requires empathy,
link |
meaning you have to truly understand how the other person, the other organism sees the world.
link |
It's like a four dimensional species talking to a two dimensional species. It's not as trivial as,
link |
to me at least, as it might at first seem. So let me reverse my position a little,
link |
because I'll say, well, the hello aliens question really combines two different scenarios
link |
that we're slipping over. So one scenario would be that the hello aliens would be like grabby
link |
aliens. They would be just fully advanced. They would have been expanding for millions of years.
link |
They would have a very advanced civilization, and then they would finally be arriving here
link |
after a billion years perhaps of expanding, in which case they're going to be crazy advanced
link |
at some maximal level. But the hello aliens about aliens we might meet soon, which might be sort of
link |
UFO aliens, and UFO aliens probably are not grabby aliens. How do you get here if you're
link |
not a grabby alien? Well, they would have to be able to travel. Oh. But they would not be expansive.
link |
So the road trip doesn't count as a grabby. So we're talking about expanding the colony,
link |
the comfortable colony. So the question is, if UFOs, some of them are aliens,
link |
what kind of aliens would they be? This is sort of the key question you have to ask in order to
link |
try to interpret that scenario. The key fact we would know is that they are here right now,
link |
but the universe around us is not full of an alien civilization. So that says right off the bat
link |
that they chose not to allow massive expansion of a grabby civilization.
link |
Is it possible that they chose it, but we just don't see them yet? These are the stragglers,
link |
the journeymen. So the timing coincidence is, it's almost surely if they are here now,
link |
they are much older than us. They are many millions of years older than us. And so they
link |
could have filled the galaxy in that last millions of years if they had wanted to.
link |
That is, they couldn't just be right at the edge. Very unlikely. Most likely they would have been
link |
around waiting for us for a long time. They could have come here any time in the last millions of
link |
years and they just chosen, they've been waiting around for this or they just chose to come
link |
recently. But the timing coincidence, it would be crazy unlikely that they just happen to be able to
link |
get here, say in the last hundred years. They would no doubt have been able to get here far
link |
earlier than that. Again, we don't know. So this is a friend like UFO sightings on earth. We don't
link |
know if this kind of increase in sightings have anything to do with actual visitations.
link |
I'm just talking about the timing. They arose at some point in space time.
link |
And it's very unlikely that that was just to the point that they could just barely get here
link |
recently. Almost surely they could have gotten here much earlier. And throughout the stretch
link |
of several billion years that earth existed, they could have been here often. Exactly. So
link |
they could have therefore filled the galaxy long time ago if they had wanted to. Let's push back
link |
on that. The question to me is, isn't it possible that the expansion of a civilization is much
link |
harder than the travel? The sphere of the reachable is different than the sphere of the colonized.
link |
So isn't it possible that the sphere of places where like the stragglers go, the different
link |
people that journey out, the explorers, is much, much larger and grows much faster than the
link |
civilization? So in which case, like they would visit us. There's a lot of visitors, the grad
link |
students of the civilization. They're like exploring, they're collecting the data, but
link |
we're not yet going to see them. And by yet, I mean across millions of years.
link |
The time delay between when the first thing might arrive and then when colonists could arrive in
link |
mass and do a mass amount of work is cosmologically short. In human history, of course, sure, there
link |
might be a century between that, but a century is just a tiny amount of time on the scales we're
link |
talking about. So this is, in computer science, ant colony optimization. It's true for ants.
link |
So it's like when the first ant shows up, it's likely if there's anything of value,
link |
it's likely the other ants will follow quickly. Yeah.
link |
Relatively short. It's also true that traveling over very long distances, probably one of the
link |
main ways to make that feasible is that you land somewhere, you colonize a bit, you create new
link |
resources that can then allow you to go farther. Many short hops as opposed to a giant long journey.
link |
Exactly. Those hops require that you are able to start a colonization of sorts along those hops.
link |
You have to be able to stop somewhere, make it into a way station such that you can then support
link |
you moving farther. So what do you think of, there's been a lot of UFO sightings. What do
link |
you think about those UFO sightings and what do you think if any of them are of extraterrestrial
link |
origin and we don't see giant civilizations out in the sky, how do you make sense of that then?
link |
I want to do some clearing of throats, which people like to do on this topic, right? They want
link |
to make sure you understand they're saying this and not that, right? So I would say the analysis
link |
needs both a prior and a likelihood. So the prior is what are the scenarios that are at all plausible
link |
in terms of what we know about the universe. And then the likelihood is the particular actual
link |
sightings, like how hard are those to explain through various means. I will establish myself
link |
as somewhat of an expert on the prior. I would say my studies and the things I've studied make me an
link |
expert and I should stand up and have an opinion on that and be able to explain it. The likelihood,
link |
however, is not my area of expertise. That is, I'm not a pilot. I don't do atmospheric studies of
link |
studies of things I haven't studied in detail, the various kinds of atmospheric phenomena or
link |
whatever that might be used to explain the particular sightings. I can just say from
link |
my amateur stance, the sightings look damn puzzling. They do not look easy to dismiss.
link |
The attempts I've seen to easily dismiss them seem to me to fail. It seems like these are
link |
pretty puzzling, weird stuff that deserve an expert's attention in terms of considering,
link |
asking what the likelihood is. So analogy I would make as a murder trial. On average, if we say,
link |
what's the chance any one person murdered another person as a prior probability, maybe one in a
link |
thousand people get murdered. Maybe each person has a thousand people around them who could
link |
plausibly have done it. So the prior probability of a murder is one in a million. But we allow
link |
murder trials because often evidence is sufficient to overcome a one in a million prior because the
link |
evidence is often strong enough, right? My guess, rough guess for the UFOs as aliens
link |
scenario, at least some of them, is the priors roughly one in a thousand,
link |
much higher than the usual murder trial, plenty high enough that strong physical evidence could
link |
put you over the top to think it's more likely than not. But I'm not an expert on that physical
link |
evidence. I'm going to leave that part to someone else. I'm going to say the prior is pretty high.
link |
This isn't a crazy scenario. So then I can elaborate on where my prior comes from.
link |
What scenario could make most sense of this data? My scenario to make sense has two main parts.
link |
First is panspermia siblings. So panspermia is the process by which life might have arrived on
link |
earth from elsewhere. And a plausible time for that, I mean, it would have to happen very early
link |
in earth's history because we see life early in history. And a plausible time could have been
link |
during the stellar nursery where the sun was born with many other stars in the same close proximity
link |
with lots of rocks flying around, able to move things from one place to another.
link |
If a rock with life on it from some rock with planet with life came into that stellar nursery,
link |
it plausibly could have seeded many planets in that stellar nursery all at the same time. They're
link |
all born at the same time in the same place, pretty close to each other, lots of rocks flying
link |
around. So a panspermia scenario would then create siblings, i.e. there would be, say, a few thousand
link |
other planets out there. So after the nursery forms, it drifts, it separates, they drift apart.
link |
And so out there in the galaxy, there would now be a bunch of other stars all formed at the same
link |
time. And we can actually spot them in terms of their spectrum. And they would have then started
link |
on the same path of life as we did with that life being seeded, but they would move at different
link |
rates. And most likely, most of them would never reach an advanced level before the deadline. But
link |
maybe one other did, and maybe it did before us. So if they did, they could know all this,
link |
and they could go searching for their siblings. That is, they could look in the sky for the other
link |
stars that match the spectrum that matches the spectrum that came from this nursery.
link |
They could identify their sibling stars in the galaxy, the thousand of them. And those would be
link |
of special interest to them because they would think, well, life might be on those. And they
link |
could go looking for them. Can we just, such a brilliant mathematical, philosophical, physical,
link |
biological idea of panspermia siblings, because we all kind of started at a similar time
link |
in this local pocket of the universe. And so that changes a lot of the math.
link |
So that would create this correlation between when advanced life might appear,
link |
no longer just random independent spaces and space time. There'd be this cluster, perhaps.
link |
And that allows interaction between non grabby alien civilizations, like kind of
link |
primitive alien civilizations, like us with others. And they might be a little bit ahead.
link |
That's so fascinating.
link |
They would probably be a lot ahead. So the puzzle is, if they happened before us,
link |
they probably happened hundreds of millions of years before us.
link |
But less than a billion.
link |
Less than a billion, but still plenty of time that they could have become grabby and filled
link |
the galaxy and gone beyond. So the fact is, they chose not to become grabby. That would
link |
have to be the interpretation. If we have panspermia siblings...
link |
Plenty of time to become grabby, you said. So they should be gone.
link |
Yes, they had plenty of time and they chose not to.
link |
Are we sure about this? A hundred million years is enough.
link |
So I told you before that I said, within 10 million years, our descendants will become
link |
And they'll have that choice. Okay.
link |
Right? And so they, clearly more than 10 million years earlier than us, so they chose not to.
link |
But still go on vacation, look around, just not grabby.
link |
If they chose not to expand, that's going to have to be a rule they set to not allow
link |
any part of themselves to do it. If they let any little ship fly away with the ability
link |
to create a colony, the game's over. Then the universe becomes grabby from their origin
link |
with this one colony, right? So in order to prevent their civilization being grabby,
link |
they have to have a rule they enforce pretty strongly that no part of them can ever try
link |
Through a global authoritarian regime or through something that's internal to them,
link |
meaning it's part of the nature of life that it doesn't want...
link |
As like a political officer in the brain or whatever.
link |
Yes. There's something in human nature that prevents you from what or like alien nature
link |
that as you get more advanced, you become lazier and lazier in terms of exploration
link |
So I would say they would have to have enforced a rule against expanding and that rule would
link |
probably make them reluctant to let people leave very far. You know, any one vacation
link |
trip far away could risk an expansion from this vacation trip. So they would probably
link |
have a pretty tight lid on just allowing any travel out from their origin in order to
link |
enforce this rule. But then we also know, well, they would have chosen to come here.
link |
So clearly they made an exception from their general rule to say, okay, but an expedition
link |
to Earth, that should be allowed.
link |
It could be intentional exception or incompetent exception.
link |
But if incompetent, then they couldn't maintain this over 100 million years, this policy of
link |
not allowing any expansion. So we have to see they have successfully, they not just
link |
had a policy to try, they succeeded over 100 million years in preventing the expansion.
link |
That's a substantial competence.
link |
Let me think about this. So you don't think there could be a barrier in 100 million years,
link |
you don't think there could be a barrier to like technological barrier to becoming expansionary.
link |
Imagine the Europeans have tried to prevent anybody from leaving Europe to go to the new
link |
world. And imagine what it would have taken to make that happen over 100 million years.
link |
Yeah, it's impossible.
link |
They would have had to have very strict, you know, guards at the borders saying, no, you
link |
But just to clarify, you're not suggesting that's actually possible.
link |
I am suggesting it's possible.
link |
I don't know how you keep, in my silly human brain, maybe it's the brain that values freedom,
link |
but I don't know how you can keep, no matter how much force, no matter how much censorship
link |
or control or so on, I just don't know how you can keep people from exploring into the
link |
mysterious, into the unknown.
link |
You're thinking of people, we're talking aliens. So remember, there's a vast space
link |
of different possible social creatures they could have evolved from, different cultures
link |
they could be in, different kinds of threats. I mean, there are many things, as you talked
link |
about, that most of us would feel very reluctant to do.
link |
This isn't one of those.
link |
Okay, so how, if the UFO sightings represent alien visitors, how the heck are they getting
link |
here under the panspermia siblings?
link |
So panspermia siblings is one part of the scenario, which is that's where they came
link |
from. And from that, we can conclude they had this rule against expansion and they've
link |
successfully enforced that. That also creates a plausible agenda for why they would be here,
link |
that is to enforce that rule on us. That is, if we go out and expanding, then we have defeated
link |
the purpose of this rule they set up.
link |
Right? So they would be here to convince us to not expand.
link |
Convince in quotes.
link |
Right? Through various mechanisms. So obviously, one thing we conclude is they didn't just
link |
destroy us. That would have been completely possible, right? So the fact that they're
link |
here and we are not destroyed means that they chose not to destroy us. They have some degree
link |
of empathy or whatever their morals are that would make them reluctant to just destroy
link |
us. They would rather persuade us.
link |
Destroy their brethren. And so they may have been, there's a difference in arrival and
link |
observation. They may have been observing for a very long time.
link |
And they arrive to try to, not to try, I don't think to try to ensure that we don't become
link |
Which is because we can see that they did not, they must have enforced a rule against
link |
that and they are therefore here to, that's a plausible interpretation why they would
link |
risk this expedition when they clearly don't risk very many expeditions over this long
link |
period to allow this one exception because otherwise, if they don't, we may become grabby.
link |
And they could have just destroyed us, but they didn't.
link |
And they're closely monitoring the technological advancing of our civilization. Like what
link |
nuclear weapons is one thing that, all right, cool. That might have less to do with nuclear
link |
weapons and more with nuclear energy. Maybe they're monitoring fusion closely. Like how
link |
clever are these apes getting?
link |
So no doubt they have a button that if we get too uppity or risky, they can push the
link |
button and ensure that we don't expand. But they'd rather do it some other way. So now
link |
that's, that explains why they're here and why they aren't out there. But there's another
link |
thing that we need to explain. There's another key data we need to explain about UFOs if
link |
we're going to have a hypothesis that explains them. And this is something many people have
link |
noticed, which is they had two extreme options they could have chosen and didn't chose.
link |
They could have either just remained completely invisible. Clearly an advanced civilization
link |
could have been completely invisible. There's no reason they need to fly around and be
link |
noticed. They could just be in orbit and in dark satellites that are completely invisible
link |
to us watching whatever they want to watch. That would be well within their abilities.
link |
That's one thing they could have done. The other thing they could do is just show up
link |
and land on the White House lawn, as they say, and shake hands, like make themselves
link |
really obvious. They could have done either of those and they didn't do either of those.
link |
That's the next thing you need to explain about UFOs as aliens. Why would they take
link |
this intermediate approach, hanging out near the edge of visibility with somewhat impressive
link |
mechanisms, but not walking up and introducing themselves nor just being completely invisible?
link |
So, okay, a lot of questions there. So one, do you think it's obvious where the White
link |
House is or the White House lawn?
link |
Obvious where there are concentrations of humans that you could go up and introduce.
link |
But is humans the most interesting thing about Earth?
link |
Are you sure about this? Because...
link |
If they're worried about an expansion, then they would be worried about a civilization
link |
that could be capable of expansion. Obviously humans are the civilization on Earth that's
link |
by far the closest to being able to expand.
link |
I just don't know if aliens obviously see...obviously see humans, like the individual
link |
humans, like the meat vehicles, as the center of focus for observing a life on a planet.
link |
They're supposed to be really smart and advanced. Like, this shouldn't be that hard for them.
link |
But I think we're actually the dumb ones, because we think humans are the important
link |
things. But it could be our ideas. It could be something about our technologies.
link |
But that's mediated with us. It's correlated with us.
link |
No, we make it seem like it's mediated by us humans. But the focus for alien civilizations
link |
might be the AI systems or the technologies themselves. That might be the organism. Like,
link |
what humans are like...human is the food, the source of the organism that's under observation,
link |
So what they wanted to have close contact with was something that was closely near humans,
link |
then they would be contacting those. And we would just incidentally see, but we would still see.
link |
But don't you think that...isn't it possible, taking their perspective,
link |
isn't it possible that they would want to interact with some fundamental aspect that
link |
they're interested in without interfering with it? And that's actually a very...no
link |
matter how advanced you are, it's very difficult to do.
link |
But that's puzzling. So, I mean, the prototypical UFO observation is a shiny,
link |
big object in the sky that has very rapid acceleration and no apparent surfaces for
link |
using air to manipulate at speed. And the question is, why that? Again, if they just...
link |
For example, if they just wanted to talk to our computer systems, they could move some sort of
link |
like a little probe that connects to a wire and reads and sends bits there. They don't need a
link |
shiny thing flying in the sky.
link |
But don't you think they would be looking for the right way to communicate, the right
link |
language to communicate? Everything you just said, looking at the computer systems,
link |
I mean, that's not a trivial thing. Coming up with a signal that us humans would not freak out
link |
too much about, but also understand, might not be that trivial.
link |
Well, so the not freak out part is another interesting constraint. So again, I said,
link |
like the two obvious strategies are just to remain completely invisible and watch,
link |
which would be quite feasible, or to just directly interact, come out and be really
link |
very direct, right? I mean, there's big things that you can see around. There's big cities,
link |
there's aircraft carriers, there's lots of... If you want to just find a big thing and come
link |
right up to it and like tap it on the shoulder or whatever, that would be quite feasible,
link |
then they're not doing that. So my hypothesis is that one of the other questions there was,
link |
do they have a status hierarchy? And I think most animals on earth who are social animals
link |
who are social animals have status hierarchy, and they would reasonably presume that we have
link |
a status hierarchy. And...
link |
Take me to your leader.
link |
Well, I would say their strategy is to be impressive and sort of get us to see them
link |
at the top of our status hierarchy. That's how, for example, we domesticate dogs, right?
link |
We convince dogs we're the leader of their pack, right? And we domesticate many animals that way,
link |
but as we just swap into the top of their status hierarchy and we say,
link |
we're your top status animal, so you should do what we say, you should follow our lead.
link |
So the idea that would be, they are going to get us to do what they want by being top status.
link |
You know, all through history, kings and emperors, et cetera, have tried to impress their citizens
link |
and other people by having the bigger palace, the bigger parade, the bigger crown and
link |
diamonds, right? Whatever, maybe building a bigger pyramid, et cetera. It's a very well
link |
established trend to just be high status by being more impressive than the rest.
link |
To push back when there's an order of several orders of magnitude of power differential,
link |
asymmetry of power, I feel like that status hierarchy no longer applies. It's like memetic
link |
theory. It's like...
link |
Most emperors are several orders of magnitude more powerful than any one member of their empire.
link |
Let's increase that by even more. So like if I'm interacting with ants,
link |
I no longer feel like I need to establish my power with ants. I actually want to lower myself
link |
to the ants. I want to become the lowest possible ant so that they would welcome me.
link |
So I'm less concerned about them worshiping me. I'm more concerned about them welcoming me.
link |
It is important that you be nonthreatening and that you be local. So I think
link |
for example, if the aliens had done something really big in the sky, 100 light years away,
link |
that would be there, not here. And that could seem threatening. So I think their strategy to
link |
be the high status would have to be to be visible, but to be here and nonthreatening.
link |
I just don't know if it's obvious how to do that. Take your own perspective. You see a planet
link |
with relatively intelligent complex structures being formed, life forms. You could see this
link |
under in Titan or something like that, Europa. You start to see not just primitive bacterial
link |
life, but multicellular life. And it seems to form some very complicated cellular colonies,
link |
structures that they're dynamic. There's a lot of stuff going on. Some gigantic cellular automata
link |
type of construct. How do you make yourself known to them in an impressive fashion
link |
without destroying it? We know how to destroy potentially.
link |
Right. So if you go touch stuff, you're likely to hurt it, right? There's a good risk of hurting
link |
something by getting too close and touching it and interacting, right?
link |
Yeah, like landing on a White House lawn.
link |
Right. So the claim is that their current strategy of hanging out at the periphery of
link |
our vision and just being very clearly physically impressive with very clear physically impressive
link |
abilities is at least a plausible strategy they might use to impress us and convince us sort of
link |
we're at the top of their status hierarchy. And I would say if they came closer, not only would
link |
they risk hurting us in ways that they couldn't really understand, but more plausibly, they would
link |
reveal things about themselves we would hate. So if you look at how we treat other civilizations
link |
on Earth and other people, we are generally interested in foreigners and people from other
link |
plant lands. And we were generally interested in their varying cult customs, et cetera,
link |
until we find out that they do something that violates our moral norms and then we hate them.
link |
And these are aliens for God's sakes, right? There's just going to be something about them
link |
that we hate. They eat babies. Who knows what it is? Something they don't think is offensive,
link |
but that they think we might find. And so they would be risking a lot by revealing a lot about
link |
themselves. We would find something we hated. Interesting. But do you resonate at all with
link |
memetic theory where like, we only feel this way about things that are very close to us.
link |
So aliens are sufficiently different to where we'll be like, fascinated, terrified or fascinated,
link |
but not like. Right, but if they want to be at the top of our status hierarchy to get us to
link |
follow them, they can't be too distant. They have to be close enough that we would see them that
link |
way. But pretend to be close enough. Right. And not reveal much that mystery that old Clint Eastwood
link |
cowboy. I mean, we're clever enough that we can figure out their agenda. That is just from the
link |
fact that we're here. If we see that they're here, we can figure out, Oh, they want us not to expand
link |
and look, they are this huge power and they're very impressive. So, and a lot of us don't want
link |
to expand. So that could easily tip us over the edge toward we already wanted to not expand. We
link |
already wanted to be able to regulate and have a central community. And here are these very advanced
link |
smart aliens who have survived for a hundred million years and they're telling us not to expand
link |
either. This is brilliant. I love this so much. Uh, the, the, so returning to panspermia siblings,
link |
just to clarify one thing in that framework, how would, who originated, who planted it?
link |
Would it be a grabby alien civilization that planted the siblings or no? The simple scenario
link |
is that life started on some other planet billions of years ago and it went through part of the
link |
stages of evolution to advance life, but not all the way to advance life. And then some rock hit
link |
it, grabbed a piece of it on the rock and that rock drifted for maybe in a million years until
link |
it happened to prong the stellar nursery where it then seeded many stars. And something about that
link |
life without being super advanced, it was nevertheless resilient to the harsh conditions
link |
of space. There's some graphs that I've been impressed by that show sort of the level of
link |
genetic information in various kinds of life on the history of earth. And basically we are now
link |
more complex than the earlier life, but the earlier life was still pretty complex. And so if
link |
you actually project this log graph in history, it looks like it was many billions of years ago
link |
when you get down to zero. So like plausible, you could say there was just a lot of evolution that
link |
had to happen before you to get to the simplest life we've ever seen in history of life on earth
link |
was still pretty damn complicated. Okay. And so that race, that's always been this puzzle. How
link |
could life get to this enormously complicated level in the short period it seems to at the
link |
beginning of earth history. So where, you know, it's only 300 million years at most when it
link |
appeared. And then it was really complicated at that point. So panspermia allows you to
link |
explain that complexity by saying, well, it's been another 5 billion years on another planet
link |
going through lots of earlier stages where it was working its way up to the level of
link |
complexity you see at the beginning of earth. We'll try to talk about other ideas of the
link |
origin of life, but let me return to UFO sightings. Is there other explanations that are possible
link |
outside of panspermia siblings that can explain no grabby aliens in the sky and yet alien arrival
link |
on earth? Well, the other categories of explanations that most people will use is, well,
link |
first of all, just mistakes, like, you know, you're, you're, you're confusing something
link |
ordinary for something mysterious, right? Or some sort of secret organization, like our
link |
government is secretly messing with us and trying to do a, you know, a false flag ops
link |
or whatever, right? You know, they're trying to convince the Russians or the Chinese that
link |
there might be aliens and scare them into not attacking or something, right? Because
link |
if you, you know, the history of World War II, say the US government did all these big
link |
fake operations where they were faking a lot of big things in order to mess with people.
link |
So that's a possibility. The government has been lying and, you know, faking things and
link |
paying people to lie about what they saw, et cetera. That's a plausible set of explanations
link |
for the range of sightings seen. And another explanation people offer is some other hidden
link |
organization on earth or some, you know, secret organization somewhere that has much more
link |
advanced capabilities than anybody's given a credit for, for some reason it's been keeping
link |
secret. I mean, they all sound somewhat implausible, but again, we're looking for maybe,
link |
you know, one in a thousand sort of priors. Question is, you know, could, could they be
link |
in that level of plausibility? Can we just linger on this? So you, first of all, you've written,
link |
talked about, thought about so many different topics. You're an incredible mind. And I just
link |
thank you for sitting down today. I'm almost like at a loss of which place we explore,
link |
but let me on this topic, ask about conspiracy theories because you've written about institutions
link |
authorities. What, this is a bit of a therapy session, but what do we make of conspiracy
link |
theories? The phrase itself is pushing you in a direction, right? So clearly in history,
link |
we've had many large coordinated keepings of secrets, right? Say the Manhattan project,
link |
right? And there was hundreds of thousands of people working on that over many years,
link |
but they kept it a secret, right? Clearly many large military operations have kept things secrets
link |
over, you know, even decades with many thousands of people involved. So clearly it's possible to
link |
keep some things secret over time periods. You know, but the more people you involve and the
link |
more time you are assuming and the more, the less centralized an organization or the less
link |
discipline they have, the harder it gets to believe. But we're just trying to calibrate
link |
basically in our minds, which kind of secrets can be kept by which groups over what time periods
link |
for what purposes, right? But let me, I don't have enough data. So I'm somebody, I, you know,
link |
I hang out with people and I love people. I love all things really. And I just, I think that most
link |
people, even the assholes have the capacity to be good and they're beautiful and I enjoy them.
link |
So the kind of data, my brain, whatever the chemistry of my brain is that sees the beautiful
link |
in things is maybe collecting a subset of data that doesn't allow me to intuit the competence
link |
that humans are able to achieve in constructing a conspiracy theory. So for example, one thing
link |
that people often talk about is like intelligence agencies, this like broad thing. They say the CIA,
link |
the FSB, the different, the British intelligence. I've fortunate or unfortunate enough, never gotten
link |
the chance that I know of to talk to any member of those intelligence agencies nor like take a
link |
peek behind the curtain or the first curtain. I don't know how many levels of curtains there are.
link |
And so I don't, I can't intuit my interactions with government. I was funded by DOD and DARPA
link |
and I've interacted, been to the Pentagon, like with all due respect to my friends, lovely friends
link |
in government. And there are a lot of incredible people, but there is a very giant bureaucracy
link |
that sometimes suffocates the ingenuity of the human spirit is one way I can put it. Meaning
link |
they are, I just, it's difficult for me to imagine extreme competence at a scale of hundreds or
link |
thousands human beings. Now that doesn't mean that's my very anecdotal data of the situation.
link |
And so I try to build up my intuition about centralized system of government, how much
link |
conspiracy is possible, how much the intelligence agencies or some other source can generate
link |
sufficiently robust propaganda that controls the populace. If you look at World War II, as you
link |
mentioned, there've been extremely powerful propaganda machines on the Nazi, on the side of
link |
Nazi Germany, on the side of the Soviet Union, on the side of the United States and all these different
link |
mechanisms. Sometimes they control the free press through social pressures. Sometimes they control
link |
the press through the threat of violence, as you do in authoritarian regimes. Sometimes it's like
link |
deliberately the dictator, like writing the news, the headlines and literally announcing it. And
link |
something about human psychology forces you to embrace the narrative and believe the narrative.
link |
And at scale that becomes reality when the initial spark was just the propaganda thought in a single
link |
individual's mind. So I can't necessarily intuit of what's possible, but I'm skeptical of the power
link |
of human institutions to construct conspiracy theories that cause suffering at scale, especially
link |
in this modern age when information is becoming more and more accessible by the populace. Anyway,
link |
that's the, I don't know if you can elucidate for us.
link |
It's called suffering at scale, but of course, say during wartime, the people who are managing
link |
the various conspiracies like D Day or Manhattan Project, they thought that their conspiracy was
link |
avoiding harm rather than causing harm. So if you can get a lot of people to think that supporting
link |
the conspiracy is helpful, then a lot more might do that. And there's just a lot of things that
link |
people just don't want to see. So if you can make your conspiracy the sort of thing that people
link |
wouldn't want to talk about anyway, even if they knew about it, you're most of the way there.
link |
So I have learned many over the years, many things that most ordinary people would never want to
link |
hear, many things that most ordinary people should be interested in, but somehow don't know,
link |
even though the data has been very widespread. So I have this book, The Elephant and the Brain,
link |
and one of the chapters is there on medicine. And basically, most people seem ignorant of the very
link |
basic fact that when we do randomized trials where we give some people more medicine than others,
link |
the people who get more medicine are not healthier. Just overall, in general, just like
link |
induce somebody to get more medicine because you just give them more budget to buy medicine, say.
link |
And not a specific medicine, just the whole category. And you would think that would be
link |
something most people should know about medicine. You might even think that would be a conspiracy
link |
theory to think that would be hidden, but in fact, most people never learn that fact.
link |
So just to clarify, just a general high level statement, the more medicine you take,
link |
the less healthy you are.
link |
Randomized experiments don't find that fact. Do not find that more medicine makes you more healthy.
link |
There's just no connection. In randomized experiments, there's no relationship between
link |
more medicine and being healthier.
link |
So it's not a negative relationship, but it's just no relationship.
link |
And so the conspiracy theory would say that the businesses that sell you medicine don't want you
link |
to know that fact. And then you're saying that there's also part of this is that people just
link |
don't want to know.
link |
They just don't want to know. And so they don't learn this. So I've lived in the Washington area
link |
for several decades now, reading the Washington Post regularly. Every week there was a special
link |
section on health and medicine. It never was mentioned in that section of the paper
link |
in all the 20 years I read that.
link |
So do you think there is some truth to this caricatured blue pill, red pill,
link |
where most people don't want to know the truth?
link |
There are many things about which people don't want to know certain kinds of truths.
link |
Yeah. That is bad looking truths, truths that discouraging, truths that sort of take away the
link |
justification for things they feel passionate about.
link |
Do you think that's a bad aspect of human nature? That's something we should try to overcome?
link |
Well, as we discussed, my first priority is to just tell people about it, to do the analysis
link |
and the cold facts of what's actually happening, and then to try to be careful about how we can
link |
improve. So our book, The Elephant in the Rain, coauthored with Kevin Simler, is about how we
link |
hide motives in everyday life. And our first priority there is just to explain to you what are
link |
the things that you are not looking at that you have reluctant to look at. And many people try
link |
to take that book as a self help book where they're trying to improve themselves and make
link |
sure they look at more things. And that often goes badly because it's harder to actually do
link |
that than you think. But we at least want you to know that this truth is available if you want
link |
to learn about it.
link |
It's the Nietzsche, if you gaze long into the abyss, the abyss gazes into you. Let's talk about
link |
this elephant in the brain. Amazing book. The elephant in the room is, quote, an important
link |
issue that people are reluctant to acknowledge or address a social taboo. The elephant in the brain
link |
is an important but unacknowledged feature of how our mind works, an introspective taboo.
link |
You describe selfishness and self deception as the core or some of the core elephants,
link |
some of the elephants, elephant offspring in the brain. Selfishness and self deception.
link |
Can you explain, can you explain why these are the taboos in our brain that we
link |
don't want to acknowledge to ourselves?
link |
Your conscious mind, the one that's listening to me that I'm talking to at the moment, you like
link |
to think of yourself as the president or king of your mind, ruling over all that you see,
link |
issuing commands that immediately obeyed. You are instead better understood as the press secretary
link |
of your brain. You don't make decisions. You justify them to an audience. That's what your
link |
conscious mind is for. You watch what you're doing and you try to come up with stories that explain
link |
what you're doing so that you can avoid accusations of violating norms. So humans compared to most
link |
other animals have norms, and this allows us to manage larger groups with our morals and norms
link |
about what we should or shouldn't be doing. This is so important to us that we needed to be
link |
constantly watching what we were doing in order to make sure we had a good story to avoid norm
link |
violations. So many norms are about motives. So if I hit you on purpose, that's a big violation.
link |
If I hit you accidentally, that's okay. I need to be able to explain why it was an accident
link |
and not on purpose.
link |
So where does that need come from for your own self preservation?
link |
Right. So humans have norms and we have the norm that if we see anybody violating a norm,
link |
we need to tell other people and then coordinate to make them stop and punish them for violating.
link |
So such benefits are strong enough and severe enough that we each want to avoid being successfully
link |
accused of violating norms. So for example, hitting someone on purpose is a big clear norm
link |
violation. If we do it consistently, we may be thrown out of the group and that would mean we
link |
would die. Okay. So we need to be able to convince people we are not going around hitting people on
link |
purpose. If somebody happens to be at the other end of our fist and their face connects, that was
link |
an accident and we need to be able to explain that. And similarly for many other norms humans
link |
have, we are serious about these norms and we don't want people to violate. We find them
link |
violating, we're going to accuse them. But many norms have a motive component. And so we are
link |
trying to explain ourselves and make sure we have a good motive story about everything we do,
link |
which is why we're constantly trying to explain what we're doing. And that's what your conscious
link |
mind is doing. It is trying to make sure you've got a good motive story for everything you're
link |
doing. And that's why you don't know why you really do things. What you know is what the good
link |
story is about why you've been doing things. And that's the self deception. And you're saying that
link |
there is a machine, the actual dictator is selfish. And then you're just the press secretary who's
link |
desperately doesn't want to get fired and is justifying all of the decisions of the dictator.
link |
And that's the self deception.
link |
Right. Now, most people actually are willing to believe that this is true in the abstract. So
link |
our book has been classified as psychology and it was reviewed by psychologists. And the basic
link |
way that psychology referees and reviewers responded is to say, this is well known. Most
link |
people accept that there's a fair bit of self deception.
link |
But they don't want to accept it about themselves.
link |
Well, they don't want to accept it about the particular topics that we talk about. So people
link |
accept the idea in the abstract that they might be self deceived or that they might not be honest
link |
about various things. But that hasn't penetrated into the literatures where people are explaining
link |
particular things like why we go to school, why we go to the doctor, why we vote, et cetera. So
link |
our book is mainly about 10 areas of life and explaining about in each area what our actual
link |
motives there are. And people who study those things have not admitted that hidden motives are
link |
explaining those particular areas.
link |
So they haven't taken the leap from theoretical psychology to actual public policy.
link |
And economics and all that kind of stuff. Well, let me just linger on this and bring up my old
link |
friends Zingman Freud and Carl Jung. So how vast is this landscape of the unconscious mind,
link |
the power and the scope of the dictator? Is it only dark there? Is it some light? Is there some
link |
The vast majority of what's happening in your head, you're unaware of. So in a literal sense,
link |
the unconscious, the aspects of your mind that you're not conscious of is the overwhelming
link |
majority. But that's just true in a literal engineering sense. Your mind is doing lots of
link |
low level things, and you just can't be consciously aware of all that low level stuff. But there's
link |
plenty of room there for lots of things you're not aware of.
link |
But can we try to shine a light at the things we're unaware of specifically? Now, again,
link |
staying with the philosophical psychology side for a moment, can you shine a light in the Jungian
link |
shadow? What's going on there? What is this machine like? What level of thoughts are happening
link |
there? Is it something that we can even interpret? If we somehow could visualize it, is it something
link |
that's human interpretable? Or is it just a kind of chaos of monitoring different systems in the
link |
body, making sure you're happy, making sure you're fed all those kind of basic forces that form
link |
abstractions on top of each other, and they're not introspective at all?
link |
We humans are social creatures. Plausibly being social is the main reason we have these unusually
link |
large brains. Therefore, most of our brain is devoted to being social. And so the things we are
link |
very obsessed with and constantly paying attention to are, how do I look to others? What would others
link |
think of me if they knew these various things they might learn about me?
link |
So that's close to being fundamental to what it means to be human, is caring what others think.
link |
Right. To be trying to present a story that would be okay for what others think. But we're very
link |
constantly thinking, what do other people think?
link |
So let me ask you this question then about you, Robin Hansen, who in many places, sometimes for
link |
fun, sometimes as a basic statement of principle, likes to disagree with what the majority of people
link |
think. So how do you explain, how are you self deceiving yourself in this task? And how are you
link |
being self, like, why is the dictator manipulating you inside your head to be so critical? Like,
link |
there's norms. Why do you want to stand out in this way? Why do you want to challenge the
link |
norms in this way?
link |
Almost by definition, I can't tell you what I'm deceiving myself about. But the more practical
link |
strategy that's quite feasible is to ask about what are typical things that most people deceive
link |
themselves about, and then to own up to those particular things.
link |
Sure. What's a good one?
link |
So for example, I can very much acknowledge that I would like to be well thought of,
link |
that I would be seeking attention and glory and praise from my intellectual work, and that that
link |
would be a major agenda driving my intellectual attempts. So if there were topics that other
link |
people would find less interesting, I might be less interested in those for that reason,
link |
for example. I might want to find topics where other people are interested, and I might want to
link |
go for the glory of finding a big insight rather than a small one, and maybe one that was
link |
especially surprising. That's also, of course, consistent with some more ideal concept of what
link |
an intellectual should be. But most intellectuals are relatively risk averse. They are in some
link |
local intellectual tradition, and they are adding to that, and they are staying conforming to the
link |
sort of usual assumptions and usual accepted beliefs and practices of a particular area
link |
so that they can be accepted in that area and treated as part of the community. But you might
link |
think for the purpose of the larger intellectual project of understanding the world better,
link |
people should be less eager to just add a little bit to some tradition, and they should be looking
link |
for what's neglected between the major traditions and major questions. They should be looking for
link |
assumptions maybe we're making that are wrong. They should be looking at things that are very
link |
surprising, things that you would have thought a priori unlikely that once you are convinced of it,
link |
you find that to be very important and a big update. So you could say that one motivation
link |
I might have is less motivated to be sort of comfortably accepted into some particular
link |
intellectual community and more willing to just go for these more fundamental long shots that should
link |
be very important if you could find them.
link |
Which would, if you can find them, would get you appreciated across a larger number of people
link |
across the longer time span of history. So like maybe the small local community will say,
link |
you suck, you must conform. But the larger community will see the brilliance of you
link |
breaking out of the cage of the small conformity into a larger cage. There's always a bigger cage
link |
and then you'll be remembered by more. Yeah. Also that explains your choice of colorful shirt that
link |
looks great in a black background. So you definitely stand out.
link |
Right. Now, of course, you could say, well, you could get all this attention by making false
link |
claims of dramatic improvement. And then wouldn't that be much easier than actually working through
link |
all the details to make true claims?
link |
Why not? Let me ask the press secretary. Why not? So of course you spoke several times about how
link |
much you value truth and the pursuit of truth. That's a very nice narrative. Hitler and Stalin
link |
also talked about the value of truth. Do you worry when you introspect as broadly as all humans
link |
might that it becomes a drug, this being a martyr, being the person who points out that the emperor
link |
wears no clothes, even when the emperor is obviously dressed, just to be the person who points
link |
out that the emperor is wearing no clothes. Do you think about that?
link |
So I think the standards you hold yourself to are dependent on the audience you have in mind.
link |
So if you think of your audience as relatively easily fooled or relatively gullible, then you
link |
won't bother to generate more complicated, deep, you know, arguments and structures and evidence
link |
to persuade somebody who has higher standards because why bother? You don't have to worry
link |
about it. Why bother? You can get away with something much easier. And of course, if you are,
link |
say, a salesperson, you know, you make money on sales, then you don't need to convince the top few
link |
percent of the most sharp customers. You can just go for the bottom 60 percent of the most gullible
link |
customers and make plenty of sales, right? So I think intellectuals have to vary. One of the main
link |
ways intellectuals vary is in who is their audience in their mind? Who are they trying to
link |
impress? Is it the people down the hall? Is it the people who are reading their Twitter feed? Is it
link |
their parents? Is it their high school teacher? Or is it Einstein and Freud and Socrates, right?
link |
So I think those of us who are especially arrogant, especially think that we're really
link |
big shot or have a chance at being a really big shot, we were naturally going to pick the
link |
big shot audience that we can. We're going to be trying to impress Socrates and Einstein.
link |
Is that why you hang out with Tyler Cohen a lot and try to convince him yourself?
link |
And you might think, you know, from the point of view of just making money or having sex or
link |
other sorts of things, this is misdirected energy, right? Trying to impress the very
link |
most highest quality minds. That's such a small sample and they can't do that much for you anyway.
link |
Yeah. So I might well have had more, you know, ordinary success in life,
link |
be more popular, invited to more parties, make more money if I had targeted a lower tier
link |
set of intellectuals with the standards they have. But for some reason I decided early on
link |
that Einstein was my audience or people like him and I was going to impress them.
link |
Yeah. I mean, you pick your set of motivations, you know, convincing,
link |
impressing Tyler Cohen is not going to help you get laid. Trust me, I tried. All right.
link |
What are some notable sort of effects of the elephant in the brain in everyday life? So you
link |
mentioned when we try to apply that to economics, to public policy. So when we think about medicine,
link |
education, all those kinds of things, what are some things that we're...
link |
The key thing is medicine is much less useful health wise than you think. So, you know,
link |
if you were focused on your health, you would care a lot less about it. And if you were focused
link |
on other people's health, you would also care a lot less about it. But if medicine is, as we
link |
suggest, more about showing that you care and let other people showing that they care about you,
link |
then a lot of priority on medicine can make sense. So that was our very earliest discussion
link |
in the podcast. You were talking about what, you know, should you give people a lot of medicine
link |
when it's not very effective? And then the answer then is, well, if that's the way that you show
link |
that you care about them and you really want them to know you care, then maybe that's what
link |
you need to do if you can't find a cheaper, more effective substitute. So if we actually just pause
link |
on that for a little bit, how do we start to untangle the full set of self deception happening
link |
in the space of medicine? So we have a method that we use in our book that is what I recommend
link |
for people to use in all these sorts of topics. The straightforward method is first, don't look
link |
at yourself. Look at other people, look at broad patterns of behavior and other people, and then
link |
ask, what are the various theories we could have to explain these patterns of behavior? And then
link |
just do the simple matching, which theory better matches the behavior they have. And the last step
link |
is to assume that's true of you too. Don't assume you're an exception. If you happen to be an
link |
exception, that won't go so well, but nevertheless, on average, you aren't very well positioned to
link |
judge if you're an exception. So look at what other people do, explain what other people do,
link |
and assume that's you too. But also in the case of medicine, there's several parties to consider.
link |
So there's the individual person that's receiving the medicine. There's the doctors that are
link |
prescribing the medicine. There's drug companies that are selling drugs. There are governments that
link |
have regulations that are lobbyists. So you can build up a network of categories of humans in this
link |
and they each play their role. So how do you introspect the sort of analyze the system at a
link |
system scale versus at the individual scale? So it turns out that in general, it's usually much
link |
easier to explain producer behavior than consumer behavior. That is, the drug companies or the
link |
doctors have relatively clear incentives to give the customers whatever they want. And so many say
link |
governments in democratic countries have the incentive to give the voters what they want.
link |
So that focuses your attention on the patient and the voter in this equation and saying,
link |
what do they want? They would be driving the rest of the system.
link |
Whatever they want, the other parties are willing to give them in order to get paid. So now we're
link |
looking for puzzles in patient and voter behavior. What are they choosing? And why do they choose
link |
that? And how much exactly? And then we can explain that potentially again, returning to
link |
the producer, but the producer being incentivized to manipulate the decision making processes of
link |
the voter and the consumer. Now, in almost every industry, producers are in general happy to lie
link |
and exaggerate in order to get more customers. This is true of auto repair as much as human
link |
body repair and medicine. So the differences between these industries can't be explained
link |
by the willingness of the producers to give customers what they want or to do various things
link |
that we have to again, go to the customers. Why are customers treating body repair different
link |
than auto repair? Yeah, and that potentially requires a lot of thinking, a lot of data
link |
collection and potentially looking at historical data too, because things don't just happen
link |
overnight. Over time, there's trends. In principle it does, but actually it's a lot,
link |
actually easier than you might think. I think the biggest limitation is just the willingness
link |
to consider alternative hypotheses. So many of the patterns that you need to rely on are actually
link |
pretty obvious, simple patterns. You just have to notice them and ask yourself, how can I explain
link |
those? Often you don't need to look at the most subtle, most difficult statistical evidence that
link |
might be out there. The simplest patterns are often enough. All right. So there's a fundamental
link |
statement about self deception in the book. There's the application of that, like we just did
link |
in medicine. Can you steel man the argument that many of the foundational ideas in the book are
link |
wrong? Meaning there's two that you just made, which is it can be a lot simpler than it looks.
link |
Can you steel man the case that it's, case by case, it's always super complicated. Like it's
link |
a complex system. It's very difficult to have a simple model about. It's very difficult to
link |
introspect. And the other one is that the human brain isn't, not just about self deception. That
link |
there's a lot of, there's a lot of motivations at play and we are able to really introspect our own
link |
mind. And like what, what's on the surface of the conscious is actually quite a good representation
link |
of what's going on in the brain. And you're not deceiving yourself. You're able to actually
link |
arrive to deeply think about where your mind stands and what you think about the world. And
link |
it's less about impressing people and more about being a free thinking individual.
link |
So when a child tries to explain why they don't have their homework assignment, they are sometimes
link |
inclined to say, the dog ate my homework. They almost never say the dragon ate my homework.
link |
The reason is the dragon is a completely implausible explanation. Almost always when we
link |
make excuses for things, we choose things that are at least in some degree plausible. It could
link |
perhaps have happened. That's an obstacle for any explanation of a hidden motive or a hidden
link |
feature of human behavior. If people are pretending one thing while really doing another,
link |
they're usually going to pick as a pretense something that's somewhat plausible. That's
link |
going to be an obstacle to proving that hypothesis if you are focused on sort of the local data that
link |
a person would typically have if they were challenged. So if you're just looking at one
link |
kid and his lack of homework, maybe you can't tell whether his dog ate his homework or not.
link |
If you happen to know he doesn't have a dog, you might have more confidence. You will need to have
link |
a wider range of evidence than a typical person would when they're encountering that actual excuse
link |
in order to see past the excuse. That will just be a general feature of it. So if I say,
link |
there's this usual story about where we go to the doctor and then there's this other explanation,
link |
it'll be true that you'll have to look at wider data in order to see that because people don't
link |
usually offer excuses unless in the local context of their excuse, they can get away with it. That
link |
is, it's hard to tell, right? So in the case of medicine, I have to point you to sort of larger
link |
sets of data. But in many areas of academia, including health economics, the researchers there
link |
also want to support the usual points of view. And so they will have selection effects in their
link |
publications and their analysis whereby they, if they're getting a result too much contrary to the
link |
usual point of view everybody wants to have, they will file drawer that paper or redo the analysis
link |
until they get an answer that's more to people's liking. So that means in the health economics
link |
literature, there are plenty of people who will claim that in fact, we have evidence that medicine
link |
is effective. And when I respond, I will have to point you to our most reliable evidence.
link |
And ask you to consider the possibility that the literature is biased in that when the evidence
link |
isn't as reliable, when they have more degrees of freedom in order to get the answer they want,
link |
they do tend to get the answer they want. But when we get to the kind of evidence that's much
link |
harder to mess with, that's where we will see the truth be more revealed. So with respect to
link |
medicine, we have millions of papers published in medicine over the years, most of which give the
link |
impression that medicine is useful. There's a small literature on randomized experiments of the
link |
aggregate effects of medicine, where there's maybe a few half dozen or so papers, where it would be
link |
the hardest to hide it because it's such a straightforward experiment done in a straightforward
link |
way that it's hard to manipulate. And that's where I will point you to.
link |
Manipulate. And that's where I will point you to, to show you that there's relatively
link |
little correlation between health and medicine. But even then, people could try to save the
link |
phenomenon and say, well, it's not hidden motives. It's just ignorance. They could say,
link |
for example, you know, medicine's complicated. Most people don't know the literature.
link |
Therefore, they can be excused for ignorance. They are just ignorantly assuming that medicine
link |
is effective. It's not that they have some other motive that they're trying to achieve.
link |
And then I will have to do, you know, as with a conspiracy theory analysis, I'm saying, well,
link |
like, how long has this misperception been going on? How consistently has it happened
link |
around the world and across time? And I would have to say, look, you know, if we're talking about,
link |
say, a recent new product, like Segway scooters or something, I could say not so many people have
link |
seen them or used them. Maybe they could be confused about their value. If we're talking
link |
about a product that's been around for thousands of years, used in roughly the same way all across
link |
the world, and we see the same pattern over and over again, this sort of ignorance mistake just
link |
doesn't work so well. It also is a question of how much of the self deception is prevalent versus
link |
foundational. Because there's a kind of implied thing where it's foundational to human nature
link |
versus just a common pitfall. This is a question I have. So, like, maybe human progress is made by
link |
people who don't fall into the self deception. It's a baser aspect of human nature, but then
link |
you escape it easily if you're motivated.
link |
The motivational hypotheses about the self deceptions are in terms of how it makes you
link |
look to the people around you. Again, the press secretary. So, the story would be, most people
link |
want to look good to the people around them. Therefore, most people present themselves in ways
link |
that help them look good to the people around them. That's sufficient to say there would be a
link |
lot of it. It doesn't need to be 100%, right? There's enough variety in people and in
link |
circumstances that sometimes taking a contrarian strategy can be in the interest of some minority
link |
of the people. So, I might, for example, say that that's a strategy I've taken. I've decided that
link |
being contrarian on these things could be winning for me in that there's a room for a small number
link |
of people like me who have these sort of messages who can then get more attention, even if there's
link |
not room for most people to do that. And that can be explaining sort of the variety, right?
link |
Similarly, you might say, look, just look at the most obvious things. Most people would like to
link |
look good, right? In the sense of physically, just you look good right now. You're wearing a nice
link |
suit, you have a haircut, you shaved, right? So, and we cut my own hair by the way. Okay.
link |
Well, that's all the more impressive. That's a counter argument for your claim.
link |
So, clearly, if we look at most people and their physical appearance, clearly, most people are
link |
trying to look somewhat nice, right? They shower, they shave, they comb their hair,
link |
but we certainly see some people around who are not trying to look so nice, right? Is that a
link |
big challenge, the hypothesis that people want to look nice? Not that much, right? We can see
link |
in those particular people's context, more particular reasons why they've chosen to be
link |
an exception to the more general rule.
link |
So, the general rule does reveal something foundational generally.
link |
That's the way things work. Let me ask you, you wrote a blog post about the general rule,
link |
let me ask you, you wrote a blog post about the accuracy of authorities since we're talking
link |
about this, especially in medicine. Just looking around us, especially during this time of the
link |
pandemic, there's been a growing distrust of authorities, of institutions, even the institution
link |
of science itself. What are the pros and cons of authorities, would you say? So, what's nice
link |
about authorities? What's nice about institutions? And what are their pitfalls?
link |
One standard function of authority is as something you can defer to, respectively,
link |
without needing to seem too submissive or ignorant or, you know, gullible. That is,
link |
you know, when you're asking what should I act on or what beliefs should I act on,
link |
you might be worried if I chose something too contrarian, too weird, too speculative,
link |
that that would make me look bad. So, I would just choose something very conservative.
link |
So, maybe an authority lets you choose something a little less conservative because the authority
link |
is your authorization. The authority will let you do it. And you can say, and somebody says,
link |
why did you do that thing? And they say, the authority authorized it. The authority tells me,
link |
I should do this. Why aren't you doing it, right?
link |
So, the authority is often pushing for the conservative?
link |
Well, no, the authority can do more. I mean, so for example, we just think about,
link |
I don't know, in a pandemic even, right? You could just think, I'll just stay home and close
link |
all the doors or I'll just ignore it, right? You could just think of just some very simple
link |
strategy that might be defensible if there were no authorities, right? But authorities might be
link |
able to know more than that. They might be able to like look at some evidence, draw a more context
link |
dependent conclusion, declare it as the authority's opinion. And then other people might follow that
link |
and that could be better than doing nothing. So, what you mentioned, WHO, the world's most
link |
beloved organization. So, this is me speaking in general, WHO and CDC has been kind of,
link |
I, depending on degrees and details, just not behaving as I would have imagined in the best
link |
possible evolution of human civilization, authorities should act. They seem to have failed
link |
in some fundamental way in terms of leadership in a difficult time for our society. Can you say what
link |
are the pros and cons of this particular authority? So, again, if there were no authorities whatsoever,
link |
no accepted authorities, then people would sort of have to sort of randomly pick different local
link |
authorities who would conflict with each other. And then they'd be fighting each other about that,
link |
or just not believe anybody and just do some initial default action that you would always do
link |
without responding to context. So, the potential gain of an authority is that they could know more
link |
than just basic ignorance. And if people followed them, they could both be more informed than
link |
ignorance and all doing the same thing. So, they're each protected from being accused or
link |
complained about. That's the idea of an authority. That would be the good. What's the con of that?
link |
Okay. How does that go wrong? So, the con is that if you think of yourself as the authority and
link |
asking what's my best strategy as an authority, it's unfortunately not to be maximally informative.
link |
So, you might think the ideal authority would not just tell you more than ignorance, it would tell
link |
you as much as possible. Okay. It would give you as much detail as you could possibly listen to and
link |
manage to assimilate. And it would update that as frequently as possible or as frequently as you
link |
were able to listen and assimilate. And that would be the maximally informative authority. The problem
link |
is there's a conflict between being an authority or being seen as an authority and being maximally
link |
informative. That was the point of my blog post that you're pointing out to here. That is, if you
link |
look at it from their point of view, they won't long remain the perceived authority if they are
link |
too cautious, incautious about how they use that authority. And one of the ways to be incautious
link |
would be to be too informative. Okay. That's still in the pro column for me because you're talking
link |
about the tensions that are very data driven and very honest. And I would hope that authorities
link |
struggle with that. How much information to provide to people to maximize outcomes.
link |
Now I'm generally somebody that believes more information is better because I trust the
link |
intelligence of people. But I'd like to mention a bigger con on authorities, which is the human
link |
question. This comes back to a global government and so on. Is that, you know, there's humans that
link |
sit in chairs during meetings and those authorities, they have different titles. It's
link |
for humans form hierarchies. And sometimes those titles get to your head a little bit
link |
and you start to want to think, how do I preserve my control over this authority? As opposed to
link |
thinking through like, what is the mission of the authority? What is the mission of WHO and
link |
the other such organization? And how do I maximize the implementation of that mission? You start to
link |
think, well, I kind of like sitting in this big chair at the head of the table. I'd like to sit
link |
there for another few years or better yet, I want to be remembered as the person who in a time of
link |
crisis was at the head of this authority and did a lot of good things. So you stop trying to do good
link |
under what good means given the mission of the authority. And you start to try to carve a
link |
narrative, to manipulate the narrative. First in the meeting room, everybody around you, just a
link |
small little story you tell yourself, the new interns, the managers throughout the whole
link |
hierarchy of the company. Okay, once everybody in the company or in the organization believes this
link |
narrative, now you start to control the release of information, not because you're trying to
link |
maximize outcomes, but because you're trying to maximize the effectiveness of the narrative that
link |
you are truly a great representative of this authority in human history. And I just feel like
link |
those human forces whenever you have an authority, it starts getting to people's heads. One of the
link |
most, me as a scientist, one of the most disappointing things to see during the pandemic
link |
is the use of authority from colleagues of mine to roll their eyes, to dismiss other human beings
link |
just because they got a PhD, just because they're an assistant, associate, full faculty, just because
link |
they are deputy head of X organization, NIH, whatever the heck the organization is,
link |
just because they got an award of some kind and at a conference they won a best paper award seven
link |
years ago and then somebody shook their hand and gave them a medal, maybe it was a president
link |
and it's been 20, 30 years that people have been patting them on the back saying how special
link |
they are, especially when they're controlling money and getting sucked up to from other scientists
link |
who really want the money in a self deception kind of way, they don't actually really care
link |
about your performance and all of that gets to your head and no longer are you the authority
link |
that's trying to do good and lessen the suffering in the world, you become an authority that just
link |
wants to maximize, self preserve yourself in a sitting on a throne of power. So this is core to
link |
sort of what it is to be an economist. I'm a professor of economics. There you go with the
link |
authority again. No, it's about saying, we often have a situation where we see a world of behavior
link |
and then we see ways in which particular behaviors are not sort of maximally socially useful.
link |
And we have a variety of reactions to that. So one kind of reaction is to sort of morally
link |
blame each individual for not doing the maximally socially useful thing under perhaps the idea that
link |
people could be identified and shamed for that and maybe induced into doing the better thing if
link |
only enough people were calling them out on it, right? But another way to think about it is to
link |
think that people sit in institutions with certain stable institutional structures and that
link |
institutions create particular incentives for individuals and that individuals are typically
link |
doing whatever is in their local interest in the context of that institution.
link |
And then perhaps to less blame individuals for winning their local institutional game
link |
and more blaming the world for having the wrong institutions. So economists are often like
link |
wondering what other institutions we could have instead of the ones we have and which of them
link |
might promote better behavior. And this is a common thing we do all across human behavior is
link |
to think of what are the institutions we're in and what are the alternative variations we could
link |
imagine and then to say which institutions would be most productive. I would agree with you that
link |
our information institutions, that is the institutions by which we collect information
link |
and aggregate it and share it with people are especially broken in the sense of far from the
link |
ideal of what would be the most cost effective way to collect and share information. But then
link |
the challenge is to try to produce better institutions. And as an academic, I'm aware that
link |
academia is particularly broken in the sense that we give people incentives to do research that's
link |
not very interesting or important because basically they're being impressive. And we actually care
link |
more about whether academics are impressive than whether they're interesting or useful.
link |
And I can go happy to go into detail with lots of different known institutions and their known
link |
institutional failings, ways in which those institutions produce incentives that are
link |
mistaken. And that was the point of the post we started with talking about the authorities. If
link |
I need to be seen as an authority, that's at odds with my being informative and I might choose to be
link |
the authority instead of being informative because that's my institutional incentives.
link |
And if I may, I'd like to, given that beautiful picture of incentives and individuals that you
link |
just painted, let me just apologize for a couple of things. One, I often put too much blame on
link |
leaders of institutions versus the incentives that govern those institutions. And as a result of that,
link |
I've been, I believe too critical of Anthony Fauci, too emotional about my criticism of
link |
Anthony Fauci. And I'd like to apologize for that because I think there's a deep, there's deeper
link |
truths to think about. There's deeper incentives to think about. That said, I do sort of, I'm a
link |
romantic creature by nature. I romanticize Winston Churchill. When I think about Nazi Germany,
link |
I think about Hitler more than I do about the individual people of Nazi Germany. You think
link |
about leaders, you think about individuals, not necessarily the parameters, the incentives that
link |
govern the system that, because it's harder. It's harder to think through deeply about the models
link |
from which those individuals arise, but that's the right thing to do. So, but also I don't apologize
link |
for being emotional sometimes and being.
link |
I'm happy to blame the individual leaders in the sense that, you know, I might say, well,
link |
you should be trying to reform these institutions if you're just there to like get promoted and look
link |
good at being at the top. But maybe I can blame you for your motives and your priorities in there,
link |
but I can understand why the people at the top would be the people who are selected for having
link |
the priority of primarily trying to get to the top. I get that.
link |
Can I maybe ask you about particular universities? They've received, like science has received an
link |
increase in distrust overall as an institution, which breaks my heart because I think science is
link |
beautiful as a, not maybe not as an institution, but as one of the things, one of the journeys that
link |
humans have taken on. The other one is university. I think university is actually a place for me,
link |
at least in the way I see it, is a place of freedom of exploring ideas, scientific ideas,
link |
engineering ideas, more than corporate, more than a company, more than a lot of domains in life.
link |
They're, it's not just in its ideal, but it's in its implementation, a place where you can
link |
be a kid for your whole life and play with ideas. And I think with all the criticism that universities
link |
still not currently receive, I think they, I don't think that criticism is representative
link |
of universities. They focus on very anecdotal evidence of particular departments, particular
link |
people, but I still feel like there's a lot of place for freedom of thought, at least MIT,
link |
at least in the fields I care about, in a particular kind of science, a particular kind
link |
of technical fields, mathematics, computer science, physics, engineering, so robotics,
link |
artificial intelligence. This is a place where you get to be a kid. Yet there is bureaucracy that's
link |
rising up. There's like more rules. There's more meetings and there's more administration
link |
having like PowerPoint presentations, which to me, you should like be more of a renegade
link |
explorer of ideas and meetings destroy, they suffocate that radical thought that happens
link |
when you're an undergraduate student and you can do all kinds of wild things when you're
link |
a graduate student. Anyway, all that to say, you've thought about this aspect too. Is there
link |
something positive, insightful you could say about how we can make for better universities
link |
in the decades to come? This particular institution, how can we improve them?
link |
I hear that centuries ago, many scientists and intellectuals were aristocrats. They had time
link |
and could, if they chose, choose to be intellectuals. That's a feature of the combination
link |
that they had some source of resources that allowed them leisure and that the kind of
link |
competition they were faced in among aristocrats allowed that sort of a self indulgence or
link |
self pursuit, at least at some point in their lives. So the analogous observation is that
link |
university professors often have sort of the freedom and space to do a wide range of things.
link |
And I am certainly enjoying that as a tenured professor.
link |
You're a really, sorry to interrupt, a really good representative of that.
link |
Just the exploration you're doing, the depth of thought, like most people are afraid to do the
link |
kind of broad thinking that you're doing, which is great.
link |
The fact that that can happen is a combination of these two things analogously. One is that
link |
we have fierce competition to become a tenured professor, but then once you become tenured,
link |
we give you the freedom to do what you like. And that's a happenstance. It didn't have to
link |
be that way. And in many other walks of life, even though people have a lot of resources,
link |
et cetera, they don't have that kind of freedom set up. So I think we're kind of,
link |
I'm kind of lucky that tenure exists and that I'm enjoying it. But I can't be too enthusiastic
link |
about this unless I can approve of sort of the source of the resources that's paying for all
link |
this. So for the aristocrat, if you thought they stole it in war or something, you wouldn't be so
link |
pleased. Whereas if you thought they had earned it or their ancestors had earned this money that
link |
they were spending as an aristocrat, then you could be more okay with that. So for universities,
link |
I have to ask, where are the main sources of resources that are going to the universities and
link |
are they getting their money's worth? Are they getting a good value for that payment?
link |
So first of all, they're students. And the question is, are students getting good value
link |
for their education? And each person is getting value in the sense that they are identified and
link |
shown to be a more capable person, which is then worth more salary as an employee later.
link |
But there is a case for saying there's a big waste to the system because we aren't actually
link |
changing the students or educating them. We're more sorting them or labeling them. And that's
link |
a very expensive process to produce that outcome. And part of the expense is the freedom from tenure,
link |
I guess. So I feel like I can't be too proud of that because it's basically a tax on all these
link |
young students to pay this enormous amount of money in order to be labeled as better. Whereas I
link |
feel like we should be able to find cheaper ways of doing that. The other main customer is
link |
researcher patrons like the government or other foundations. And then the question is,
link |
are they getting their money worth out of the money they're paying for research to happen?
link |
And my analysis is they don't actually care about the research progress. They are mainly
link |
buying an affiliation with credentialed impressiveness on the part of the researchers.
link |
They mainly pay money to researchers who are impressive and have high, you know,
link |
impressive affiliations. And they don't really much care what research project happens as a result.
link |
Is that a cynical? So there's a deep truth to that cynical perspective. Is there
link |
a less cynical perspective that they do care about the long term investment into the progress
link |
of science and humanity? Well, they might personally care, but they're stuck in an equilibrium.
link |
Wherein they, basically most foundations like governments or research or, you know,
link |
the Ford Foundation, they are, the individuals there are rated based on the prestige they bring
link |
to that organization. And even if they might personally want to produce more intellectual
link |
progress, they are in a competitive game where they don't have tenure and they need to produce
link |
this prestige. And so once they give grant money to prestigious people, that is the thing that
link |
shows that they have achieved prestige for the organization. And that's what they need to do in
link |
order to retain their position. And you do hope that there's a correlation between prestige and
link |
actual competence. Of course, there is a correlation. The question is just, could we do
link |
this better some other way? I think it's almost, I think it's pretty clear we could. What is harder
link |
to do is move the world to a new equilibrium where we do that instead. What are the components
link |
of the better ways to do it? Is it money? So how, the sources of money and how the money is
link |
allocated to give the individual researchers freedom? Years ago I started studying this topic
link |
exactly because this was my issue and this was many decades ago now. And I spent a long time
link |
and my best guess still is prediction markets, betting markets. So if you as a research
link |
paper patron want to know the answer to a particular question, like what's the mass of
link |
the electron neutrino, then what you can do is just subsidize a betting market in that question.
link |
And that will induce more research into answering that question because the people who then
link |
answer that question can then make money in that betting market with the new information they gain.
link |
So that's a robust way to induce more information on a topic. If you want to induce an
link |
accomplishment, you can create prizes. And there's of course a long history of prizes to induce
link |
accomplishments. And we moved away from prizes, even though we once used them far more often than
link |
we did today. And there's a history to that. And for the customers who want to be affiliated with
link |
impressive academics, which is what most of the customers want, students, journalists, and patrons,
link |
I think there's a better way of doing that, which I just wrote about in my second most recent blog
link |
post. Can you explain? Sure. What we do today is we take sort of acceptance by other academics
link |
recently as our best indication of their deserved prestige. That is recent publications, recent
link |
job affiliation, institutional affiliations, recent invitations to speak, recent grants.
link |
We are today taking other impressive academics, recent choices to affiliate with them as our best
link |
guesstimate of their prestige. I would say we could do better by creating betting markets in what the
link |
distant future will judge to have been their deserved prestige looking back on them. I think
link |
most intellectuals, for example, think that if we looked back two centuries, say to intellectuals
link |
from two centuries ago, and tried to look in detail at their research and how it influenced
link |
future research and which path it was on, we could much more accurately judge their actual
link |
deserved prestige. That is who was actually on the right track, who actually helped, which will be
link |
different than what people at the time judged using the immediate indications at the time of
link |
which position they had or which publications they had or things like that. So in this way,
link |
if you think from the perspective of multiple centuries, you would higher prioritize true
link |
novelty, you would disregard the temporal proximity, like how recent the thing is,
link |
and you would think like, what is the brave, the bold, the big, a novel idea that this sense,
link |
and you would actually, you would be able to rate that because you could see the path
link |
with which ideas took, which things had dead ends, which led to what other followings. You could,
link |
looking back centuries later, have a much better estimate of who actually had what long term
link |
effects on intellectual progress. So my proposal is we actually pay people in several centuries to
link |
do this historical analysis. And we have prediction markets today where we buy and sell
link |
assets, which will later off pay off in terms of those final evaluations. So now we'll be inducing
link |
people today to make their best estimate of those things by actually looking at the details of
link |
people and setting the prices accordingly. So my proposal would be we rate people today on those
link |
prices today. So instead of looking at their list of publications or affiliations, you look at the
link |
actual price of assets that represent people's best guess of what the future will say about them.
link |
That's brilliant. So this concept of idea futures, can you elaborate what this would entail?
link |
I've been elaborating two versions of it here. So one is if there's a particular question,
link |
say the mass of the electron neutrino, and what you as a patron want to do is get an answer to
link |
that question, then what you would do is subsidize the betting market in that question under the
link |
assumption that eventually we'll just know the answer and we can pay off the bets that way.
link |
And that is a plausible assumption for many kinds of concrete intellectual questions like what's the
link |
mass of the electron neutrino. In this hypothetical world that you're constructing that may be a real
link |
world, do you mean literally financial? Yes. Literal. Very literal. Very cash. Very direct
link |
and literal. Yes. Or crypto. Well, crypto is money. Yes, sure. So the idea would be research labs
link |
would be for profit. They would have as their expense paying researchers to study things and
link |
then their profit would come from using the insights the researchers gains to trade in these
link |
financial markets. Just like hedge funds today make money by paying researchers to study firms
link |
and then making their profits by trading on those that that insight in the ordinary financial market.
link |
And the market would, if it's efficient, would be able to become better and better at predicting
link |
the powerful ideas that the individual is able to generate. The variance around the mass of the
link |
electron neutrino would decrease with time as we learned that value of that parameter better and
link |
any other parameters that we wanted to estimate. You don't think those markets would also respond
link |
to recency of prestige and all those kinds of things? They would respond, but the question is
link |
if they might respond incorrectly, but if you think they're doing it incorrectly, you have a
link |
profit opportunity where you can go fix it. So we'd be inviting everybody to ask whether they can
link |
find any biases or errors in the current ways in which people are estimating these things from
link |
whatever clues they have. Right. There's a big incentive for the correction mechanism in academia
link |
currently. There's not, it's the safe choice to go with the prestige. Exactly. And there's no.
link |
Even if you privately think that the prestige is over overrated. Even if you think strongly that
link |
it's overrated. Still you don't have an incentive to defy that publicly. You're going to lose a lot
link |
unless you're a contrarian that writes brilliant blogs and then you could talk about it in the
link |
pocket. Right. I mean, initially this was my initial concept of having these betting markets
link |
on these key parameters. And what I then realized over time was that that's more what people
link |
pretend to care about. What they really mostly care about is just who's how good. And that's
link |
what most of the system is built on is trying to rate people and rank them. And so I designed this
link |
other alternative based on historical evaluation centuries later, just about who's how good,
link |
because that's what I think most of the customers really care about.
link |
Customers. I like the word customers here. Humans. Right. Well, every major area of life,
link |
which, you know, has specialists who get paid to do that thing must have some customers from
link |
elsewhere who are paying for it. Well, who are the customers for the mass of the neutrino?
link |
Yes. I, I, I understand that a sense people who are willing to pay. Right. For a thing.
link |
That's an important thing to understand about anything. Who are the customers? So when I think
link |
and what's the product, like medicine, education, academia, military, et cetera, that's part of the
link |
hidden motives analysis. Often people have a thing they say about what the product is and who the
link |
customer is. And maybe you need to dig a little deeper to find out what's really going on.
link |
Or a lot deeper. You, uh, you've written that you seek out quote view quakes. You're able as a,
link |
uh, as an intelligent black box word generating machine, you're able to generate a lot of sexy
link |
words. I like it. I love it. View quakes, which are insights, which dramatically changed my
link |
worldview, your worldview. Uh, you write, I loved science fiction as a child studied physics and
link |
artificial intelligence for a long time each, and now study economics and political science,
link |
all fields full of such insights. So let me ask, what are some view quakes or a beautiful,
link |
surprising idea to you from each of those fields, physics, AI, economics, political science?
link |
I know it's a tough question. Something that springs to mind about physics, for example,
link |
that just as beautiful. I mean, right from the beginning, say special relativity was a big
link |
surprise. Uh, you know, most of us have a simple concept of time and it seems perfectly adequate
link |
for everything we've ever seen. And to have it explained to you that you need to sort of have a
link |
mixture concept of time and space where you put it into the space time construct, how it looks
link |
different from different perspectives. That was quite a shock. And that was, you know, such a
link |
shock that it makes you think, what else do I know that, you know, isn't the way it seems. Certainly
link |
quantum mechanics is certainly another enormous shock in terms of from your point, you know,
link |
you have this idea that there's a space and then there's, you know, point particles at points and
link |
maybe fields in between. And, um, quantum mechanics is just a whole different representation. It looks
link |
nothing like what you would have thought as sort of the basic representation of the physical world.
link |
And that was quite a surprise. What would you say is the catalyst for the, for the view quake in
link |
theoretical physics in the 20th century? Where does that come from? So the interesting thing
link |
about Einstein, it seems like a lot of that came from like almost thought experiments. It wasn't
link |
almost experimentally driven. Um, and with, actually, I don't know the full story of quantum
link |
mechanics, how much of it is experiment, like where, if you, if you look at the full trace of
link |
idea generation there, uh, of all the weird stuff that falls out of quantum mechanics, how much of
link |
that was the experimentalist? How much was it the theoreticians? But usually in theoretical
link |
physics, the theories lead the way. So maybe can you, uh, can you elucidate like what, what is the
link |
catalyst for these? The remarkable thing about physics and about many other areas of academic
link |
intellectual life is that it just seems way overdetermined. That is, if it hadn't been for
link |
Einstein or if it hadn't been for Heisenberg, certainly within a half a century, somebody else
link |
would have come up with essentially the same things. Is that something you believe or is that
link |
something? Yes. So I think when you look at sort of just the history of physics and the history of
link |
other areas, you know, some areas like that, there's just this enormous convergence that the,
link |
the different kinds of evidence that was being collected was so redundant in the sense that so
link |
many different things revealed the same things that eventually you just kind of have to accept it
link |
because it just gets obvious. So if you look at the details, of course, you know, Einstein did it
link |
for somebody else and it's well worth celebrating Einstein for that. And, you know, we, by
link |
celebrating the particular people who did something first or came across something first,
link |
we are encouraging all the rest to move a little faster, to try to, to push us all a little faster,
link |
which is great. But I still think we would have gotten roughly to the same place within a half
link |
century. So sometimes people are special because of how much longer it would have taken. So some
link |
people say general relativity would have taken longer without Einstein than other things. I mean,
link |
Heisenberg quantum mechanics, I mean, there were several different formulations of quantum mechanics
link |
all around the same few years, means no one of them made that much of a difference. We would have
link |
had pretty much the same thing regardless of which of them did it exactly when. Nevertheless,
link |
I'm happy to celebrate them all. But this is a choice I make in my research. That is, when there's
link |
an area where there's lots of people working together, you know, who are sort of scoping each
link |
other and getting a result just before somebody else does, you ask, well, how much of a difference
link |
would I make there? At most, I could make something happen a few months before somebody else. And so
link |
I'm less worried about them missing things. So when I'm trying to help the world, like doing research,
link |
I'm looking for neglected things. I'm looking for things that nobody's doing it. If I didn't do it,
link |
nobody would do it. Nobody would do it. Or at least for a long time. In the next 10, 20 years,
link |
kind of thing. Right, exactly. Same with general relativity, just, you know, who would do it?
link |
It might take another 10, 20, 30, 50 years. So that's the place where you can have the
link |
biggest impact is finding the things that nobody would do unless you did them.
link |
And then that's when you get the big view quake, the insight. So what about artificial
link |
intelligence? Would it be the EMs, the emulated minds? What idea, whether that struck you in the
link |
shower one day or that you just...
link |
Clearly, the biggest view quake in artificial intelligence is the realization of just how
link |
complicated our human minds are. So most people who come to artificial intelligence from other
link |
fields or from relative ignorance, a very common phenomenon, which you must be familiar with,
link |
is that they come up with some concept and then they think that must be it. Once we implement this
link |
new concept, we will have it. We will have full human level or higher artificial intelligence,
link |
right? And they're just not appreciating just how big the problem is, how long the road is,
link |
just how much is involved, because that's actually hard to appreciate. When we just think,
link |
it seems really simple. And studying artificial intelligence, going through many particular
link |
problems, looking at each problem, all the different things you need to be able to do
link |
to solve a problem like that, makes you realize all the things your minds are doing that you
link |
are not aware of. That's that vast subconscious that you're not aware of. That's the biggest
link |
view quake from artificial intelligence by far for most people who study artificial intelligence,
link |
is to see just how hard it is. I think that's a good point. But I think it's a very early
link |
view quake. It's when the Dunning Kruger crashes hard. It's the first realization that humans are
link |
actually quite incredible. The human mind, the human body is quite incredible. There's a lot
link |
of different parts to it. But then, see, it's already been so long for me to think about
link |
it. It's already been so long for me that I've experienced that view quake that, for me,
link |
I now experience the view quakes of, holy shit, this little thing is actually quite powerful,
link |
like neural networks. I'm amazed. Because you've become almost cynical after that first view quake
link |
of, like, this is so hard. Like, evolution did some incredible work to create the human mind.
link |
But then you realize, just like you have, you've talked about a bunch of simple models
link |
that simple things can actually be extremely powerful, that maybe emulating the human mind
link |
is extremely difficult. But you can go a long way with a large neural network. You can go a long way
link |
with a dumb solution. It's that Stuart Russell thing with the reinforcement learning. Holy crap,
link |
you can go quite a long way with a simple thing. But we still have a very long road to go,
link |
but not unless... I can't, I refuse to sort of know. The road is full of surprises. So long is
link |
an interesting, like you said, with the six hard steps that humans have to take to arrive at where
link |
we are from the origin of life on Earth. So it's long, maybe, in the statistical improbability of
link |
the steps that have to be taken. But in terms of how quickly those steps could be taken,
link |
I don't know if my intuition says it's, if it's hundreds of years away or if it's a couple of
link |
years away, I prefer to measure... Pretty confidence, at least a decade. And
link |
mildly confidence, at least three decades. I can steal man either direction. I prefer to
link |
measure that journey in Elon Musk's. That's a new... Well, we don't get Elon Musk very often,
link |
so that's a long timescale. For now, I don't know, maybe you can clone or maybe multiply or
link |
even know what Elon Musk, what that is. What is that? What is... That's a good question.
link |
Exactly. Well, that's an excellent question. How does that fit into the model of the three
link |
parameters that are required for becoming a grabby alien civilization? That's the question of how
link |
much any individual makes in the long path of civilization over time. Yes. And it's a favorite
link |
topic of historians and people to try to focus on individuals and how much of a difference they
link |
make. And certainly, some individuals make a substantial difference in the modest term,
link |
right? Like, you know, without Hitler being Hitler in the role he took, European history would
link |
have taken a different path for a while there. But if we're looking over like many centuries
link |
longer term things, most individuals do fade in their individual influence.
link |
So, I mean... Even Einstein. Even Einstein, no matter how sexy your hair is, you will also be
link |
forgotten in the long arc of history. So you said at least 10 years. So let's talk a little bit about
link |
this AI point of where, how we achieve, how hard is the problem of solving intelligence
link |
is the problem of solving intelligence by engineering artificial intelligence
link |
that achieves human level, human like qualities that we associate with intelligence. How hard
link |
is this? What are the different trajectories that take us there? One way to think about it
link |
is in terms of the scope of the technology space you're talking about. So let's take the biggest
link |
possible scope, all of human technology, right? The entire human economy. So the entire economy
link |
is composed of many industries, each of which have many products with many different technologies
link |
supporting each one. At that scale, I think we can accept that most innovations are a small
link |
fraction of the total. That is usually have relatively gradual overall progress. And that
link |
individual innovations that have a substantial effect that total are rare and their total effect
link |
is still a small percentage of the total economy. There's very few individual innovations that
link |
made a substantial difference to the whole economy. What are we talking? Steam engine,
link |
shipping containers, a few things. Shipping containers deserves to be up there with steam
link |
engines, honestly. Can you say exactly why shipping containers... Shipping containers
link |
revolutionized shipping. Shipping is very important. But placing that at shipping containers.
link |
So you're saying you wouldn't have some of the magic of the supply chain, all that,
link |
without shipping containers. That made a big difference, absolutely. Interesting. That's
link |
something to look into. We shouldn't take that tangent, although I'm tempted to. But anyway,
link |
so there's a few, just a few innovations. Right. So at the scale of the whole economy, right?
link |
Right. Now, as you move down to a much smaller scale, you will see individual innovations
link |
having a bigger effect, right? So if you look at, I don't know, lawnmowers or something,
link |
I don't know about the innovations lawnmower, but there were probably like steps where you
link |
just had a new kind of lawnmower and that made a big difference to mowing lawns because you're
link |
focusing on a smaller part of the whole technology space, right? And sometimes like military
link |
technology, there's a lot of military technologies, a lot of small ones, but every once in a while,
link |
a particular military weapon like makes a big difference. But still, even so, mostly overall,
link |
they're making modest differences to something that's increasing relatively soon. Like US military
link |
is the strongest in the world consistently for a while. No one weapon in the last 70 years has
link |
made a big difference in terms of the overall prominence of the US military, right? Because
link |
that's just saying, even though every once in a while, even the recent Soviet hyper missiles or
link |
whatever they are, they aren't changing the overall balance dramatically, right?
link |
So when we get to AI, now I can frame the question, how big is AI? Basically, so one way of
link |
thinking about AI is it's just all mental tasks. And then you ask what fraction of tasks are mental
link |
tasks? And then I go, a lot. And then if I think of AI as like half of everything, then I think,
link |
well, it's got to be composed of lots of parts where any one innovation is only a small impact,
link |
right? Now, if you think, no, no, no, AI is like AGI. And then you think AGI is a small thing,
link |
right? There's only a small number of key innovations that will enable it. Now you're
link |
thinking there could be a bigger chunk that you might find that would have a bigger impact. So
link |
the way I would ask you to frame these things in terms of the chunkiness of different areas of
link |
technology, in part, in terms of how big they are. If you take 10 chunky areas and you add them
link |
together, the total is less chunky. Yeah. But don't you, are you able until you solve
link |
the fundamental core parts of the problem to estimate the chunkiness of that problem?
link |
Well, if you have a history of prior chunkiness, that could be your best estimate for future
link |
chunkiness. So for example, I mean, even at the level of the world economy, right? We've had this,
link |
what, 10,000 years of civilization. Well, that's only a short time. You might say, oh, that doesn't
link |
predict future chunkiness. But it looks relatively steady and consistent. We can say even in computer
link |
science, we've had 70 years of computer science. We have enough data to look at chunkiness of
link |
computer science. Like when were there algorithms or approaches that made a big chunky difference
link |
and how large a fraction of that was that? And I'd say mostly in computer science,
link |
most innovation has been relatively small chunks. The bigger chunks have been rare.
link |
Well, this is the interesting thing. This is about AI and just algorithms in general is
link |
page rank. So Google's, right? So sometimes it's a simple algorithm that by itself is not that useful,
link |
but the scale of context and in a context that's scalable, depending on the context,
link |
all of a sudden the power is revealed. And there's something, I guess that's the nature of chunkiness
link |
is that things that can reach a lot of people simply can be quite chunky.
link |
So one standard story about algorithms is to say algorithms have a fixed cost plus a marginal cost.
link |
And so in history, when you had computers that were very small, you tried all the algorithms
link |
that had low fixed costs and you look for the best of those. But over time, as computers got bigger,
link |
you could afford to do larger fixed costs and try those. And some of those had more effective
link |
algorithms in terms of their marginal cost. And that, in fact, that roughly explains the
link |
longterm history where in fact, the rate of algorithmic improvement is about the same as
link |
the rate of hardware improvement, which is a remarkable coincidence. But it would be explained
link |
by saying, well, there's all these better algorithms you can't try until you have a big enough computer
link |
to pay the fixed cost of doing some trials to find out if that algorithm actually saves you
link |
on the marginal cost. And so that's an explanation for this relatively continuous history. So we have
link |
a good story about why hardware is so continuous. And you might think, why would software be so
link |
continuous with the hardware? But if there's a distribution of algorithms in terms of their fixed
link |
costs, and it's, say, spread out at a wide log normal distribution, then we could be sort of
link |
marching through that log normal distribution, trying out algorithms with larger fixed costs and
link |
finding the ones that have lower marginal costs.
link |
So would you say AGI, human level, AI, even EM, M, emulated minds, is chunky? Like a few
link |
breakthroughs can take us.
link |
So an M is by its nature chunky in the sense that if you have an emulated brain and you're
link |
25% effective at emulating it, that's crap. That's nothing. Okay. You pretty much need to
link |
emulate a full human brain.
link |
Is that obvious? Is that obvious?
link |
It's pretty obvious. I'm talking about like, you know, so the key thing is you're emulating
link |
various brain cells. And so you have to emulate the input output pattern of those cells. So if
link |
you get that pattern somewhat close, but not close enough, then the whole system just doesn't have
link |
the overall behavior you're looking for, right?
link |
But it could have functionally some of the power of the overall system.
link |
So there'll be some threshold. The point is when you get close enough, then it goes over the
link |
threshold, right? It's like taking a computer chip and deleting every 1% of the gates, right?
link |
No, that's very chunky. But the hope is that emulating the human brain, I mean, the human
link |
brain itself is not...
link |
Right. So it has a certain level of redundancy and a certain level of robustness. And so there's
link |
some threshold when you get close to that level of redundancy or robustness, then it starts to
link |
work. But until you get to that level, it's just going to be crap, right? It's going to be just a
link |
big thing that isn't working for us. So we can be pretty sure that emulations is a big chunk in an
link |
economic sense, right? At some point, you'll be able to make one that's actually effective in
link |
enable substituting for humans. And then that will be this huge economic product that people will
link |
try to buy like crazy.
link |
You'll bring a lot of value to people's lives, so they'll be willing to pay for it.
link |
Right. But it could be that the first emulation costs a billion dollars each, right? And then we
link |
have them, but we can't really use them. They're too expensive. And then the cost slowly comes
link |
down. And now we have less of a chunky adoption, right? That as the cost comes down, then we use
link |
more and more of them in more and more contexts. And that's a more continuous curve. So it's only
link |
if the first emulations are relatively cheap that you get a more sudden disruption to society.
link |
And that could happen if sort of the algorithm is the last thing you figure out how to do or
link |
What about robots that capture some magic in terms of social connection? The robots, like we have a
link |
robot dog on the carpet right there. Robots that are able to capture some magic of human connection
link |
as they interact with humans, but are not emulating the brain. What about those? How far away?
link |
So we're thinking about chunkiness or distance now. So if you ask how chunky is the task of making
link |
a, you know, emulatable robot or something, which chunkiness and time are correlated.
link |
Right. But it's about how far away it is or how suddenly it would happen. Chunkiness is how
link |
suddenly and difficulty is just how far away it is. But it could be a continuous difficulty. It
link |
would just be far away, but we'll slowly steadily get there. Or there could be these thresholds where
link |
we reach a threshold and suddenly we can do a lot better.
link |
Yeah. That's a good question for both. I tend to believe that all of it, not just the M, but AGI
link |
too is chunky and human level intelligence embodied in robots is also chunky.
link |
The history of computer science and chunkiness so far seems to be my rough best guess for the
link |
penis of AGI. That is, it is chunky.
link |
It's modestly chunky, not that chunky. Right.
link |
Our ability to use computers to do many things in the economy has been moving relatively steadily.
link |
Overall, in terms of our use of computers in society,
link |
they have been relatively steadily improving for 70 years.
link |
No, but I would say that's hard. Yeah. Okay. Okay. I would have to really think about that
link |
because neural networks are quite surprising.
link |
Sure. But every once in a while we have a new thing that's surprising. But if you stand back,
link |
you know, we see something like that every 10 years or so, some new innovation that has a big effect.
link |
So, moderately chunky. Yeah.
link |
But the history of the level of disruption we've seen in the past would be a rough
link |
estimate of the level of disruption in the future. Unless the future is,
link |
we're going to hit a chunky territory, much chunkier than we've seen in the past.
link |
Well, I do think there's, it's like, like Kuhnian, like revolution type.
link |
It seems like the data, especially on AI, is difficult to reason with because it's so recent,
link |
it's such a recent field. Wow, AI's been around for 50 years.
link |
I mean, 50, 60, 70, 80 years being recent. Okay.
link |
It's enough time to see a lot of trends.
link |
A few trends, a few trends. I think the internet, computing, there's really a lot of interesting
link |
stuff that's happened over the past 30 years that I think the possibility of revolutions
link |
is likelier than it was in the... I think for the last 70 years,
link |
there have always been a lot of things that look like they had a potential for revolution.
link |
So we can't reason well about this. I mean, we can reason well by looking
link |
at the past trends. I would say the past trend is roughly your best guess for the future.
link |
No, but if I look back at the things that might've looked like revolutions in the 70s and 80s and 90s,
link |
they are less like the revolutions that appear to be happening now, or the capacity of revolution
link |
that appear to be there now. First of all, there's a lot more money to be made. So there's a lot more
link |
incentive for markets to do a lot of kind of innovation, it seems like in the AI space.
link |
But then again, there's a history of winters and summers and so on.
link |
So maybe we're just like riding a nice wave right now.
link |
One of the biggest issues is the difference between impressive demos and commercial value.
link |
So we often through the history of AI, we saw very impressive demos
link |
that never really translated much into commercial value.
link |
Somebody who works on and cares about autonomous and semi autonomous vehicles,
link |
tell me about it. And there again, we return to the number of Elon Musk's per earth per year
link |
generated. That's the M. Coincidentally, same initials as the M.
link |
Very suspicious, very suspicious. We're going to have to look into that. All right. Two more fields
link |
that I would like to force and twist your arm to look for view quakes and for beautiful ideas,
link |
economics. What is a beautiful idea to you about economics? You mentioned a lot of them.
link |
Sure. So as you said before, there's going to be the first view cake most people encounter that
link |
makes the biggest difference on average in the world, because that's the only thing most people
link |
ever see is the first one. And so with AI, the first one is just how big the problem is. But
link |
once you get past that, you'll find others. Certainly for economics, the first one is just
link |
the power of markets. You might have thought it was just really hard to figure out how to optimize
link |
in a big, complicated space. And markets just do a good first pass for an awful lot of stuff.
link |
And they are really quite robust and powerful. And that's just quite the view quake, where you just
link |
say, if you want to get in the ballpark, just let a market handle it and step back. And that's true
link |
for a wide range of things. It's not true for everything, but it's a very good first approximation.
link |
Most people's intuitions for how they should limit markets are actually messing them up.
link |
They're that good in sense. Most people, when you go, I don't know if we want to trust that.
link |
Well, you should be trusting that. What are markets? Just a couple of words. So the idea
link |
is if people want something, then let other companies form to try to supply that thing.
link |
Let those people pay for their cost of whatever they're making and try to offer that product
link |
to those people. Let many such firms enter that industry and let the customers decide
link |
which ones they want. And if the firm goes out of business, let it go bankrupt and let other
link |
people invest in whichever ventures they want to try to attract customers to their version
link |
of the product. And that just works for a wide range of products and services.
link |
And through all of this, there's a free exchange of information too.
link |
There's a hope that there's no manipulation of information and so on.
link |
Even when those things happen, still just the simple market solution is usually better
link |
than the things you'll try to do to fix it.
link |
Than the alternative.
link |
That's a view, Craig. It's surprising. It's not what you would have initially thought.
link |
That's one of the great, I guess, inventions of human civilization that trust the markets.
link |
Now, another view, Craig, that I learned in my research that's not all of economics,
link |
but something more specialized is the rationality of disagreement. That is,
link |
basically people who are trying to believe what's true in a complicated situation would not actually
link |
disagree. And of course, humans disagree all the time. So it was quite the striking fact for me to
link |
learn in grad school that actually rational agents would not knowingly disagree. And so that makes
link |
disagreement more puzzling and it makes you less willing to disagree.
link |
Humans are, to some degree, rational and are able to...
link |
Their priorities are different than just figuring out the truth.
link |
Are different than just figuring out the truth.
link |
Which might not be the same as being irrational.
link |
That's another tangent that could take an hour.
link |
In the space of human affairs, political science, what is a beautiful, foundational,
link |
interesting idea to you, a view, Craig, in the space of political science?
link |
It's the main thing that goes wrong in politics is people not agreeing on what the best thing to do is.
link |
That's a wrong thing.
link |
So that's what goes wrong. That is where you say, what's fundamentally behind most
link |
political failures? It's that people are ignorant of what the consequences of policy is.
link |
And that's surprising because it's actually feasible to solve that problem,
link |
which we aren't solving.
link |
So it's a bug, not a feature that there's an inability to arrive at a consensus.
link |
So most political systems, if everybody looked to some authority, say, on a question and that
link |
authority told them the answer, then most political systems are capable of just doing that thing.
link |
That is. And so it's the failure to have trustworthy authorities
link |
that is sort of the underlying failure behind most political failure.
link |
We invade Iraq, say, when we don't have an authority to tell us that's a really stupid
link |
thing to do. And it is possible to create more informative trustworthy authorities.
link |
That's a remarkable fact about the world of institutions that we could do that, but we aren't.
link |
Yeah, that's surprising. We could and we aren't.
link |
Right. Another big view, Craig, about politics is from the elephant in the brain that most people,
link |
when they're interacting with politics, they say they want to make the world better,
link |
they make their city better, their country better, and that's not their priority.
link |
They want to show loyalty to their allies. They want to show their people they're on their side,
link |
yes. Or their various tribes they're in, that's their primary priority and they do accomplish that.
link |
Yeah. And the tribes are usually color coded, conveniently enough.
link |
What would you say, you know, it's the Churchill question. Democracy is the crappiest form of
link |
government, but it's the best one we got. What's the best form of government for this, our 7 billion
link |
human civilization and the maybe as we get farther and further. You mentioned a lot of stuff
link |
that's fascinating about human history as we become more forager like and looking out beyond
link |
what's the best form of government in the next 50, 100 years as we become a multi planetary species.
link |
So, the key failing is that we have existing political institutions and related institutions
link |
like media institutions and other authority institutions, and these institutions sit in
link |
a vast space of possible institutions. And the key failing, we're just not exploring that space.
link |
And the key failing, we're just not exploring that space. So, I have made my proposals in that space,
link |
and I think I can identify many promising solutions. And many other people have made many
link |
other promising proposals in that space. But the key thing is we're just not pursuing those
link |
proposals. We're not trying them out on small scales, we're not doing tests, we're not exploring
link |
the space of these options. That is the key thing we're failing to do. And if we did that, I am
link |
confident we would find much better institutions than one we're using now, but we would have to
link |
actually try. So, a lot of those topics, I do hope we get a chance to talk again. You're a fascinating
link |
human being. So, I'm skipping a lot of tangents on purpose that I would love to take. You're such a
link |
brilliant person on so many different topics. Let me take a stroll into the deep human psyche of
link |
Robin Hansen himself. So, first... May not be that deep.
link |
I might just be all on the surface. What you see is what you get. There might not be much hiding
link |
behind it. Some of the fun is on the surface. I actually think this is true of many of the most
link |
successful, most interesting people you see in the world. That is, they have put so much effort
link |
into the surface that they've constructed. And that's where they put all their energy. Somebody
link |
might be a statesman or an actor or something else, and people want to interview them and they
link |
want to say, what are you behind the scenes? What do you do in your free time? Those people don't
link |
have free time. They don't have another life behind the scenes. They put all their energy into
link |
that surface, the one we admire, the one we're fascinated by. And they kind of have to make up
link |
the stuff behind the scenes to supply it for you, but it's not really there. Well, there's several
link |
ways of phrasing that. So, one of it is authenticity, which is if you become the thing you are on the
link |
surface, if the depths mirror the surface, then that's what authenticity is. You're not hiding
link |
something. You're not concealing something. To push back on the idea of actors, they actually have
link |
often a manufactured surface that they put on and they try on different masks and the depths are
link |
very different from the surface. And that's actually what makes them very not interesting
link |
to interview. If you are an actor who actually lives the role that you play, so like, I don't
link |
know, Clint Eastwood type character who clearly represents the cowboy, like at least rhymes or
link |
echoes the person you play on the surface, that's authenticity. Some people are typecasts and they
link |
have basically one persona they play in all of their movies and TV shows. And so those people,
link |
it probably is the actual persona that they are, or it has become that over time. Clint Eastwood
link |
would be one. I think of Tom Hanks as an ever. I think they just always play the same person.
link |
And you and I are just both surface players. You're the fun, brilliant thinker and I am the
link |
suit wearing idiot full of silly questions. All right. That said, let's put on your wise
link |
sage hat and ask you, what advice would you give to young people today in high school and college
link |
about life, about how to live a successful life in career or just in general that they can be proud
link |
of? Most young people, when they actually ask you that question, what they usually mean is how can
link |
I be successful by usual standards? I'm not very good at giving advice about that because that's
link |
not how I tried to live my life. So I would more flip it around and say, you live in a rich society
link |
and you will have a long life. You have many resources available to you. Whatever career you
link |
take, you'll have plenty of time to make progress on something else. Yes, it might be better if you
link |
find a way to combine your career and your interests in a way that gives you more time
link |
and energy, but there are often big compromises there as well. So if you have a passion about some
link |
topic or some thing that you think just is worth pursuing, you can just do it. You don't need other
link |
people's approval. And you can just start doing whatever it is you think it's worth doing. It
link |
might take you decades, but decades are enough to make enormous progress on most all interesting
link |
things. And don't worry about the commitment of it. I mean, that's a lot of what people worry
link |
about is, well, there's so many options. And if I choose a thing and I stick with it, I sacrifice
link |
all the other paths I could have taken. So I switched my career at the age of 34 with two
link |
kids, age zero and two, went back to grad school in social science after being a research software
link |
engineer. So it's quite possible to change your mind later in life.
link |
How can you have an age of zero?
link |
Okay. Oh, you index was zero. I got it. Okay.
link |
Right. People also ask what to read and I say, textbooks. Until you've read lots of textbooks
link |
or maybe review articles, I'm not so sure you should be reading blog posts and Twitter feeds
link |
and even podcasts. I would say at the beginning, this is our best, humanity's best summary of how
link |
to learn things is crammed into textbooks. Especially the ones on like introduction to
link |
biology, introduction to everything. Just read all the algorithms, read as many textbooks as
link |
you can stomach. And then maybe if you want to know more about a subject, find review articles.
link |
Right. You don't need to read the latest stuff for most topics.
link |
Yeah. And actually textbooks often have the prettiest pictures.
link |
And depending on the field, if it's technical, then doing the homework problems at the end,
link |
it's actually extremely, extremely useful. Extremely powerful way to understand something
link |
if you allow it. I actually think of like high school and college, which you kind of remind me
link |
of, people don't often think of it that way, but you will almost not again get an opportunity
link |
to spend the time with a fundamental subject and like, and everybody's forcing you, like
link |
everybody wants you to do it. And like, you'll never get that chance again to sit there,
link |
even though it's outside of your interest, biology. Like in high school, I took AP biology,
link |
AP chemistry. I'm thinking of subjects I never again really visited seriously. And it was so
link |
nice to be forced into anatomy and physiology, to be forced into that world, to stay with it,
link |
to look at the pretty pictures, to certain moments to actually for a moment, enjoy the beauty of
link |
these, of like how a cell works and all those kinds of things. And you're somehow that stays
link |
like the ripples of that fascination that stays with you, even if you never do those,
link |
even if you never utilize those learnings in your actual work.
link |
A common problem, at least of many young people I meet is that they're like feeling
link |
idealistic and altruistic, but in a rush. So, you know, the usual human tradition that goes back,
link |
you know, hundreds of thousands of years is that people's productivity rises with time and maybe
link |
peaks around the age of 40 or 50. The age of 40, 50 is when you will be having the highest income,
link |
you'll have the most contacts, you will sort of be wise about how the world works.
link |
Expect to have your biggest impact then. Before then, you can have impacts, but you're also mainly
link |
building up your resources and abilities. That's the usual human trajectory. Expect that to be
link |
true of you too. Don't be in such a rush to like accomplish enormous things at the age of 18 or
link |
whatever. I mean, you might as well practice trying to do things, but that's mostly about
link |
learning how to do things by practicing. There's a lot of things you can't do unless you just
link |
keep trying them. And when all else fails, try to maximize the number of offspring,
link |
however way you can. That's certainly something I've neglected. I would tell my younger version
link |
of myself, try to have more descendants. Yes, absolutely. It matters more than I gave,
link |
I realized at the time. Both in terms of making copies of yourself in mutated form
link |
and just the joy of raising them. Sure. I mean, the meaning even, you know, so in the literature on
link |
the value people get out of life, there's a key distinction between happiness and meaning.
link |
So happiness is how do you feel right now about right now and meaning is how do you feel about
link |
your whole life? And, you know, many things that produce happiness don't produce meaning as
link |
reliably. And if you have to choose between them, you'd rather have meaning. And meaning is more
link |
goes along with sacrificing happiness sometimes. And children are an example of that. You get a lot
link |
more meaning out of children, even if they're a lot more work. What do you think kids, children
link |
are so magical, like raising kids? I would love to have kids. And whenever I work with robots,
link |
there's some of the same magic when there's an entity that comes to life. And in that case,
link |
I'm not trying to draw too many parallels, but there is some echo to it, which is when you
link |
program a robot, there's some aspect of your intellect that is now instilled in this other
link |
moving being that's kind of magical. Well, why do you think that's magical? And you said happiness
link |
and meaning as opposed to a short. Why is it meaningful? It's overdetermined. Like I can give
link |
you several different reasons, all of which is sufficient. And so the question is, we don't know
link |
which ones are the correct reasons. It's overdetermined. Look it up. So, you know, I meet a
link |
lot of people interested in the future, interested in thinking about the future. They're thinking
link |
about how can I influence the future? But overwhelmingly in history so far, the main way
link |
people have influenced the future is by having children, overwhelmingly. And that's just not an
link |
incidental fact. You are built for that. That is, you're the sequence of thousands of generations,
link |
each of which successfully had a descendant. And that affected who you are. You just have to expect
link |
and it's true that who you are is built to be, you know, expect to have a child, to want to have a
link |
child, to have that be a natural and meaningful interaction for you. And it's just true. It's just
link |
one of those things you just should have expected and it's not a surprise. Well, to push back and
link |
sort of in terms of influencing the future, as we get more and more technology, more and more of us
link |
are able to influence the future in all kinds of other ways, right? Being a teacher, educator. Even
link |
so, though, still most of our influence in the future has probably happened being kids, even
link |
though we've accumulated more ways, other ways to do it. You mean at scale. I guess the depth of
link |
influence, like really how much of much effort, how much of yourself you really put into another human
link |
being. Do you mean both the raising of a kid or you mean raw genetic information? Well, both, but
link |
raw genetics is probably more than half of it. More than half. More than half. Even in this modern
link |
world? Yeah. Genetics. Let me ask some dark, difficult questions, if I might. Let's take a
link |
stroll into that place that may or may not exist, according to you. What's the darkest place you've
link |
ever gone to in your mind, in your life, a dark time, a challenging time in your life that you had to overcome?
link |
You know, probably just feeling strongly rejected. And so I've been, I'm apparently somewhat
link |
emotionally scarred by just being very rejection averse, which must have happened because
link |
some rejections were just very scarring. At a scale in what kinds of communities? On the
link |
individual scale? I mean, lots of different scales, yeah. All the different, many different scales. Still
link |
that rejection stings. Hold on a second, but you are a contrarian thinker. You challenge the
link |
norms. Why, if you were scarred by rejection, why welcome it in so many ways at a much
link |
larger scale, constantly with your ideas? It could be that I'm just stupid, or that I've just categorized
link |
them differently than I should or something. You know, the most rejection that I've faced hasn't been
link |
because of my intellectual ideas. So the intellectual ideas haven't been the thing
link |
to risk the rejection. The one that, the things that challenge your mind taking you to a dark
link |
place are the more psychological rejections. So. Well, you just asked me, you know, what took me to a
link |
dark place. You didn't specify it as sort of an intellectual dark place, I guess. Yeah, I just
link |
meant like what? So intellectual is disjoint or at least at a more surface level than something
link |
emotional? Yeah, I would just think, you know, there are times in your life when, you know,
link |
you're just in a dark place and that can have many different causes. And most, you know, most
link |
intellectuals are still just people and most of the things that will affect them are the kinds of
link |
things that affect people. They aren't that different necessarily. I mean, that's going to be true for,
link |
like, I presume most basketball players are still just people. If you ask them what was the worst
link |
part of their life, it's going to be this kind of thing that was the worst part of life for most
link |
people. So rejection early in life? Yeah, I think, I mean, not in grade school probably, but, you know,
link |
yeah, sort of, you know, being a young nerdy guy and feeling, you know, not in much demand or interest
link |
or, you know, later on, lots of different kinds of rejection. But yeah, but I think that's, you know,
link |
most of us like to pretend we don't that much need other people. We don't care what they think.
link |
I know it's a common sort of stance if somebody rejects you or something, I didn't care about them
link |
anyway. I, you know, didn't, but I think to be honest, people really do care. Yeah, we do seek
link |
that connection, that love. What do you think is the role of love in the human condition?
link |
Um, opacity, in part. That is, love is one of those things where we know at some level it's
link |
important to us, but it's not very clearly shown to us exactly how or why or in what ways.
link |
There are some kinds of things we want where we can just clearly see that we want and why that we
link |
want it, right? We know when we're thirsty, and we know why we were thirsty, and we know what to
link |
do about being thirsty, and we know when it's over that we're no longer thirsty. Love isn't like that.
link |
It's like, what do we seek from this? We're drawn to it, but we do not understand why
link |
we're drawn exactly. Because it's not just affection, because if it was just affection,
link |
we don't seem to be drawn to pure affection. We don't seem to be drawn to somebody who's like a
link |
servant. We don't seem to be necessarily drawn to somebody that satisfies all your needs or something
link |
like that. So it's clearly something we want or need, but we're not exactly very clear about it,
link |
and that is kind of important to it. So I've also noticed there are some kinds of things
link |
you can't imagine very well. So if you imagine a situation, there's some aspects of the situation
link |
that you can clearly, you can imagine it being bright or dim, you can imagine it being windy,
link |
or you can imagine it being hot or cold. But there's some aspects about your emotional stance
link |
in a situation that's actually just hard to imagine or even remember. You can often remember
link |
an emotion only when you're in a similar sort of emotion situation, and otherwise, you just can't
link |
bring the emotion to your mind, and you can't even imagine it, right? So there's certain kinds of
link |
emotions you can have, and when you're in that emotion, you can know that you have it, and you
link |
can have a name, and it's associated. But later on, I tell you, remember joy, and it doesn't come to
link |
mind. I'm not able to replay it. Right. And it's the sort of reason why we have, one of the reasons
link |
that pushes us to re consume it and reproduce it is that we can't reimagine it. Well, it's interesting
link |
because there's a Daniel Kahneman type of thing of reliving memories, because I'm able to summon
link |
some aspect of that emotion, again, by thinking of that situation from which that emotion came.
link |
Right. So like a certain song, you can listen to it, and you can feel the same way you felt the
link |
first time you remember that song associated with it. Right. So you need to remember that situation
link |
in some sort of complete package. Yes. You can't just take one part off of it, and then if you get
link |
the whole package again, if you remember the whole feeling. Yes. Or some fundamental aspect of that
link |
whole experience that arouse from which the feeling arose. And actually, the feeling is probably
link |
different in some way. It could be more pleasant or less pleasant than the feeling you felt
link |
originally, and that morphs over time every time you replay that memory. It is interesting. You're
link |
not able to replay the feeling perfectly. You don't remember the feeling. You remember the facts of the
link |
events. So there's a sense of which over time we expand our vocabulary as a community of language,
link |
and that allows us to sort of have more feelings and know that we are feeling them. Because you can
link |
have a feeling but not have a word for it, and then you don't know how to categorize it or even
link |
what it is and whether it's the same as something else. But once you have a word for it, you can
link |
sort of pull it together more easily. And so I think over time we are having a richer palette of
link |
feelings because we have more words for them. What has been a painful loss in your life?
link |
Maybe somebody or something that's no longer in your life, but played an important part of your life.
link |
That's a concept. No, it has to be...
link |
I mean, but I was once younger. I had health and I had vitality. I was
link |
insomere. I mean, you know, I've lost that over time.
link |
Do you see that as a different person? Maybe you've lost that person.
link |
Yes, absolutely. I'm a different person than I was when I was younger, and I don't even remember
link |
exactly what he was. So I don't remember as many things from the past as many people do. So in
link |
some sense, I've just lost a lot of my history by not remembering it. And I'm not that person
link |
anymore. That person is gone and I don't have any of their abilities.
link |
Is it a painful loss, though?
link |
Or is it a... Why is it painful? Because you're wiser.
link |
There's so many things that are beneficial to getting older.
link |
Right. But I just was this person and I felt assured that I could continue to be that person.
link |
And you're no longer that person.
link |
And he's gone. And I'm not him anymore. And he died without fanfare or a funeral.
link |
And that the person you are today talking to me, that person will be changed, too.
link |
Yes. And maybe in 20 years, he won't be there anymore.
link |
And the future person, we'll look back. The future version of you will...
link |
For Ems, this will be less of a problem. For Ems, they would be able to save an archived
link |
copy of themselves at each different age. And they could turn it on periodically and go back
link |
To replay. You think some of that will be... So with emulated minds, with Ems,
link |
there's a digital cloning that happens. And do you think that makes you less special if you're
link |
clonable? Does that make you the experience of life, the experience of a moment, the scarcity
link |
of that moment, the scarcity of that experience, isn't that a fundamental part of what makes
link |
that experience so delicious, so rich of feeling?
link |
I think if you think of a song that lots of people listen to that are copies all over the
link |
world, we're going to call that a more special song.
link |
So there's a perspective on copying and cloning where you're just scaling happiness versus
link |
I mean, each copy of a song is less special if there are many copies, but the song itself is
link |
more special if there are many copies.
link |
In a mass, right, you're actually spreading the happiness even if it diminishes over a
link |
large number of people at scale and that increases the overall happiness in the world.
link |
And then you're able to do that with multiple songs.
link |
Is a person who has an identical twin more or less special?
link |
Well, the problem with identical twins is, you know, it's like just two with M's.
link |
Right, but two is different than one.
link |
So I think an identical twin's life is richer for having this other identical twin, somebody
link |
who understands them better than anybody else can.
link |
From the point of view of an identical twin, I think they have a richer life for being
link |
part of this couple, each of which is very similar.
link |
Now, if you said, will the world, you know, if we lose one of the identical twins, will
link |
the world miss it as much because you've got the other one and they're pretty similar?
link |
Maybe from the rest of the world's point of view, they suffer less of a loss when they
link |
lose one of the identical twins.
link |
But from the point of view of the identical twin themselves, their life is enriched by
link |
See, but the identical twin copying happens at the place of birth.
link |
It's different than copying after you've done some of the environment, like the nurture
link |
at the teenage or in the 20s after going to college.
link |
Yes, that'll be an interesting thing for M's to find out all the different ways that
link |
they can have different relationships to different people who have different degrees of similarity
link |
But it seems like a rich space to explore and I don't feel sorry for them.
link |
This sounds like an interesting world to live in.
link |
And there could be some ethical conundrums there.
link |
There will be many new choices to make that they don't make now.
link |
So, and I discussed that in the book Age of M.
link |
Like, say you have a lover and you make a copy of yourself, but the lover doesn't make
link |
Well now, which one of you or are both still related to the lover?
link |
Socially entitled to show up.
link |
Yes, so you'll have to make choices then when you split yourself, which of you inherit
link |
which unique things.
link |
Yeah, and of course there'll be an equivalent increase in lawyers.
link |
Well, I guess you can clone the lawyers to help manage some of these negotiations of
link |
how to split property.
link |
The nature of owning, I mean, property is connected to individuals, right?
link |
You only really need lawyers for this with an inefficient, awkward law that is not very
link |
transparent and able to do things.
link |
So, you know, for example, an operating system of a computer is a law for that computer.
link |
When the operating system is simple and clean, you don't need to hire a lawyer to make a
link |
key choice with the operating system.
link |
You don't need a human in the loop.
link |
You just make a choice, right?
link |
So ideally we want a legal system that makes the common choices easy and not require much
link |
And that's the digitization of things further enables that.
link |
So the loss of a younger self, what about the loss of your life overall?
link |
Do you ponder your death, your mortality?
link |
Are you afraid of it?
link |
I am a cryonics customer.
link |
That's what this little tag around my deck says.
link |
It says that if you find me in a medical situation, you should call these people to enable the
link |
cryonics transfer.
link |
So I am taking a long shot chance at living a much longer life.
link |
Can you explain what cryonics is?
link |
So when medical science gives up on me in this world, instead of burning me or letting
link |
worms eat me, they will freeze me or at least freeze my head.
link |
And there is damage that happens in the process of freezing the head.
link |
But once it's frozen, it won't change for a very long time.
link |
Chemically, it'll just be completely exactly the same.
link |
So future technology might be able to revive me.
link |
And in fact, I would be mainly counting on the brain emulation scenario, which doesn't
link |
require reviving my entire biological body.
link |
It means I would be in a computer simulation.
link |
And so I think I've got at least a 5% shot at that.
link |
And that's immortality.
link |
But most likely it won't happen.
link |
And therefore, I'm sad that it won't happen.
link |
Do you think immortality is something that you would like to have?
link |
Well, I mean, just like infinity, I mean, you can't know until forever, which means
link |
So all you can really, you know, the better choice is at each moment, do you want to keep
link |
So I would like at every moment to have the option to keep going.
link |
The interesting thing about human experience is that the way you phrase it is exactly right.
link |
At every moment, I would like to keep going.
link |
But the thing that happens, you know, leave them wanting more of whatever that phrase
link |
is, the thing that happens is over time, it's possible for certain experiences to become
link |
bland and you become tired of them.
link |
And that actually makes life really unpleasant.
link |
Sorry, makes that experience really unpleasant.
link |
And perhaps you can generalize that to life itself if you have a long enough horizon.
link |
Might happen, but might as well wait and find out.
link |
But then you're ending on suffering, you know?
link |
So in the world of brain emulations, I have more options.
link |
You can return yourself.
link |
That is, I can make copies of myself, archive copies at various ages.
link |
And at a later age, I could decide that I'd rather replace myself with a new copy from
link |
So does a brain emulation still operate in physical space?
link |
So can we do, what do you think about like the metaverse and operating in virtual reality
link |
so we can conjure up not just emulate, not just your own brain and body, but the entirety
link |
of the environment?
link |
Well, most brain emulations will, in fact, most of their time in virtual reality.
link |
But they wouldn't think of it as virtual reality.
link |
They would just think of it as their usual reality.
link |
I mean, the thing to notice, I think, in our world, most of us spend most time indoors.
link |
And indoors, we are surrounded by walls covered with paint and floors covered with
link |
Most of our environment is artificial.
link |
It's constructed to be convenient for us.
link |
It's not the natural world that was there before.
link |
A virtual reality is basically just like that.
link |
It is the environment that's comfortable and convenient for you.
link |
But when it's the right, that environment for you, it's real for you.
link |
Just like the room you're in right now most likely is very real for you.
link |
You're not focused on the fact that the paint is hiding the actual studs behind the
link |
wall and the actual wires and pipes and everything else.
link |
The fact that we're hiding that from you doesn't make it fake or unreal.
link |
What are the chances that we're actually in the very kind of system that you're describing
link |
where the environment and the brain is being emulated and you're just replaying an experience
link |
when you first did a podcast with Lex after?
link |
And now, the person that originally launched this already did hundreds of podcasts with
link |
This is just the first time and you like this time because there's so much uncertainty.
link |
It could have gone any direction.
link |
At the moment, we don't have the technical ability to create that emulation.
link |
So we'd have to be postulating that in the future we have that ability and then they
link |
choose to evaluate this moment now to simulate it.
link |
Don't you think we could be in the simulation of that exact experience right now and we
link |
wouldn't be able to know?
link |
So one scenario would be this never really happened.
link |
This only happens as a reconstruction later on.
link |
That's different than the scenario that this did happen the first time and now it's happening
link |
again as a reconstruction.
link |
That second scenario is harder to put together because it requires this coincidence where
link |
between the two times we produce the ability to do it.
link |
But don't you think replay of memories, poor replay of memories is something that might
link |
be a possible thing in the future?
link |
You're saying it's harder than conjure up things from scratch.
link |
It's certainly possible.
link |
So the main way I would think about it is in terms of the demand for simulation versus
link |
other kinds of things.
link |
So I've given this a lot of thought because I first wrote about this long ago when Bostrom
link |
first wrote his papers about simulation argument and I wrote about how to live in a simulation.
link |
And so the key issue is the fraction of creatures in the universe that are really experiencing
link |
what you appear to be really experiencing relative to the fraction that are experiencing
link |
it in a simulation way, i.e., simulated.
link |
So then the key parameter is at any one moment in time, creatures at that time, many of them,
link |
most of them are presumably really experiencing what they're experiencing, but some fraction
link |
of them are experiencing some past time where that past time is being remembered via their
link |
So to figure out this ratio, what we need to think about is basically two functions.
link |
One is how fast in time does the number of creatures grow?
link |
And then how fast in time does the interest in the past decline?
link |
Because at any one time, people will be simulating different periods in the past with different
link |
I love the way you think so much.
link |
That's exactly right, yeah.
link |
So if the first function grows slower than the second one declines, then in fact, your
link |
chances of being simulated are low.
link |
So the key question is how fast does interest in the past decline relative to the rate
link |
at which the population grows with time?
link |
Does this correlate to you earlier suggested that the interest in the future increases
link |
over time, are those correlated interest in the future versus interest in the past?
link |
Like, why are we interested in the past?
link |
So, but the simple way to do it is, as you know, like Google Ngrams has a way to type
link |
in a word and see how interest in it declines or rises over time, right?
link |
You can just type in a year and get the answer for that.
link |
If you type in a particular year, like 1900 or 1950, you can see with Google Ngram, how
link |
interest in that year increased up until that date and decreased after it.
link |
And you can see that interest in a date declines faster than does the population grow with
link |
That is brilliant.
link |
That is so interesting.
link |
And so you have the answer.
link |
And that was your argument against, not against to this particular aspect of the simulation,
link |
how much past simulation there will be, replay of past memories.
link |
First of all, if we assume that like simulation of the past is a small fraction of all the
link |
creatures at that moment.
link |
And then it's about how fast.
link |
Now, some people have argued plausibly that maybe most interest in the past falls with
link |
this fast function, but some unusual category of interest in the past won't fall that fast
link |
And then that eventually would dominate.
link |
So that's a other hypothesis you want.
link |
So that very outlier specific kind of, yeah, okay.
link |
Like really popular kinds of memories, like probably sexual.
link |
In a trillion years, there's some small research institute that tries to randomly select from
link |
all possible people in history or something to simulate.
link |
So the question is how big is this research institute and how big is the future in a trillion
link |
And that would be hard to say.
link |
But if we just look at the ordinary process by which people simulate recent errors.
link |
So if you look at, it's also true for movies and plays and video games,
link |
overwhelmingly they're interested in the recent past.
link |
There's very few video games where you play someone in the Roman Empire.
link |
But even fewer where you play someone in the ancient Egyptian Empire.
link |
Yeah, just different.
link |
It's just declined very quickly.
link |
But every once in a while that's brought back.
link |
But yeah, you're right.
link |
I mean, just if you look at the mass of entertainment, movies and games, it's focusing on the present
link |
And maybe some, I mean, where does science fiction fit into this?
link |
Because it's sort of, what is science fiction?
link |
I mean, it's a mix of the past and the present and some kind of manipulation of that to make
link |
it more efficient for us to ask deep philosophical questions about humanity.
link |
The closest genre to science fiction is clearly fantasy.
link |
Fantasy and science fiction in many bookstores and even Netflix or whatever categories, they're
link |
just lumped together.
link |
So clearly they have a similar function.
link |
So that the function of fantasy is more transparent than the function of science fiction.
link |
So use that as your guide.
link |
What's fantasy for is just to take away the constraints of the ordinary world and imagine
link |
stories with much fewer constraints.
link |
That's what fantasy is.
link |
You are much less constrained.
link |
What's the purpose to remove constraints?
link |
Is it to escape from the harshness of the constraints of the real world?
link |
Or is it to just remove constraints in order to explore some, get a deeper understanding
link |
I mean, why do people read fantasy?
link |
I'm not a cheap fantasy reading kind of person.
link |
One story that sounds plausible to me is that there are sort of these deep story structures
link |
that we love and we want to realize.
link |
And then many details of the world get in their way.
link |
Fantasy takes all those obstacles out of the way and lets you tell the essential hero story
link |
or the essential love story, whatever essential story you want to tell.
link |
Well, the reality and constraints are not in the way.
link |
And so science fiction can be thought of as like fantasy, except you're not willing to
link |
admit that it can't be true.
link |
So the future gives the excuse of saying, well, it could happen.
link |
And you accept some more reality constraints for the illusion, at least, that maybe it
link |
could really happen.
link |
Maybe it could happen.
link |
And that, it stimulates the imagination.
link |
Imagination is something really interesting about human beings.
link |
And it seems also to be an important part of creating really special things is to be
link |
able to first imagine them.
link |
With you and Nick Bostrom, where do you land on the simulation and all the mathematical
link |
ways of thinking it and just the thought experiment of it?
link |
Are we living in a simulation?
link |
That was just the discussion we just had.
link |
That is, you should grant the possibility of being a simulation.
link |
You shouldn't be 100% confident that you're not.
link |
You should certainly grant a small probability.
link |
The question is, how large is that probability?
link |
Are you saying we would be, I misunderstood because I thought our discussion was about
link |
replaying things that already happened.
link |
But the whole question is, right now, is that what I am?
link |
Am I actually a replay from some distant future?
link |
But it doesn't necessarily need to be a replay.
link |
It could be a totally new.
link |
You could be, you don't have to be an NPC.
link |
Clearly, I'm in a certain era with a certain kind of world around me.
link |
So either this is a complete fantasy or it's a past of somebody else in the future.
link |
No, it could be a complete fantasy though.
link |
But then you have to talk about what's the fraction of complete fantasies.
link |
I would say it's easier to generate a fantasy than to replay a memory.
link |
Oh, but the fraction is important.
link |
We just look at the entire history of everything.
link |
We just say, sure, but most things are real.
link |
Most things aren't fantasies.
link |
Therefore, the chance that my thing is real.
link |
So the simulation argument works stronger about sort of the past.
link |
We say, ah, but there's more future people than there are today.
link |
So you being in the past of the future makes you special relative to them,
link |
which makes you more likely to be in a simulation.
link |
If we're just taking the full count and saying, in all creatures ever,
link |
what percentage are in simulations?
link |
Probably no more than 10%.
link |
So what's the good argument for that?
link |
That most things are real?
link |
Because a classroom says the other way, right?
link |
In a competitive world, in a world where people have to work and have to get things done,
link |
then they have a limited budget for leisure.
link |
And so, you know, leisure things are less common than work things, like real things.
link |
But if you look at the stretch of history in the universe, doesn't the ratio of leisure increase?
link |
Isn't that where we, isn't that the forger?
link |
Right, but now we're looking at the fraction of leisure,
link |
which takes the form of something where the person doing the leisure doesn't realize it.
link |
Now there could be some fraction of that, but that's much smaller, right?
link |
Or somebody is clueless in the process of supporting this leisure, right?
link |
It might not be the person leisureing, somebody,
link |
they're a supporting character or something,
link |
but still that's got to be a pretty small fraction of leisure.
link |
What, you mentioned that children are one of the things that are a source of meaning.
link |
Broadly speaking, then let me ask the big question.
link |
What's the meaning of this whole thing?
link |
Robin, meaning of life.
link |
What is the meaning of life?
link |
We talked about alien civilizations, but this is the one we got.
link |
Where are the aliens?
link |
Where are the human?
link |
Seem to be conscious, be able to introspect.
link |
What's, why are we here?
link |
This is the thing I told you before about how we can predict that
link |
future creatures will be different from us.
link |
We, our preferences are this amalgam of various sorts of random sort of patched together
link |
preferences about thirst and sex and sleep and attention and all these sorts of things.
link |
So we don't understand that very well.